Mining an ePR system using a treatment plan navigator for radiation toxicity to evaluate proton therapy treatment protocol for prostate cancer

PDF

Download

Open document

Flip pages

Summarize

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

MINING AN EPR SYSTEM USING A TREATMENT PLAN NAVIGATOR FOR
RADIATION TOXICITY TO EVALUATE PROTON THERAPY TREATMENT
PROTOCOL FOR PROSTATE CANCER

by

Anh Hong Tu Le

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(BIOMEDICAL ENGINEERING)

August 2010

Copyright 2010 Anh Hong Tu Le
ii

Dedication
To my dearest parents, Oanh and Chinh, for their unconditional love and the opportunity
for the first step to make my dreams a reality,
my awesome sister, Trai, for her devoted love, enormous assistance, thoughtfulness and
logical advice,
my little brother, Tuan Anh, for his encouragement and forever young wisdom,
my coolest brother-in-law, Mark, for his eccentricity, generosity and the amazing fun,
my adorable niece and nephew, Mai and Ladislaus, for their innocence and growing
curiosity of a wonderful world,
and my best friend, Lan, for her great love, friendship, support and objective opinions.
iii

Acknowledgments
This work has been done with the partial contribution and collaboration of the following
colleagues and friends:
• HK Huang, DSc Advisor
• Brent Liu, PhD Advisor
• Jorge Documet, PhD Initial ePR framework
• Ashley Sullivan, BS Integration of DICOM image display
• Ruchi Deshpande, MS Non-DICOM module for clinical data forms
• James Sayre, PhD Statistical analysis
• Maria Y.Y. Law, PhD Radiation therapy knowledge advice
• Reinhard W. Schulte, MD, MS, Associate Professor, Director of Translational
Research of LLUMC for his valuable clinical suggestions, advice, and input.
• Jerry M. Slater, MD, Professor and Chair of the Department of Radiation
Medicine, Medical Director and Director of Clinical Affairs for the department at
LLUMC for allowing facility usage, patient data collection and technical support.
In addition, I would like to express my deepest gratitude to my advisors: Professor HK
(Bernie) Huang for his guidance, great vision and inspiration, and Professor Brent J. Liu
for his patience, supervision to keep me on track and most of all for believing in me.
I would like to give my warmest thanks to all other members of my PhD Committee,
Professors K. Kirk Shung and Jill L. McNitt-Gray and Maria Y.Y. Law for their helpful
advices, and especially to Professor David Z. D'Argenio for his wisdom and
iv

encouragement throughout the years of my graduate studies at the University of Southern
California.
I would also like to thank all members of the Image Processing and Informatics
Laboratory for their support and feedback as a friend, colleague and member of our
IPILab family.
Last but not least, I would like to give special thanks to all my friends and relatives who
have supported me for many years.

v

Table of Contents
Dedication ......................................................................................................................... ii
Acknowledgments.............................................................................................................. iii
List of Tables ..................................................................................................................... ix
List of Figures ..................................................................................................................... x
Abbreviations ................................................................................................................... xiv
Abstract ...................................................................................................................... xvi
Chapter 1 Introduction ..................................................................................................... 1
1.1 Prostate Cancer ...............................................................................................1
1.1.1 Available Treatments ..........................................................................1
1.2 Proton Therapy................................................................................................2
1.2.1 Physics & Treatment Centers..............................................................2
1.2.2 Radiation Treatment Side Effects .......................................................5
1.3 Current Challenges in Proton Therapy for Prostate Cancer ............................6
1.4 Proton Therapy ePR System ...........................................................................8
1.5 Aim and Scope ................................................................................................9
1.6 Chapter Summary .........................................................................................11
Chapter 2 Proton Therapy Clinical Workflow and Data ................................................ 12
2.1 Introduction ...................................................................................................12
2.2 Proton Therapy Workflow ............................................................................12
2.3 Proton Therapy Data Objects: DICOM–RT and DICOM–RT–ION
Objects ..........................................................................................................17
2.3.1 RT Structure Set ................................................................................17
2.3.2 RT Dose ............................................................................................18
2.3.3 RT Image ..........................................................................................18
2.3.4 RT ION Plan .....................................................................................18
2.3.5 RT ION Beams Treatment Record ...................................................19
2.3.6 DICOM RT and DICOM RT ION Dependencies ............................19
2.4 Treatment Protocols ......................................................................................20
2.4.1 Dose Escalation Protocol ..................................................................20
2.4.2 Hypofractionation Protocol ...............................................................22
2.5 Clinical Outcomes of Prostate Cancer Patient Treated with Proton
Therapy .........................................................................................................23
2.5.1 Local Tumor Control ........................................................................23
2.5.2 Radiation Toxicity ............................................................................23
Chapter 3 Proton Therapy ePR System .......................................................................... 25
3.1 Data Model and Standards ............................................................................25
vi

3.1.1 The Data Model ................................................................................25
3.1.2 Standards for Data and Image Distribution ......................................26
3.1.2.1 DICOM (Digital Imaging and Communication in
Medicine) ............................................................................. 26
3.1.2.2 HTTPS (Hypertext Transfer Protocol Secure) ..................... 26
3.1.2.3 JPEG (Joint Photographic Expert Group), GIF
(Graphics Interchange Format), and PGN (Portable
Network Graphics) ............................................................... 27
3.2 The ePR Data Flow Model ...........................................................................27
3.3 The ePR System Architecture .......................................................................30
3.4 Data Gateway ................................................................................................31
3.4.1 DICOM Module ................................................................................31
3.4.2 Non-DICOM Module........................................................................31
3.5 The ePR Server .............................................................................................32
3.5.1 Storage/ Database ..............................................................................32
3.5.2 Data Quantification Module .............................................................34
3.5.3 Image Processing Engine ..................................................................35
3.6 Decision Support Tools.................................................................................35
3.7 Visualization and Display Tools ...................................................................36
Chapter 4 Data Collection .............................................................................................. 38
4.1 Introduction ...................................................................................................38
4.2 Data Types ....................................................................................................38
4.3 DICOM Files ................................................................................................40
4.4 Non-DICOM Related Clinical Data ..............................................................41
4.4.1 Initial Patient Data Form ...................................................................42
4.4.2 Follow-Up Data Form .......................................................................45
4.5 Patient Datasets .............................................................................................46
4.6 Data Organization .........................................................................................47
4.7 Summary of Data Collection ........................................................................48
4.7.1 Demonstration Patients Descriptions ................................................48
4.7.2 Hypofractionation Patients ................................................................50
Chapter 5 Intelligent Treatment Plan Navigator ............................................................ 53
5.1 The Need for Treatment Planning Navigation Tool .....................................53
5.2 Features of the Intelligent Treatment Plan Navigator (ITPN) ......................54
5.2.1 Dose Volume Histogram (DVH) and features development ............54
5.2.2 Functions of the ITPN .......................................................................55
5.2.3 ITPN Mock-up GUI ..........................................................................59
5.3 Summary .......................................................................................................59
Chapter 6 Data Mining Using Knowledge Driven Search Engine ................................ 61
6.1 Significance of Data Mining using Knowledge Driven Seach Engine .........61
6.2 Knowledge of DICOM RT and DICOM RT-ION Data ...............................61
6.2.1 DVH Data .........................................................................................62
6.2.2 3D Dose and 2D Dose Correlation ...................................................63
vii

6.2.3 Quantification of Received Dose on Organs at Risk ........................64
6.3 Knowledge of Outcomes Data ......................................................................65
6.3.1 Toxicity Grade Computation ............................................................66
6.4 The Integration of Knowledge and the Development of the Search
Engine ...........................................................................................................67
6.5 Summary .......................................................................................................68
Chapter 7 System Implementation ................................................................................. 70
7.1 System Implementation ................................................................................70
7.1.1 Web Technology ...............................................................................70
7.1.2 Database ............................................................................................70
7.1.3 Graphical user interface (GUI) .........................................................72
7.2 Laboratory Evaluation of the ePR System ....................................................72
7.2.1 Hardware & Software .......................................................................73
7.2.2 Communication & Data ....................................................................73
7.2.2.1 DICOM Communication and DICOM Objects of
Treatment Plans ................................................................... 73
7.2.2.2 Web Communication & Non-DICOM Data ........................ 74
7.2.2.3 Data Distribution .................................................................. 75
7.2.3 Features: ITPN & Search Engine ......................................................77
7.3 Summary .......................................................................................................78
Chapter 8 Data and System Integration for Improving Patient Outcome Study ............ 79
8.1 Introduction ...................................................................................................79
8.2 Existing Dataflow for Collecting Hypofractionation Patient Data ...............80
8.3 Improved Dataflow for Collecting Hypofractionation Patient Data
with ePR System ...........................................................................................82
8.4 Comparison of Dataflow without and with ePR Implementation .................84
8.5 Improvement and Efficiency in Data Collection and Integration with
ePR System ...................................................................................................87
Chapter 9 Knowledge Discovery for Decision Support in Treating Prostate
Cancer Patient with Hypofractionation Protocol Proton Therapy ................. 89
9.1 Introduction ...................................................................................................89
9.2 Existing Hypofractionation Protocol Workflow ...........................................90
9.3 Enhanced Hypofractionation Protocol Workflow with ePR System ...........92
9.4 Clinical Example of Utilizing Knowledge Discovery for Decision
Support ..........................................................................................................94
9.4.1 Description of Hypothetical Patient ..................................................94
9.4.2 Clinical Example: Patient Criteria Assessment using the ePR .........94
9.4.3 Clinical Example: Knowledge Discovery Utilizing the ePR ............97
9.4.3.1 Data Mining ......................................................................... 97
9.4.3.2 Selection of Hypofractionation Patients Matched
Search Criteria ................................................................... 100
9.4.3.3 Modification of Treatment Plan to Suite the New
Patient ................................................................................ 102
viii

9.5 Summary .....................................................................................................108
Chapter 10 Current Status and Future Direction ............................................................ 110
10.1 Current status ..............................................................................................110
10.2 Future Research Direction ..........................................................................111
References ..................................................................................................................... 114
Alphabetical References ................................................................................................. 116
Appendices ..................................................................................................................... 118
Appendix A Patient Demo S ................................................................................118
Appendix B Follow-up Questionnaires ...............................................................122

ix

List of Tables
Table 1-1 Status and Name & Location of Proton Therapy Facilities ................................4
Table 1-2 Proton ePR System Specifications ...................................................................10
Table 4-1 Proton therapy data collected from one patient ................................................39
Table 4-2 Number of objects from one patient .................................................................47
Table 4-3 Summary of main pretreatment clinical data (n=39) ........................................50
Table 4-4 Summary of dosimetrist and technical parameters in treatment plan
(n=39) ...............................................................................................................51
Table 4-5 Summary of follow-up clinical data (n=39) .....................................................51
Table 6-1 Dose limit for critical structures .......................................................................63
Table 8-1 Differences between hypofractionation dataflow with and without ePR
system ..............................................................................................................85
Table 8-2 Time improvement of workflow and data completeness .................................87
Table 9-1 Patient criteria for hypofractionation protocol treatment .................................91
Table 9-2 Patient New information and treatment plan in comparison to search
criteria ..............................................................................................................98
Table 9-3 Comparison of volume percentage of the tumor target and critical
structures between previous plan and modified plan of Pt. New ...................105

x

List of Figures
Figure 1-1 Graph showing a comparison between Photon Depth Dose (6MV) and
Proton Depth Dose (130, 150, 170, 190 MeV) percentages as measured
in water. The maximum dose from the proton can be deposited at a
certain depth. This can be utilized to treat cancer tumors at a certain
depth while minimizing dose to surrounding normal healthy tissue [2]. ...........4
Figure 1-2 Floor Plan of the Loma Linda Proton Treatment Center [3]. ............................5
Figure 1-3 A Medical Imaging Informatics approach towards development of
decision-support tools for the DICOM-RT based ePR system. The final
results are add-on features for the ePR system to provide decision-
support for new patient cases. This methodology can be applied to
different lesion types as well as treatment types to quickly research and
develop new decision-support tools [2]. ............................................................9
Figure 2-1 Flow chart illustrates proton therapy workflow. Yellow boxes indicate
the data (both DICOM objects and data forms) that are generated within
the workflow. (a) Workflow steps from Consultation to Treatment
Planning Phase; (b) Workflow steps from Treatment Delivery to
Follow-up Phase. (DVH = dose-volume histogram, DRR = digital
reconstructed radiograph, QA = quality assurance, CT = computed
tomography, TPS = treatment planning system, DICOM RT =
radiotherapy DICOM object, DICOM RT-ION = radiotherapy ion,
including protons and heavy ionized particles beam, DICOM object.) ...........13
Figure 2-2 RT object dependency of planning CT images ...............................................19
Figure 3-1 Data Model consisting of three different data types: images, RT and
RT-ION objects, and clinical data forms .........................................................25
Figure 3-2 The Architecture of the Proton Therapy ePR and Its Data Flow Model.
The data flow depicts roadmap of how data can be integrated into the
proton therapy ePR system. (A) The Data Gateway use to receive data.
(B) The ePR Server has the storage and database to archive and quantify
stored data. (C) The Decision Support Tools give users the ability to
analyze and interact with treatment plans and patient data to improve
treatment for future patient. (D) The Visualization Module contains all
GUI for user interaction in data acquiring, data storage and display and
decision support tools. KB: knowledge base, GUI: graphical user
interface ............................................................................................................30

xi

Figure 3-3 DICOM Database Schema. This schema shows relation following the
DICOM Standard. Each Patient has 1-n studies, each study has 1-n
series, each series can contain one or more objects of diagnosis image,
RT Image, RT Structure Set, RT Dose, RT-ION Plan and/or RT-ION
Treatment Record. ............................................................................................33
Figure 3-4 Entity-Relationship Example of a Sample Knowledge Base for a
Clinical Scenario to perform treatment plan assessment. The classes
defined have attributes that are extracted from the standardized
DICOM-RT and DICOM-RT-ION data integrated in the ePR system.
Each class carries a Primary Key (PK) Identifier, and can contain
Foreign Keys (FK) which link it to another class object. ................................34
Figure 3-5 Screen shot of a Patient Folder displaying patient data in 7 tabs. ...................37
Figure 4-1 Modality Simulator .........................................................................................41
Figure 4-2 Prostate Initial Data Form of patient treated hypofractionation protocol
treatment ..........................................................................................................43
Figure 4-4 Follow-up Data Form: Toxicity Data Sheet ....................................................46
Figure 4-5 DICOM Data Directory Hierarchy of File System of the ePR Server ............48
Figure 5-1 Dose volume histogram (DVH) curves in a prostate treatment plan (red-
prostate; yellow-bladder; green-rectum.) .........................................................55
Figure 5-2 Features of the ITPN. (1a) Navigation along single DVH curve
navigation along x-axis and y-axis using arrow buttons; (1b) Multiple
DVH curves navigation; (2) Display 2D CT Images with contours and
isodose curve overlaid; (3) Display quantitative measurement of
overdose volume in critical structures; and (4) Display 3D surface of
chosen critical structure, dose distribution;. (Orange-bladder; Green-
rectum; Pink-prostate; Brown-isodose surface). ..............................................58
Figure 5-3 Prototype GUI of ITPN with four regions: (A) DVH display region; (B)
2D image display with contours and isodose curves overlaid; (C)
Quantitative measurements display; and (D) 3D rendering image
display; .............................................................................................................59
Figure 6-1 Dose and Volume Knowledge extracted from DICOM RT objects ...............62
Figure 6-2 Example of dose volume histogram (DVH) curves in a prostate
treatment plan (red-prostate; yellow-bladder; green-rectum.) The green
O shows that the dose of 42.50 Gy is delivered to maximum 14.49%
volume of the rectum. ......................................................................................63
xii

Figure 6-3 Example of Toxicity Knowledge. The weekly toxicity grade is the
worst (highest) grade in five categories. The total worst toxicity grade is
the worst grade of four weekly grades. In this example, the worst
toxicity grade is 2. ............................................................................................66
Figure 6-4 Knowledge for Toxicity Grade. Top: ctcaemap table structure contains
the information published in CTCAE v.3 to evaluate radiation toxicity
grade. Bottom: examples of knowledge data stored in ctcaemap table in
ePR database. ...................................................................................................67
Figure 6-5 Knowledge base database schema example combining toxicity and
treatment plan information ...............................................................................68
Figure 7-1 Screenshot of database tables of the ePR System for prostate cancer
patients; (A) Tables of DICOM data, (B) Tables of user management,
(C) Tables of DICOM extraction knowledge, and (D) Tables of clinical
data and knowledge discovery. ........................................................................71
Figure 7-2 Testing configuration at IPILab ......................................................................73
Figure 7-3 Verification of DICOM Data ..........................................................................74
Figure 7-4 Non-DICOM data was verified at the ePR server database ............................75
Figure 7-5 Initial data displayed on Firefox browser ........................................................76
Figure 7-6 Screenshot of the ITPN with DICOM data displayed on Firefox
browser .............................................................................................................77
Figure 8-1 Existing proton therapy dataflow in the research of hypofractionation
protocol for selected patients at Loma Linda University Medical Center.
The arrows at the bottom depict the time required for each step and
information extraction. .....................................................................................82
Figure 8-2 New improved hypofractionation dataflow in research with ePR
deployed. The ouput is the aggregated data in the database of the ePR
system (item 5) which can be easily accessed by researchers/ clinicians
at the web-client workstation. The arrows at the bottom depict the time
required for each step which is instantly. .........................................................84
Figure 8-3 Screenshot of Initial Data Form filled in electronically inside white
rectangles. ........................................................................................................86
Figure 8-4 Screenshot of Follow-up Form filled in electronically inside white
rectangles. ........................................................................................................86
xiii

Figure 9-1 Existing hypofractionation workflow consisted of 4 main workflow
steps. Steps 1-4 are current general proton therapy clinical workflow.
Two additional steps are added in the hypofractionation research
workflow, steps 1a and 2a, to ensure the patient is in low and immediate
group risks. .......................................................................................................91
Figure 9-2 Enhanced hypofractionation workflow with ePR system. Blue boxes:
Existing hypofractionation protocol workflow. Yellow boxes: additional
workflow steps with ePR implemented. Steps 2b, c, and d: Additional
steps to utilize ePR system for decision support. .............................................93
Figure 9-3 Screenshot of Patient New’s information, in red boxes and circles,
registered to the ePR to determine his eligibility for hypofractionation
protocol. ...........................................................................................................96
Figure 9-4 Screenshot of Pt. New criteria assessment result that he satisfied the
criteria. .............................................................................................................97
Figure 9-5 Step-by-step guideline of the clinical example ...............................................97
Figure 9-6 Screenshot of knowledge-based search engine query input for Pt. New
(see red ellipses and boxes in Figure 9-3) ........................................................99
Figure 9-7 Screenshot of knowledge-based search results displaying previous
patients that have undergone new hypofractionation treatment protocol
who have toxicity grades ranging from 0 to 3 ...............................................100
Figure 9-8 Difference in the shapes of the prostates, red contours, of Pt. New (top)
and Pt. Twenty-One (bottom) ........................................................................102
Figure 9-9 Screenshot of beam margin configuration of Pt. Demo showing the
beam margin of 10mm around prostate. ........................................................103
Figure 9-10 Screenshot of ITPN shows overlapping region of the isodose curve
95% (orange), target prostate (pink) and crtical organs: rectum (green)
and bladder (yellow). The isodose 95% covered entire prostate. ..................105
Figure 9-11 Screenshots DVH curves on the ITPN of Pt. New shows 100% volume
of prostate receiving 55Gy but a decreasing volume percentage of
rectum receiving 55Gy from 25.70% to 18.18% with the modified
hypofractionation plan of 7mm beam margin was used. ...............................107

xiv

Abbreviations
AJCC American Joint Committee on Cancer
CSS Cascade Stylesheet
CT Computed Tomography
CTCAE Common Terminology Criteria for Adverse Events v3.0
CTV Clinical Tumor Volume
DB Database
DICOM Digital Imaging and Communications in Medicine
DRR Digital Reconstructed Radiograph
DVH Dose Volume Histogram
ePR Electronic Patient Record
EBRT External Beam Radiotherapy
GTV Gross Tumor Volume
GUI Graphical User Interface
HIPAA Health Insurance Portability and Accountability Act
HTTPS Hypertext Transfer Protocol Secure
ITPN Intelligent Treatment Plan Navigator
IPILab Image Processing and Informatics Laboratory
IRB Institutional Review Board
KB Knowledge Base
LLUMC Loma Linda University Medical Center
NCI National Cancer Institute
NAPT National Association for Proton Therapy
xv

MS Modality Simulator
MV Megavolt
OAR Organ at Risk
PACS Picture Archiving and Communication System
PTV Planning Tumor Volume
PSA Prostate Specific Antigen
PT Proton Therapy
QA Quality Assurance
ROI Region of Interest
RT Radiation Therapy/Radiotherapy
TP Treatment Plan
TPS Treatment Planning System
USC University of Southern California
WS Workstation

xvi

Abstract
Radiotherapy (RT) is an image-intensive specialty in medicine that utilizes radiation for
treatment of cancer. Similar to x-ray beam treatment, proton therapy (PT) utilizes high
energy particle proton beam to kill cancer cells at the target region. Due to the physical
properties of the proton beam, PT delivers dose with higher accuracy than conventional
radiotherapy with x-ray beams and no exit dose. In PT, patient data are scattered in
different systems within the oncology department which hinders the research on many
topics such as whether PT with higher dose or different delivery schedules is more
beneficial to reducing toxicity of prostate cancer patients. This research aims to introduce
a data mining method and a treatment plan navigator utilizing the infrastructure and data
repository of a PT electronic patient record (ePR) system that can ultimately help to
minimize radiation toxicity in prostate cancer patients and improve patient outcomes in
future proton therapy treatment.
The workflow of a proton therapy treatment in oncology department was reviewed, and a
clinical data model and data flow was designed. A prototype PT ePR system compliant
with Digital Imaging and Communications In Medicine (DICOM) was developed to store
prostate cancer patient images, treatment plans and related clinical data. The ePR system
consists of four main components: 1) Data Gateway; 2) ePR Server; 3) Decision Support
Tools; and 4) Visualization and Display Tools. The decision support tools and
visualization tools are currently developed based on DICOM images, DICOM-RT and
DICOM-RT-ION objects with collaboration with James M. Slater, M.D. Proton
Treatment and Research Center, Loma Linda University Medical Center (LLUMC) and
xvii

its oncologists and physicists. The data from prostate cancer patients treated with
hypofractionation protocol proton therapy at LLUMC was used for evaluating the
effectiveness of the ePR system. Each patient dataset includes a set of computed
tomography (CT) DICOM images and four DICOM RT and RT-ION objects. In addition,
clinical outcomes data collected from PT cases were included in the overall database to
form a knowledge base for future outcomes analysis. A data mining search engine and an
intelligent treatment plan navigator (ITPN) has been developed and integrated with the
ePR system to utilize the knowledge base and data for decision support, which was
evaluated using a hypothetical patient.
In general, the ePR system with centralized data shows promise in its ability to facilitate
the proton therapy workflow. The scope of the PT ePR system in this dissertation is for
prostate cancer patient treated with hypofractionation protocol in proton therapy. The
system was evaluated with 41 patients collected for this protocol. The goals of my
research in the ePR System for prostate cancer patients are: (1) design a system based on
the above specifications tailored to proton therapy, (2) to improve efficiency in data
collection and integration to facilitate outcomes analysis with this system and (3) to
utilize knowledge extracted from the collected data to help clinical decision making
process for a better treatment plan of prostate cancer patient treated with new
hypofractionation treatment protocol.

1

Chapter 1 Introduction
1.1 Prostate Cancer
Prostate cancer is a type of cancer that forms in tissues of the prostate (a gland in the
male reproductive system found below the bladder and in front of the rectum). Prostate
cancer usually occurs in older men.
The National Cancer Institute (NCI) estimates that there will be 192,280 new cases and
37,360 deaths from prostate cancer in the United States in the year 2009 [1].
1.1.1 Available Treatments
When a patient is diagnosed with cancer, there are multiple ways to decide what
treatment will be best for the patient. These include surgery, chemotherapy and ionized
radiation beam. Usually, the treatment is carefully chosen using one or a combination of
two or three treatment types.
Men with prostate cancer have similar treatment options but these options vary between
patients. The right treatment depends mainly on age, the cancer pathology condition or
tumor grade, which is quantified using a Gleason score [2], the number of biopsy tissue
samples that contain cancer cells, the stage of the cancer, symptoms, and general health.
Radiation therapy is a treatment option for men with any stage of prostate cancer. In
addition, men with early stage prostate cancer may choose radiation therapy instead of
surgery. It also may be used after surgery to destroy any cancer cells that remain in the
area. In later stages of prostate cancer, radiation treatment may be used to relieve pain.
2

Radiation therapy (also called radiotherapy) uses high-energy rays to kill cancer cells.
The radiation is planned to maximize the dose delivered to the tumor target while
minimizing the radiation to other healthy tissue. Usually, two types of radiation therapy
are used to treat prostate cancer with some men receiving both types. The two types are:
• External radiation: The radiation comes from a large machine outside the body.
The types of external beam include x-ray beam (or conventional radiation
therapy), proton beam, and heavy particle beam (i.e. carbon ion beam). These
types of treatment utilize computer systems called treatment planning systems
(TPS) to estimate the delivery of radiation dose to more closely target the cancer
and lessen the damage to healthy tissue near the prostate.
• Internal radiation (implant radiation or brachytherapy): The radiation comes from
radioactive material usually contained in very small implants called seeds. Dozens
of seeds are placed inside needles, and the needles are inserted into the prostate.
Once the needles are removed, the seeds are left behind. These seeds give off the
radiation for months and will not need to be removed once the radiation is gone.
My area of research focuses on the external proton beam treatment method.
1.2 Proton Therapy
1.2.1 Physics & Treatment Centers
Proton Therapy (PT) is a particular treatment that utilizes energized charged particles,
protons, to deliver dose to the target region. Protons are energized to specific velocities
which determine where they will deposit maximum energy within the body to destroy
cancerous cells, allowing for maximum dose to the target region while minimizing dose
3

to surrounding tissues. This is due to the fact that the proton depth dose (Bragg Peak) is
inversely proportional to the square of the particle velocity. In comparison to proton,
photon depth dose is proportional to an exponential function. Figure 1-1 shows a
comparison between 6megavolt (MV) Photons from a Linear Accelerator utilized in
traditional radiation therapy and different energy Protons. Each of the different energy
protons have minimal dose in water but deposit their maximum dose at a target depth.
This translates to less dose to normal healthy tissue in the body while depositing most of
the energy within the target tumor located at a certain depth within the body. In addition,
proton beams have no exit dose which also minimizes damage to healthy tissue beyond
the target tumor. Proton Therapy is especially effective for types of cancer that require
controlled high concentration dose and tumors that are close to sensitive tissue. Some
examples of the types of cancer treated include: prostate, brain, spinal cord, head and
neck, base of skull, eye, lung, and tumors in children. [3]
According to the National Association for Proton Therapy (NAPT), there are only a total
of 12 proton centers in the US as of April 2010. Table 1-1 shows the working status on
the left column, and names and locations of these centers on the right column. [4]

4

Figure 1-1 Graph showing a comparison between Photon Depth Dose (6MV) and Proton
Depth Dose (130, 150, 170, 190 MeV) percentages as measured in water. The
maximum dose from the proton can be deposited at a certain depth. This can be
utilized to treat cancer tumors at a certain depth while minimizing dose to
surrounding normal healthy tissue [2].
Table 1-1 Status and Name & Location of Proton Therapy Facilities
Working Status Name and Location of Proton Therapy Centers
Currently
Operating
James M. Slater, M.D. Proton Treatment and Research Center at
Loma Linda University Medical Center
Francis H. Burr Proton Center at Mass. General Hospital
Midwest Proton Radiotherapy Institute at Indiana University
The University of Florida Proton Therapy Institute
M.D. Anderson Cancer Center's Proton Center, Houston
ProCure Proton Therapy Center, Oklahoma City, located at the
INTEGRIS Cancer Campus
The Roberts Proton Therapy Center at University of Pennsylvania
Health System
Under Construction Hampton University Proton Therapy Institute
Northern Illinois University Proton Therapy Center
CDH Proton Therapy Center, a ProCure Center, Warrenville, Illinois
ProCure Proton Therapy Center in partnership with Princeton
Radiation Oncology Group and CentraState Healthcare System
In Development South Florida Proton Center
There are few PT sites across the country due to the complex and expensive system
requirements which include a synchrotron, linear accelerator, and large rotating gantry
and require a very large square area footprint. Figure 1-2 shows the floor plan of the
5

Proton Treatment Center located at Loma Linda University Medical Center (LLUMC)
[4].

Figure 1-2 Floor Plan of the Loma Linda Proton Treatment Center [3].
LLUMC was opened in 1990 as the first national hospital-based proton treatment center
for cancer and named the James M. Slater, M.D. Proton Treatment and Research Center
in 2007. Since 1990, it has treated more than ten thousand patients, of which men with
prostate cancer comprise approximately 65% of all patients treated with protons at
LLUMC; this population represents the largest cohort of patients treated with protons for
prostate cancer anywhere in the world [5]. My research is in collaboration with Dr. Jerry
D. Slater and his group at LLUMC.
1.2.2 Radiation Treatment Side Effects
The negative treatment side effects depend mainly on the dose and type of radiation. The
patient is likely to be very tired during radiation therapy, especially in the later weeks of
6

treatment. If a patient has external radiation, he may have diarrhea or frequent and
uncomfortable urination. Some men have lasting bowel or urinary problems. The skin in
the treated area may become red, dry, and tender. Internal radiation therapy may cause
incontinence. These side effects usually go away. However, both internal and external
radiation and its associated side effects can cause impotence. All of these side effects are
generally called radiation toxicity.
1.3 Current Challenges in Proton Therapy for Prostate Cancer
Uncertainty of dose and treatment schedule
Conventional x-ray radiation therapy, given in sufficient doses, can control many cancers.
However, because of the difficulty for physicians to adequately conform the irradiation
pattern to the cancer, healthy tissues may receive a similar dose and can be damaged.
Consequently, a less than desired dose is frequently used to reduce damage to healthy
tissues and to avoid unwanted side effects. On the other hand, in proton therapy, the
treatment usually has greater tumor control than conventional radiation therapy. The
power of protons is that higher doses of radiation can be used to control and manage
cancer while significantly reducing damage to surrounding healthy tissue and vital
organs. Unfortunately, the tumor is not just a separate target that can be irradiated without
side effects to other healthy tissues of nearby critical organs. In most cases, the tumor is
in a position that is very close to other critical organs so that the acceptable margin is
relatively small. Therefore, in close proximity, these critical organs are still exposed to a
certain amount of radiation dose.
7

Similar to conventional radiation therapy, PT has a long and established role in the
curative treatment of organ-confined prostate cancer [6]. However, the optimal radiation
dose and treatment schedule still remains unknown [7].
High cost of proton treatment
Despite the superior physical characteristics of proton which allows the beam to radiate
precisely to specific tumor targets without significant exit dose and minimizing damage
to healthy tissue near the tumor, PT has been limited in popularity among other types of
radiotherapy due to the complexity and cost of proton facilities.
Data scattered among many systems
In general, there are different stand-alone and proprietary information systems in a proton
therapy department. Each system has its own data “storage area” and proprietary data
formats. Typically, the treatment plans are stored in treatment planning system (TPS),
treatment records stored in another information system or patient chart, and digital
images are stored in separate digital imaging devices. The patient medical history and
initial treatment data are usually stored in an oncology information system. The patient
outcomes data are usually collected and stored in another information system which is
not readily available to physicians or researchers. Currently, all of these data is “linked”
by paper records or patient charts. Moreover, it is common practice that the hardcopy
film images are usually stored separately in the film library. Therefore, if the paper record
is lost, the patient’s treatment information is lost. Even more disturbing is that the
treatment plans from an old TPS cannot be retrieved for review after a system upgrade
making historical review challenging.
8

Challenges in data comparison and outcomes analysis between multiple treatment
protocols
Proton therapy along with other radiation treatment methods face common problems of
disparate data as well as a lack of information technology readiness. Archiving,
visualizing, and distributing treatment plans and patient outcomes is a big challenge for
clinicians and researchers in the field. They require a system that can collect all relevant
information for proton therapy treatment of prostate cancer. This system consolidates and
makes centrally accessible all cancer patients’ treatment related information. This
information can be used to compare treatment plans and outcomes results of patients of
the same treatment protocol or different treatment protocols, which ultimately improves
treatment planning for new patients.
1.4 Proton Therapy ePR System
The electronic patient record (ePR) is an emerging concept to provide support for the
hospital or clinical-based healthcare information system. The ePR combines image data
from an image repository called Picture Archiving and Communication System (PACS)
with other information systems and clinical services.
An ePR consists of five major functions: 1) Accepts direct digital input of patient data; 2)
Analyzes across patients and providers; 3) Provides clinical decision support; 4) Supports
outcomes analysis and patient and physician profiling; 5) Distributes information across
different platforms and health information systems [8].
The research of the Proton Therapy ePR System follows the medical imaging informatics
methodology, shown in Figure 1-3 [5], to develop the knowledge base, the data mining,
9

and quantification and visualization tools which ultimately become add-on features to the
Proton ePR system. This methodology can be applied to different lesion types as well as
treatment types to quickly research and develop new decision-support tools.

Figure 1-3 A Medical Imaging Informatics approach towards development of decision-
support tools for the DICOM-RT based ePR system. The final results are add-on
features for the ePR system to provide decision-support for new patient cases.
This methodology can be applied to different lesion types as well as treatment
types to quickly research and develop new decision-support tools [2].
The ePR consists of a data model derived from existing clinical workflow, a database, a
web server and a graphical user interface. Additionally, the ePR provides storage and a
method to collect all relevant patient data, which equips the clinician with tools to
improve treatment planning and achieve the goal of minimizing radiation toxicity in
prostate cancer patients while ultimately improving the patient’s quality of life. There is
no such current system available or developed for proton therapy.
1.5 Aim and Scope
An ePR system with centralized PT related data can facilitate the efficiency of proton
therapy workflow by providing clinicians the ability to view imaging studies, evaluate
proton treatment plans and explore treatment outcomes. Specifically, the PT ePR can help
10

to minimize radiation toxicity in prostate cancer patients through the utilization of data
mining and treatment plan navigation tool.
Based on the medical imaging informatics methodology, shown in Figure 1-3, the Proton
Therapy ePR System has been developed to be an accurate repository for all patient data
with the specifications shown in Table 1-2.
Table 1-2 Proton ePR System Specifications
Items Descriptions
Hardware Computer server that can support web-based medical
imaging intensive applications
Software Web-based server-client architecture
Communication/
Security
DICOM*, HTTPS**
Data types Treatment plans, patient clinical data
Compliance DICOM*, HIPAA***
Features Data mining search engine, treatment plan navigator
* Digital imaging and Communications in Medicine
** Hypertext Transfer Protocol Secure
*** Health Insurance Portability and Accountability Act
The scope of the PT ePR system in this dissertation is for prostate cancer patient treated
with hypofractionation protocol in proton therapy. The system is evaluated with 41
patients collected for this protocol.
The goals of my research in the ePR System for prostate cancer patients are: (1) design a
system based on the above specifications tailored to proton therapy, (2) to improve
efficiency in data collection and integration to facilitate outcomes analysis with this
system and (3) to utilize knowledge extracted from the collected data to help clinical
11

decision making process for a better treatment plan of prostate cancer patient treated with
new hypofractionation treatment protocol.
1.6 Chapter Summary
The development of the ePR with decision support tools involves (1) clinical workflow
analysis, (2) system architecture and design, (3) data collection for knowledge discovery,
(4) decision support tools design, and (5) system implementation, testing and evaluation.
Following this direction, the chapters are arranged as follows:
Chapter 2: Proton Therapy Clinical Workflow and Data
Chapter 3: Proton Therapy ePR System
Chapter 4: Data Collection
Chapter 5: Intelligent Treatment Plan Navigator
Chapter 6: Data Mining Using Knowledge Driven Search Engine
Chapter 7: System Implementation
Chapter 8: Data and System Integration for Improving Patient Outcome Study
Chapter 9: Knowledge Discovery for Decision Support in Treating Prostate Cancer
Patient with Hypofractionation Protocol Proton Therapy
Chapter 10: Current Status and Future Direction

12

Chapter 2 Proton Therapy Clinical Workflow and Data
2.1 Introduction
Using high-energy protons for medical treatment was first proposed in 1946. Proton
therapy (PT) was first used as a method of radiotherapy for cancer less than 10 years.
There are only a dozen of proton therapy centers in comparison to the thousands of
conventional x-ray and brachytherapy radiotherapy centers nationwide [9]. In general, it
is believed that proton therapy provides effectiveness of a superior and more advanced
form of radiation treatment resulting in less morbidity with minimal to no side effects.
For prostate cancer, proton therapy has been proven to deliver high doses to the prostate
tumor to achieve high levels of tumor eradication without any substantial increase in
morbidity seen with conventional therapy or with other forms of high-dose therapy; and
reduce volume of pelvic tissue receiving lower does (<40Gy). The immediate benefits are
not substantial but the long-term benefits may prove to be [10].
2.2 Proton Therapy Workflow
Figure 2-1 shows the general clinical workflow model for proton therapy at the James M.
Slater, M.D. Proton Treatment and Research Center (LLUMC). This workflow shows
steps after the patient has been diagnosed with cancer, consulted with the physicians,
discussed available treatments, and determined to treat the tumor with proton therapy.
The workflow describes general procedures involved in the entire treatment from when
the patient enters the oncology department for consultation to follow-up after the
treatment for prostate cancer.
13

(a) Workflow steps from Consultation to Treatment Planning Phase
Figure 2-1 Flow chart illustrates proton therapy workflow. Yellow boxes indicate the
data (both DICOM objects and data forms) that are generated within the
workflow. (a) Workflow steps from Consultation to Treatment Planning Phase;
(b) Workflow steps from Treatment Delivery to Follow-up Phase. (DVH = dose-
volume histogram, DRR = digital reconstructed radiograph, QA = quality
assurance, CT = computed tomography, TPS = treatment planning system,
DICOM RT = radiotherapy DICOM object, DICOM RT-ION = radiotherapy ion,
including protons and heavy ionized particles beam, DICOM object.)
14

(Figure 2-1 Continued)

(b) Workflow steps from Treatment Delivery to Follow-up Phase

The following paragraphs explain the workflow in more detail with numbered workflow
steps corresponding to Figure 2-1:
1. The patient enters the oncology department to schedule treatment.
15

2. The physician consults with the patient about the treatment, including an
explanation of the treatment duration, steps involved and side effects. For
prostate cancer, the total dose is delivered into many fractions, one fraction per
day; this makes the duration of the treatment ranging from 7 to 8 weeks.
3. The patient is scheduled for body immobilization. At LLUMC, a plastic cradle is
created with injected foam cushioning to immobilize the body for imaging,
planning, and treatment.
4. A diagnostic CT is acquired to plan the treatment with the patient immobilized
within the plastic cradle.
5. Staff loads the CT images to treatment planning system (TPS).
6. The oncologist prescribes the doses, and contours the tumor or prostate regions
and critical structures such as the rectum and bladder.
7. The dosimetrist or physicist defines the initial treatment plan based on the
oncologist’s dose prescription. Once the initial plan is computed, the oncologist
reviews the results and requests any necessary changes for the dosimeters and
physicist to make.
8. The planning and review process may be an iterative process until the plan is
finally approved by the oncologist.
9. Once the treatment plan has been approved, QA and setup is performed and a
simulated treatment plan will be executed in order to make any fine-tune
adjustments to the overall plan.
10. After the QA procedure is completed, the tumor volume structure and beam shape
are exported to a manufacturer to build a 3D computer assisted bolus made of
16

high-grade wax and apertures made from an alloy called cerrobend that will shape
the proton beam conformal to the target tumor(s).
11. Before the actual radiation treatment commences, the accuracy of the treatment
plan in terms of field sizes, setup, shielding positions needs to be verified at the
proton treatment room. For such verification, a portal image is obtained at the
proton beam gantry, either as a film image or with a digital image device installed
in the treatment room as a digital portal image. The image is then compared with
the reference DRR or simulator image.
12. The PT treatment session is then executed by the radiation therapist within the
gantry and the corresponding PT plan data are stored in the treatment planning
systems. The dose of each treatment is recorded into the patient chart.
13. After the treatment is completed, all patient treatment results and side-effects will
be recorded in the patient follow up. Follow-up data forms will be collected at
each post radiation therapy session in a follow-up visit which occurs every three
months for two years, then every six months for up to three years. The prostate
specific antigen (PSA) levels and toxicity grades are reported in these follow-up
forms.
As shown in Figure 2-1, the yellow boxes indicate data generated and collected from the
proton workflow. The data includes both textual data and DICOM objects, which is
defined by the Digital Imaging and Communications in Medicine (DICOM) Standard.
Unless all equipment used in a department is from a single software vendor, it is
necessary that images and other data should be transferable between different computing
systems in a standardized fashion so that it can be interpreted by different software
17

applications. The DICOM standard facilitates interoperability of systems claiming
conformance in a multi-vendor environment and has attained the general acceptance as
the standard for such data transfer in medical imaging and informatics. In addition to the
DICOM standard formats for diagnostic images, such as CT, MR, US, etc.; the DICOM
Standard has provided support for the transfer of radiotherapy-related data between
devices found within and outside a radiotherapy department by introducing seven
DICOM-RT objects for x-ray radiotherapy (Supplement 11, 1996 and Supplement 29,
1998) and two more DICOM-RT-ION for proton therapy or ion therapy in general
(Supplement 102, 2004) [11, 12]. These will be discussed in more detail in the following
section.
2.3 Proton Therapy Data Objects: DICOM–RT and DICOM–RT–ION
Objects
2.3.1 RT Structure Set
To carry out radiotherapy treatment planning, the target tissue and organs at risk (OAR)
are defined. This process of segmentation of the tomographic images or drawing contours
of target tumor and OARs leads to a set of structures, which are defined by the DICOM
RT structure set object. The data structure allows for different contour types, including
the body or external contour, avoidance structures or OARs and the target planning tumor
volume (PTV), clinical tumor volume (CTV) or gross tumor volume (GTV). Note that
PTV, CTV and GTV terms are used more often in X-ray treatment and brachytherapy
treatment plans. These structures have special meanings that must be maintained across
the different computing systems. An RT Structure Set is linked to an image (see Figure
2-2 in Section 2.3.6) from which the original contours were derived. In addition to
18

contours, the standard allows the definition of markers and isocentres, although
interpretation of these may not be the same.
2.3.2 RT Dose
Treatment planning systems calculate the radiation dose distribution as a matrix of points
with associated doses. These dose grid files are supported in the DICOM RT dose object.
Definitions also exist in the DICOM RT dose specification to store relationships between
dose and structures through dose volume histograms and dose region of interest (ROI)
statistics. Isodose curves may also be defined. These functionalities are slowly being
implemented by the manufacturers but is not available in the newest version of the TPS at
LLUMC and not heavily used among other proton therapy TPS.
2.3.3 RT Image
The RT Image object addresses the requirements for image transfer found in general
radiotherapy applications performed on conventional simulators, virtual simulators, and
portal imaging devices. Such images may either be acquired directly from the device, or
digitized using a film digitizer. Numeric beam data parameters may also be recorded with
the image, indicating the parameter values at the time the image was taken or created.
2.3.4 RT ION Plan
The RT Ion Plan addresses the requirements for transfer of treatment plans generated by
manual entry, a virtual simulation system, or a treatment planning system before or
during a course of proton therapy treatment. Such plans may contain fractionation
information, and define proton beams.
19

2.3.5 RT ION Beams Treatment Record
The RT Ion Beams Treatment Record addresses the requirements for transfer of treatment
session reports generated by a treatment verification system during a course of proton
beam treatment, with optional cumulative summary information. It may also be used for
transfer of treatment information during delivery.
Although not in widespread use, the DICOM RT ION treatment record object completes
the list of available data objects under DICOM RT and DICOM RT ION for proton
treatment. It is provided for completeness and for the purposes of future compatibility. It
is more usual for this type of information to be interfaced directly with the electronic
patient record (ePR).
2.3.6 DICOM RT and DICOM RT ION Dependencies
In the radiotherapy process, a number of DICOM objects are created. There is a system
of dependencies in which each DICOM RT object contains references the DICOM
objects which preceded it. Figure 2-2 shows an example of this dependency.

Figure 2-2 RT object dependency of planning CT images
20

2.4 Treatment Protocols
External radiation beam, including both x-ray and proton beam, has been used to treat
more than 26,000 men in the United States annually [13]. Since the patient treated with
conventional doses receive the total dose in the range of 60-71 Gy, many studies suggest
that conventional dose is unable to eradicate prostate cancer in a significant proportion of
cases, with a resultant increase in prostate-specific antigen (PSA) levels, secondary
treatment, and, ultimately, clinical recurrence [14-16]. Currently, two treatment
protocols are being studied for PT of prostate cancer: escalation dose protocol and
hypofractionation dose protocol. The former approach delivers higher dose to prostate
gland in small dose-per-fraction, which is the same as the conventional dose rate. It is
currently in clinical use. The latter method exposes fewer doses and fractions to the target
tumor but in larger dose-per-fraction as compared to escalated dose treatment. This
protocol is currently being evaluated and not in full clinical use. These two protocols will
be discussed in further detail in the next sections.
2.4.1 Dose Escalation Protocol
In current practice, it is believed that higher radiation doses to tumor result in better rates
of local tumor control, but the higher doses to normal tissues increase the risk for
complication and side effects. Therefore, much research and technology have been
focused on technological developments to optimize therapeutic ratio by maximizing
tumor dose and minimizing normal tissue dose [17]. Similar in prostate cancer,
improving therapeutic ratio is to maximize dose to the prostate cancer and minimize dose
to nearby healthy tissue, particular rectum and bladder tissues.
21

Studies suggest a benefit of dose escalation for the treatment of prostate cancer,
especially for patients with a prostate-specific antigen (PSA) level ≥10 nanograms per
milliliter (ng/mL) [16-19]. However, higher dose carries a potential risk of greater rates
of chronic toxicity.
Typically, patients treated with external beam radiation therapy (EBRT) receive daily
dose treatment, also called dose-per-fraction, in five days per week for a total duration of
7-8 weeks [18]. Recent studies strongly suggest that radiation doses in the range of 78-81
Gy result in an improvement in five year chemical freedom relapse, even in “low risk”
patients [19, 20]. At LLUMC, current institutional standard for early-stage disease is to
treat the prostate and immediately adjacent paraprostatic tissues with conformal proton
beams alone to a radiation dose of ~81 CGE [Cobalt-Gray Equivalent, in which 1 proton
Gy is assumed to equal 1.1 photon Gy], delivered one fraction, 1.8 Gy, of total dose each
day, five days a week over a period of nine weeks. The prolonged course, although well
tolerated, can lead to logistical difficulties in terms of requiring patients (many of whom
reside outside of the immediate area) to remain at LLUMC for periods of up to three
months; this can serve as an absolute treatment barrier to some patients who would
otherwise be candidates for conformal proton beam therapy but who are unable to remain
at LLUMC for such a prolonged treatment course. A reduction in treatment time could
therefore allow more patients to take advantage of this treatment modality that is
currently available. But for now, this protocol is implemented in the clinical environment.

22

2.4.2 Hypofractionation Protocol
Hypofractionation protocol is defined as an increase of daily dose or dose-per-fraction to
decrease the number of fractions per course of treatment. Recent studies suggest that it is
possible to increase the dose-response of tumors and normal tissue to fractionated
radiotherapy by employing unconventional fractionation schedules. The use of
hypofractionation has a long and generally successful history in conformal proton beam
therapy. Its use has been predicated by both the superior physical characteristics of
protons as compared to x-rays (i.e., low entrance dose, sharply defined high dose area,
lack of exit dose).
Currently, LLUMC is developing a hypofractionation protocol for treatment of a prostate
cancer patient, in which the number of fractions will be reduced to half by increasing the
dose-per-fraction to nearly double at 3Gy per fraction of 60Gy total dose delivered over a
period of four weeks. This protocol promises to improve treatment time for patient
convenience and may drop the overall cost of treatment by up to 50% without increasing
treatment related toxicity [21].
The treatment with hypofractionated regimens (shorter overall treatment time, larger
treatment fractions) may prove to be more effective than conventional fractionation in
proton beam treatment for prostate cancer. Therefore, my research to develop the ePR
system can facilitate data collection and patient outcomes analysis for this new
hypofractionation treatment protocol.

23

2.5 Clinical Outcomes of Prostate Cancer Patient Treated with Proton
Therapy
There are two types of clinical outcomes collected for prostate cancer patients: local
tumor control outcomes and radiation toxicity.
2.5.1 Local Tumor Control
The outcomes of local tumor control are based on the survival status, PSA, and
reoccurrence or metastasis after the proton treatment. This information is collected when
the patient comes back for follow-up.
2.5.2 Radiation Toxicity
Radiation toxicity is evaluated based on symptoms of the side effects developed after
proton treatment and frequency or severity of these symptoms. Side effects manifested
within 90 days of initiation of radiotherapy are considered "acute", and "late" for those
side effects manifested thereafter. The Cancer Radiation Therapy Program, National
Cancer Institute has published the Common Terminology Criteria for Adverse Events
(CTCAE) (formerly known as Common Toxicity Criteria) as standards used to grade,
assign attribution and report side effects or radiation toxicity experienced by patients on
clinical trials. The latest version is CTCAE v.3 published on August 9, 2006, including
all Adverse Events applicable to all oncology clinical trials regardless of chronicity or
modality [22]. This research follows the CTCAE standard when discussing radiation
toxicity and developing tools to characterize, quantify, or mine for radiation toxicity
values.
24

For prostate cancer, side effects and symptoms in five categories, such as
dermatology/skin, include gastrointestinal, renal/genitourinary, and sexual/reproductive
function, and constitutional symptoms, are collected at LLUMC to evaluate the radiation
toxicity. In each category, there are many symptoms, in which each is described in five
levels of frequency or severity. Each level is given a grade from 1 to 5, as grade 1 is the
least toxicity and grade 5 is highest toxicity, normally resulting from death of a patient
caused by the symptom. It is generally accepted in many studies that the highest grade
toxicity of any of the nearby organs constitutes the overall radiation toxicity of the patient
due to the PT treatment session. Therefore, my research utilizes this rule when designing
and development tools to mine radiation toxicity outcomes.

25

Chapter 3 Proton Therapy ePR System
The new trend of information technology in medical imaging and informatics is towards
the development of an electronic patient record (ePR), in which all health and medical
information of each patient are organized under the patient’s name and identification
number. Consequently, the patient-centric ePR provides support for the hospital- or
clinical-based healthcare information systems [12]. The proton therapy ePR (PT ePR)
combines PACS image data with other clinical related data utilized in proton therapy
treatment to facilitate further medical research, education, and clinical services.
3.1 Data Model and Standards
3.1.1 The Data Model
Figure 3-1 shows the Data Model developed for the ePR System. This Data Model
depicts the relation of data objects organized for one patient and consists of 3 different
data types: images, RT and RT-ION objects, and clinical data forms.

Figure 3-1 Data Model consisting of three different data types: images, RT and RT-ION
objects, and clinical data forms
26

3.1.2 Standards for Data and Image Distribution
A technical standard is an established norm and requirements in which a set of rules,
formats, instructions and terminologies are used to specify how many devices and/or
software components communicate and exchange information. In the context of the ePR
system, these standards provide the current practice and methods in which patient data
and images are archived, distributed, assessed and viewed in an efficient, convenient and,
most importantly, standardized fashion within the ePR system and with other information
systems within the oncology department.
3.1.2.1 DICOM (Digital Imaging and Communication in Medicine)
DICOM, mentioned in Section 2.3, is the de facto standard for distributing and viewing
any type of medical images regardless of the origin (radiology, oncology, pathology,
endoscopy etc.). The DICOM standard is enforced where possible to ensure the readiness
to interact with other medical devices. The current usage of DICOM standard in this
research includes DICOM communication for data transfer in/out of the ePR system, and
DICOM data objects, shown in Figure 3-1.
3.1.2.2 HTTPS (Hypertext Transfer Protocol Secure)
HTTP is an application-level protocol for distributed, collaborative, hypermedia
information systems. Its use for retrieving inter-linked resources led to the establishment
of the World Wide Web. HTTPS is the combination of HTTP for distributing information
on the network and cryptographic protocol for security. Due to web-based design of the
ePR system, the HTTPS protocol is used to allow user interactions with various
components within the system and other information systems through a web-browser,
27

such as Internet Explorer and Mozilla Firefox, in a streamline and secure manner. This is
especially important for patient privacy and security under the Health Insurance
Portability and Accountability Act (HIPAA).
3.1.2.3 JPEG (Joint Photographic Expert Group), GIF (Graphics Interchange Format), and
PGN (Portable Network Graphics)
Currently, the available web-browsers do not have a universal communication
mechanism for display of native DICOM images and other DICOM-RT and DICOM-RT
objects. Thus, all DICOM image and information in other DICOM objects are converted
to formats which can be interpreted and displayed on the web-browser. JPEG format is
chosen to display DICOM medical images and information. GIF and PNG are mostly
used for icons in the ePR system.
3.2 The ePR Data Flow Model
The design of the data flow model is based on the general proton therapy workflow
described in Section 2.2. As shown in Figure 3-2, this data flow model illustrates
methods and a roadmap for integrating the different data types from proton treatment into
a PT ePR system. Figure 3-2 also shows the four main components of the ePR: Data
Gateway, ePR Server, Decision Support Tools, and Visualization and Display Tools,
which will be described in the following sections.
The following numbered list explains the data flow in relation to the data and main
components of the ePR system. Note that the arrows in Figure 3-2 depict the dataflow
direction and each number of each item in this list refers to the numerical data flow step
labeled next to the arrow. The summary of the general PT workflow is shown on the far
28

left of Figure 3-2. To the right of each workflow phase is the data generated from that
phase.
Input Data
1. The initial data form is collected at the consultation section between the
oncologist and prostate cancer patient. The data in this form will be input into the
Non-DICOM Module residing in the Data Gateway.
2. The CT images and DICOM RT objects, such as RT Structure Set, RT Plan, RT
Dose and RT Image containing DDR Image is sent from the TPS system to the
Data Gateway and received by the DICOM Module. Another RT Image
containing the portal image is sent from the digital portal system, residing in the
treatment, to the DICOM Module.
3. The treatment records are collected from the patient chart and sent to the non-
DICOM module.
4. The follow-up data forms are collected and entered into the Non-DICOM Module
in the Data Gateway.
Data Gateway
5. The Data Gateway GUI shows the status of data acquisition. The GUI is used to
receive commands from physicists/oncologist/clinicians for data query/retrieve
and entry.
6. The Non-DICOM Module converts treatment records to DICOM RT-ION
Treatment Record objects which are transferred to the DICOM Module. Other
29

clinical data from Initial Data Forms and Follow-up Data Forms, controlled by the
Non-DICOM Module, are sent to ePR Server for archival.
7. All DICOM objects from the DICOM Module will be sent to the ePR Server for
archival.
ePR Server
8. DICOM objects and data in Initial Data Form and Follow-up Data Forms are
displayed on the ePR Server GUI. This GUI also shows the status of stored data
and all the processes running on the ePR Server.
Decision Support Tools
The Decision Support Tools are user interactive and consists of the following:
9. The Treatment Plan data is assessed using the Intelligent TP Navigator (ITPN).
10. All the data residing in the ePR can also be accessed through the Knowledge Base
& Data Mining Search Engine.
11. The clinicians can interact with the data and use the decision support tools
through the Decision Support Tools GUI.
30

Figure 3-2 The Architecture of the Proton Therapy ePR and Its Data Flow Model.
The data flow depicts roadmap of how data can be integrated into the proton
therapy ePR system. (A) The Data Gateway use to receive data. (B) The ePR
Server has the storage and database to archive and quantify stored data. (C) The
Decision Support Tools give users the ability to analyze and interact with
treatment plans and patient data to improve treatment for future patient. (D) The
Visualization Module contains all GUI for user interaction in data acquiring, data
storage and display and decision support tools.
KB: knowledge base, GUI: graphical user interface
3.3 The ePR System Architecture
The ePR is a web-based system that complements the current TPS in a radiotherapy
department. Its primary goal is to be integrated in the workflow of the current planning of
cancer cases, thus the ePR system should not be considered as a replacement to any
existing RT application software. The ePR system is designed with four main
components: the Data Gateway, the ePR Server, Decision Support Tools, and
Visualization and Display Tools. The thick-line box in Figure 3-2 shows the architecture
31

of the entire ePR system in which the design has taken into account the capability to add
extra necessary modules in future, symbolized by the (…) in the figure.
3.4 Data Gateway
3.4.1 DICOM Module
The DICOM module allows the ePR to receive DICOM and DICOM-RT objects from
either the PACS or any TPS that can export DICOM-RT objects.
The DICOM Receiver always has the DICOM StoreSCP process continuously available
to receive DICOM object from another DICOM node. It also allows the users to upload
data to the ePR through an upload web GUI. Upon receiving the DICOM object, the
DICOM Receiver transfers it to the ePR Server and triggers the DICOM Extractor to
update the database and obtain the knowledge information.
Query/Retrieve Tool gives the user the ability to query and retrieve DICOM studies from
PACS, TPS or any DICOM storage node to the ePR Server.
3.4.2 Non-DICOM Module
The Non-DICOM module has two components: Text Processing Module and DICOM RT
Converter. The former is designed to handle clinical data in non-DICOM format (text,
Excel spreadsheet, etc.) and the latter to convert data which are defined in DICOM
standard to DICOM objects.
Text Processing module is designed to receive data forms (initial data and follow-up
data) collected from proton therapy workflow, in which the data is not in DICOM
standard format.
32

DICOM RT Converter converts the data, which has been already defined in the DICOM
standard, into DICOM RT objects. For example, the treatment records collected from the
patient chart are readily defined within the DICOM standard and will be converted to
DICOM RT ION Treatment Record objects before archiving into the ePR Server. Patient
clinical information in initial data forms and outcomes data in follow-up data forms do
not have a standardized format nor is it defined in the DICOM Standard. Therefore, it
will not be converted to DICOM objects for now; this information will be kept as text
information in the database.
3.5 The ePR Server
The ePR Server is the primary processor and monitoring engine of the whole ePR system.
It provides the computational power for all modules and services including storage,
database, data quantification module, decision support tools and future extra add-on
modules. The ePR Server also performs tasks to monitor and manage the entire system to
ensure it is working properly. As shown in Figure 3-2, the current ePR Server consists of
three components: Storage/Database, Data Quantification Module and Image Processing
Engine.
3.5.1 Storage/ Database
The data collected in the ePR is archived in File System or the Database. After being
imported to the ePR Server, all DICOM files will be stored in the file system, in which
the files are organized as the following: PatientID\Study ID\Series ID\DICOM Files;
DICOM metadata will be stored in database. Non-DICOM data from clinical data forms
are mainly stored in the database under the same patient identification.
33

The database design consists of two parts: the DICOM schema and the knowledge
schema. The DICOM database schema, as shown in Figure 3-3, follows the DICOM
standard and contains the information from the DICOM header to keep track of the
objects and patient demographics. The knowledge database schema contains information
extracted from DICOM objects and clinical data, as shown in Figure 3-4.
FK fk_SeriesNo
...
Diagnosis Image
PK ImageNo
FK fk_SeriesNo
...
RT Image
PK ImageNo
FK fk_SeriesNo
...
RT Dose
PK DoseNo
FK fk_SeriesNo
...
RT Structure Set
PK StructNo
PK: Primary Key
FK: Foreign Key
RT Object Level
FK fk_PatientNo
...
Study
PK StudyNo
Study
FK fk_StudyNo
...
Series
PK SeriesNo
Series
…
Patient
PK PatientNo
Patient
1 1 1
1, n 1, n
1, n
1, n
1, n
1, n
1, n
1, n
FK fk_SeriesNo
...
RT-ION Plan
PK PlanNo
FK fk_SeriesNo
...
PK RecordNo
RT-ION Treatment
Record

Figure 3-3 DICOM Database Schema. This schema shows relation following the
DICOM Standard. Each Patient has 1-n studies, each study has 1-n series, each
series can contain one or more objects of diagnosis image, RT Image, RT
Structure Set, RT Dose, RT-ION Plan and/or RT-ION Treatment Record.
34

.
Figure 3-4 Entity-Relationship Example of a Sample Knowledge Base for a Clinical
Scenario to perform treatment plan assessment. The classes defined have
attributes that are extracted from the standardized DICOM-RT and DICOM-RT-
ION data integrated in the ePR system. Each class carries a Primary Key (PK)
Identifier, and can contain Foreign Keys (FK) which link it to another class
object.
3.5.2 Data Quantification Module
The main function of the Data Quantification module is to extract and classify data
elements from DICOM objects and update the knowledge database as shown in Figure
3-4. These data elements can be derived from the knowledge database schema which can
be defined into class objects or tables. A few are shown in Figure 3-4 along with their
attributes: Class Object 1) DVH; Class Object 2) Isodose curve; Class Object 3) Critical
Structure; and Class Object 4) CT image. Then, for each of these classes, attributes can
be defined as shown. For example, for the CT Image class object, there are the primary
key (PK) identifier and five attributes: Critical Structure Curve; Isodose Curve; Spatial
Coordinates of the image including x, y, and z-directions; Pointer to the image data; and
DICOM header data. The relationships between each of the class objects are through the
Foreign Keys (FK). For example, in Figure 3-4, Isodose Curve FK1, and Critical
Structure FK2 are related to CT Image object. This is because a CT image would contain
multiple isodose curves and multiple critical structures. Once the knowledge has been
35

defined and knowledge database schema developed, the knowledge can be extracted and
stored within the database. A search engine is built to perform queries on the quantified
knowledge for automatic extraction of particular knowledge to be developed in Chapter
6.
3.5.3 Image Processing Engine
The Image Processing Engine contains libraries for image manipulation as well as image
analysis and a rendering engine. The image manipulation includes the ability to change
window level (contrast and brightness), pan, and zoom in/out of an image. This engine
also provides 3D rendering services and calculates volume of structures included in
DICOM objects.
3.6 Decision Support Tools
The Decision Support Tools utilize the processing modules and data residing inside the
ePR Server as an input for their applications.
The Intelligent Treatment Plan Navigator (ITPN), discussed later in detail in Chapter 5,
uses the Image Processing Engine to display DICOM images in 2-D and 3-D mode
together with quantified knowledge in DICOM objects, such as dose, volume and contour
information to help the user to evaluate treatment plans.
The function of the Knowledge Base (KB) Module is to first extract quantified data in the
DICOM object and clinical data to populate the database and then make these data
available to data mining applications to provide decision support for clinicians. The data
in both the DICOM header and the body are extracted accordingly. The KB also provides
36

methods to extract data from clinical forms. The knowledge extracted by the KB will be
explained later in detail in Data Collection, Chapter 4.
The KB Module and Data Mining Tools are developed under the guidance of the experts,
who are the oncologists and physicists. The design shows how quantified knowledge can
be mined to promote better proton treatment delivery as an example of decision support
tools in Chapter 6.
3.7 Visualization and Display Tools
The ePR Visualization and Display module is the most important part to facilitate the
user to interface with tools to improve proton treatment plans and enhance patient
outcome. Figure 3-5 depicts the general concept of an ePR system, where all patient data
is organized in one place. The patient demographics are shown on the top. The top left
region is the ePR menu allowing users to control and manage the system; such as, data
communication tools, system status, data summary, etc. Each tab on the GUI presents a
separate view for different types of data and purpose. The design currently has seven tabs
(see red numerals in Figure 3-5 Screen shot of a Patient Folder): (1) Summary tab shows
the overall patient record including treatment’s starting and ending date, prescription
dose, dose-per-fraction, delivered dose to tumor and critical structures, and toxicity
grade; (2) Initial Data tab allows users to input and view data from the Initial Data Form;
(3) TP Overview tab shows the summary of treatment plan; (4) TP Evaluation tab
contains tools, eg. Intelligent Treatment Plan Navigator discussed in Chapter 5, allowing
user to evaluate treatment plan; (5) TP Comparison tab allows user to compare two
treatment plans of current patient, usually a quality assured plan and an approved plan;
37

(6) Follow-up Data tab gives the user the feature to input and display data from multiple
follow-up data forms; and (7) TP Knowledge Base Search tab allows user access to the
data mining tool, discussed in Chapter 6.

Figure 3-5 Screen shot of a Patient Folder displaying patient data in 7 tabs.
To satisfy the need of the clinicians, the Graphical User Interface (GUI) is implemented
under guidance of oncologists and physicists. Since the ePR System is the repository for
many types of data from one patient, the GUI is designed to tailor both the needs of
viewing all data under one patient folder and displaying each type of data separately and
continuously. The data displays include The Initial Data Form Display, Images and
Treatment Plan Display; and the Follow-up Data Display. The Intelligent TP Navigator
GUI, and KB & Data Mining Search Engine GUI are examples of Decision Support
Tools and how Decision Support Tools are designed to enhance user-data interaction for
knowledge discovery. A clinical example will be discussed in Chapter 9.
Different patient data are
displayed in 7 tabs
Patient demographics
ePR menu to
access tools
(1) (2) (3) (4) (5) (6) (7)
38

Chapter 4 Data Collection
4.1 Introduction
Similar to conventional radiation therapy, patient data in proton therapy are scattered in
many stand-alone imaging and informatics systems. Each system has its own “storage
area” and has only limited interface or integration with other systems. The PT ePR
system proposed in my research is a centralized system that integrates all the disparate
data and provides a “one-stop shop” for proton treatment data, treatment outcome and
key related data of prostate cancer data [11, 12, 23]. Therefore, data collection not only
verifies the concept of an ePR system but also helps to validate the data mining capability
of the ePR to minimize radiation toxicity in prostate cancer patient, as indicated in
Section 1.5.
A USC IRB (Institutional Review Board) study number HS-08-00652, was submitted to
collect prostate cancer patient data, including CT images, treatment plans and clinical
data. This IRB study was approved on March 2, 2009, which allows 300 data sets to be
collected from LLUMC. Only 41 patient cases were collected in this scope of research.
These data was anonymized before integrating into the ePR system for testing and
evaluating.
4.2 Data Types
Referring back to the yellow boxes in Figure 2-1, the following steps in proton therapy
workflow generate data and are included in the data collection:
• Step 2 – consultation section between oncologist and prostate cancer patient:
initial data form on paper.
39

• Step 4 – Acquisition of CT images in the immobilization device: DICOM Image.
• Step 7-8 – Treatment planning process performed by the physicist/dosimetrist and
oncologist on TPS: DICOM RT Structure Set, DICOM RT-ION Plan, DICOM
RT Dose, and DICOM RT Image.
• Step 11 – Verification using portal image: DICOM RT Image or digitized manual
portal film.
• Step 12 – Delivery of proton treatment on a proton beam machine: treatment
records in paper patient chart.
• Step 13 – Follow-up: follow-up data form in paper.
Table 4-1 summarizes data collected from one patient.
Table 4-1 Proton therapy data collected from one patient
Steps Name Media Type Digital
Format
Location
2 Patient Initial Data
Form
Text Pre-treatment
Clinical Data
No Research
Spreadsheet
4 CT DICOM Image Yes PT Data Server
7, 8 RT Structure Set DICOM Contours Yes TPS WS
7, 8 RT-ION Plan DICOM Plan Yes TPS WS
7, 8 RT Dose DICOM Dose Image Yes TPS WS
7, 8, 11 RT Image DICOM RT Image Yes TPS WS
12 RT-ION Treatment
Record*
DICOM Treatment
Record
No Patient Chart
13 Follow-up Data
Form
Text Outcome Data No Research
Spreadsheet
(* not currently collected since this data type is not significant for current status of the system)

40

4.3 DICOM Files
DICOM files are data in standardized digital format and are collected from two different
information systems. DICOM RT objects are obtained from Odyssey version 4.5,
Permedics Inc. TPS Workstation using the DICOM export function to DICOM files.
Because the DICOM CT Images were previously pushed to TPS for treatment planning
purpose, CT images were also collected at the TPS.
These DICOM objects are imported to the ePR system using a Modality Simulator (MS),
a screenshot of the MS is shown in Figure 4-1, and received by the DICOM Module in
Data Gateway component, see Figure 3-2. The MS is a software application which acts as
a device that sends DICOM objects from one DICOM node to another. A DICOM node
consists of three attributes: an IP address, a communication port and an application entity
(AE) title. A computer can have multiple DICOM nodes by using different port numbers
and AE titles. In Figure 4-1, IP, PORT and RemoteHost are three attributes of the ePR
Server DICOM node where the DICOM data will be received and archived. The
LocalHost’s AE tile is required from the computer used to send data to ePR Server. This
localhost’s AE title needs to be configured at ePR Server so that the ePR can recognize
where the data is sent from to maintain the data integrity and security.
41

Figure 4-1 Modality Simulator
4.4 Non-DICOM Related Clinical Data
The clinicians at LLUMC usually store and provide patient data on paper chart and Excel
spreadsheets. These clinical data are manually inputted into the ePR system through a
data form input webpage. This web-based application can be utilized in the future to
replace paper patient data forms once the ePR system is deployed for clinical usage. The
data entry web-pages are part of the Non-DICOM Module within the Data Gateway
component. (See Figure 3-2)
The collected DICOM-RT Dose at LLUMC did not contain the data points to reconstruct
the Dose Volume Histogram, a graph summarizing 3D dose distribution; therefore, these
data were collected from the PT Department Archive and imported to ePR system using
the non-DICOM Module in the Data Gateway.

42

4.4.1 Initial Patient Data Form
The initial patient data form allows the ePR to collect patient pre-treatment data such as
age, performance status (KPS), prior cancer, other diseases, prior prostate treatment,
initial diagnosis with biopsies results, pathology condition (Gleason score), clinical T
Stage (AJCC 6
th
Edition), and pre-diagnostic and pretreatment PSA level. These data are
critical for physicians to evaluate treatment outcomes. Figure 4-2 shows a scanned paper
initial data form with above mentioned information outlined in red boxes. An electronic
form was developed so that the user can input the information directly into the ePR
system and can be used in place of paper forms once the ePR system is deployed in the
clinical environment.
43

Figure 4-2 Prostate Initial Data Form of patient treated hypofractionation protocol
treatment

44

(Figure 4-2 Continued)

45

The data was collected on an excel spreadsheet from the research department at LLUMC
before imported into the ePR database for system evaluation.
4.4.2 Follow-Up Data Form
The follow-up data form provides data for the ePR system regarding the treatment
clinical outcome. These data includes Prostate Specific Antigent (PSA) levels after the
treatment and five adverse event types identified in CTCAE to evaluate proton therapy
radiation toxicity. The form collects patient’s symptoms and frequency/severity of each
symptom in five adverse events, which includes gastrointestinal, urinary, constitutional,
skin and sexual function. Side effects manifested within 90 days of initiation of proton
therapy were considered "acute", and "late" those manifest thereafter. For acute toxicity,
the follow-up data is collected weekly after the first day of treatment up to four weeks.
Figure 4-3 shows the existing excel sheet to collect acute toxicity for a hypofractionation
protocol as a clinical example. For late toxicity, the form is collected every three monthly
in first two years, six month for the next three years and annually thereafter. An assigned
nurse calls the patient to get answers for the questionnaire in Appendix B to complete the
follow-up forms of late toxicity.
Four follow-up data forms of four weeks after the treatment of each patient were
collected. Similar to the Initial Data Form, an electronic form was developed to have
users input the clinical data on the Follow-up Data Form directly to the ePR system
through web browsers. The data on these follow-up data was imported to the ePR system
in the same manner as the Initial Data Form and used as input for developing the data
mining module discussed in Chapter 6.
46

Figure 4-3 Follow-up Data Form: Toxicity Data Sheet
4.5 Patient Datasets
A data set from one prostate cancer patient, shown in Table 4-1, includes the following
items:
• CT images: one CT study including many CT images; the number of images in
one CT study varies between several dozen to hundreds of images
• RT-Dose: one DICOM-RT Dose for each CT images. Due to the technical
limitation of current TPS application used at LLUMC, this TPS produces one 2D
RT Dose object for each CT slide, instead of one multi-framed RT Dose object
for the whole CT study.
• RT Structure Set: one object
• RT ION Plan: one object
47

• RT Image: several objects of DRR images
• RT ION Treatment Record: one bitmap image file scanned from patient chart
• Initial Data Form and Follow-up Data Form: one Excel file; this file contains
multiple spreadsheets in which the clinician re-entered the patient data from paper
data forms.
Table 4-2 Number of objects from one patient
Data Types Number of objects
CT Images* Several dozens to hundreds
RT Dose* Same as number of CT images
RT Structure Set* 1
RT ION Plan* 1
RT Image* 0 – 5
RT ION Treatment Record** 1
Initial Data and Follow-up Data 1***
* DICOM file;
** Bitmap image file;
*** All patient data is in 1 Excel file.

In my research, I collected 41 patient datasets, each dataset contains a similar number of
objects as shown in Table 4-2.
4.6 Data Organization
All DICOM files are stored in the file system using a complete folder structure, which
specifies patient ID, study UID, and series UID as names of subfolder. This is based on
the hierarchical DICOM data model of the real world [24]. Figure 4-4 shows the
directory hierarchy of DICOM storage on the ePR server.
48

Figure 4-4 DICOM Data Directory Hierarchy of File System of the ePR Server
After storing the DICOM objects, the ePR system extracts the DICOM header
information in these objects to populate the database. The data from the Initial Data Form
and the Follow-up Data Form are stored directly in the database without extracting any
metadata.
4.7 Summary of Data Collection
There are a total of 41 cases collected. Two of the 41 cases were collected to evaluate the
system, which will be described further.
4.7.1 Demonstration Patients Descriptions
Since there are two organs at risks in prostate treatment, two sample cases of patients
who had radiation toxicity grader greater than 3 were collected for evaluation purposes:
one patient had bladder related complications and another had rectum related
complications. (See Appendix A for more details) These patients were created by an
expert in hypofractionation protocol at LLUMC.
49

Demonstration Patient 1: Late Rectal Toxicity
The patient is a 60 year old man who is in early stage of prostate cancer. The patient was
diagnosed with clinical stage T1c Nx Mx prostate cancer. The pretreatment PSA level
was 5.2 ng/ml. He was treated with protons for a final tumor dose of 60 Gy. The patient
agreed to be treated on a hypofractionation Phase I/II study protocol at Loma Linda
University Medical Center. The treatment consists of 20 fractions and was given over 33
calendar days. This patient developed no acute toxicity but late toxicity of grade 3 in
gastrointestinal category due to proctitis beginning 34 months after radiation treatment.
Demonstration Patient 2: Late Bladder Toxicity
The patient is a 75 year old man who was in early stage of prostate cancer. The patient
was diagnosed with clinical stage T2a Nx Mx prostate cancer. Over the last five years, he
presented with slowly increasing prostate-specific antigen (PSA) levels from 4.1 ng/ml to
the highest of 7.7 ng/ml. The patient agreed to be treated on a hypofractionation Phase
I/II study protocol at Loma Linda University Medical Center. He was treated with
protons for a final tumor dose of 60 Gy. The treatment consists of 20 fractions and was
given over 33 calendar days. This patient developed acute toxicity of grade 1 and late
toxicity of grade 3 in genitourinary category due to cystitis beginning 34 months after
radiation treatment.
The patient in example 1 is used to explain the decision support capability of the system
in Chapter 9.

50

4.7.2 Hypofractionation Patients
All 39 prostate patients collected in this research were treated with hypofractionation
protocol, total dose of 60Gy and dose per fraction of 3Gy, and had no prior prostate
cancer.
Table 4-3 summarizes the distribution of the main pretreatment clinical parameters
collected on the initial data form. Table 4-4 lists the distribution of dosimetry and
technical parameters in the treatment plans. Table 4-5 summarizes follow-up
characteristics in the follow-up data forms.
Table 4-3 Summary of main pretreatment clinical data (n=39)
Characteristic Mean Standard
Deviation
Range Missing
Data
Age (years-old) 62.59 8.85 33 - 75 0
Pretreatment PSA (ng/ml) 4.83 2.24 0.27 - 8.67 3
Gleason score 6 0 6 - 6 1
Percent biopsy scores (%) 23.12 21.35 5.26 - 100 7
T Stage T1c T1a - T2c 1

51

Table 4-4 Summary of dosimetrist and technical parameters in treatment plan (n=39)
Characteristic Mean Standard
Deviation
Range Missing
Data
Prostate volume (cc) 69.65 25.03 33.93 - 145.78 0
Mean rectal Dose (%) 12.53 4.15 6.1 - 22.7 0
Rectal V60 *(%) 2.25 2.57 0 - 12.9 0
Rectal V55 *(%) 8.85 4.57 1.5 - 21.5 0
Rectal V50 *(%) 11.68 5.30 3.4 - 25.6 0
Rectal V45 *(%) 13.94 5.82 4.8 - 28.7 0
Mean bladder dose (%) 7.79 5.36 2.2 - 26.3 0
Bladder V60 *(%) 0.60 1.15 0 - 6.2 0
Bladder V55 *(%) 4.22 3.87 0.2 - 18.7 0
Bladder V50 *(%) 5.91 5.00 1.1 - 23.6 0
Bladder V45 *(%) 6.98 5.69 1.6 - 26.1 0
V
45
-V
60
= volume receiving ≥ 45-60
* These data was obtained from the ePR

Table 4-5 Summary of follow-up clinical data (n=39)
Characteristic Mean Standard
Deviation
Range Missing
Data
Acute Toxicity * 1.29 0.73 0 - 2 25
Late Toxicity* 0.63 0.67 0 - 2 1
PSA 3 months after
treatment (ng/ml)
2.1 1.27 0.02 - 5.4 6
* These data was calculated using the ePR
The initial data form, summarized in Table 4.3, for the sample of n=39 subjects indicated
a 0.096 missing value rate. The follow-up form, summarized in Table 4.5, for the sample
of n=39 subjects indicated a severe missing value rate of 0.602. The SAS multiple
imputation procedure showed no discernable missing value pattern in both data forms.
This highlights the problem of missing data due to human handling that will be avoided
52

with the implementation of the ePR system in the clinical setting for data collection once
the data is received in the system.

53

Chapter 5 Intelligent Treatment Plan Navigator
5.1 The Need for Treatment Planning Navigation Tool
As described in Section 2.4, the hypofractionation protocol treatment is currently being
developed at LLUMC to improve treatment schedule for patient convenience and
reducing cost of PT treatment. In the treatment planning steps of this protocol, both
physicists and the oncologists need to evaluate the following listed below in addition to
the standard treatment plan workflow, shown in Error! Reference source not found.:
• DVH curves of tumors and critical structures to make sure they fall under
hypofractionation protocol dose constraints, i.e. no more than 15% volume of
rectum receives the dose of 60Gy.
• Dose distribution to identify the overdose area and volume for the overdosed
regions in the critical structures.
In order to perform this task, they rely on the cumulative dose volume histogram (DVH),
and the dose distribution, isodose curves and structures' contours overlaid on CT images.
However, existing treatment plan tools are limited in their ability to provide intelligent
navigation of these data for the necessary evaluation of complex treatment. Therefore, the
intelligent treatment planning navigator (ITPN) has been developed as a decision support
tool to enhance the treatment evaluation process for hypofractionation protocol treatment.

54

5.2 Features of the Intelligent Treatment Plan Navigator (ITPN)
Because the ITPN is developed based on the knowledge of the dose volume histogram
(DVH) I will first introduce the concept of DVH, which is used to evaluate treatment plan
in PT and RT. Next I will describe the features and steps of using the ITPN for evaluating
treatment plan. Finally I will present the mock-up GUI of the ITPN.
5.2.1 Dose Volume Histogram (DVH) and features development
The dose volume histogram (DVH) is a concept used in treatment planning with purpose
of summarizing 3D dose distribution in a graphical 2D format. The "volume" referred to
in DVH analysis can be a target of radiation treatment, a healthy organ nearby a target, or
an arbitrary structure. As shown in Figure 5-1, the target is prostate; healthy organs
nearby the prostate include rectum, bladder. The cumulative DVH is plotted with doses
along the x-axis; the value of percentage volume on y-axis represents the volume of
structure receiving greater than or equal to that dose. The cumulative DVH takes on the
appearance of a smooth line graph. The lines always slope and start from top-left to
bottom-right. For a structure receiving a very homogenous dose—100% of the volume
receiving exactly 60 Gy for example—the cumulative DVH will appear as a horizontal
line at the top of the graph, at 100% volume as plotted vertically, with a vertical drop at
60 Gy on the x-axis. In Figure 5-1, from the DVH curve, the prostate receives nearly a
total dose of 60 Gy.
In general, the DVH curves and dose distributions are important indicators from the
treatment plan that show the impact of dose to the prostate and particular critical
55

structures. A drawback of the DVH methodology is that it offers no spatial information;
i.e., a DVH does not show where within a structure a dose is received.

Figure 5-1 Dose volume histogram (DVH) curves in a prostate treatment plan (red-
prostate; yellow-bladder; green-rectum.)
5.2.2 Functions of the ITPN
In this research, following terminology described in aforementioned section, the
development of an intelligent treatment plan navigator has the following features:
• Navigate along a single DVH curve to verify dose of the tumor and critical
structures along with correlated dose distribution volumes in 2D and 3D
• Navigate through multiple DVH curves with the same corresponding volume or
dose along with correlated dose distribution volumes in 2D and 3D
• Visualization and quantification of overlapped or non-overlapped region between
the structure volume and isodose using 3D surface rendering.
• Automatic visualization and quantification of overdose regions in critical
structures.
56

The ITPN in this context means being able to visualize and interpret various parameters,
such as dose distribution, structures, and location of delivered dose, etc., to use DVH
curve as a guide to traverse through 2D image or 3D volume of the anatomy structures in
a treatment plan.
The features of the ITPN help to provide the user with the ability to navigate and evaluate
the complex treatment plans such as hypofractionation protocols for prostate cancer but
can be easily extended to all types of treatment plans and cancer types.
Figure 5-2 shows the workflow steps of the ITPN that the user interacts with the DVH
data, CT images, structures contours, dose distribution to evaluate the treatment plan.
Note that the caption numbers in Figure 5-2 are corresponding to the step numbers in the
following paragraphs.
1. Utilizing the ITPN, the user can first review the DVH curves of the tumor and
critical structures (rectum and bladder). The user can navigate along single DVH
curve by dragging the black cross-hair (see 1a in Figure 5-2) both along the x-axis
for received dose and y-axis (percentage of volume that receives a dose equal or
greater than an indicated dose on x-axis). The user can also select multiple DVH
curves to navigate in the same manner (see 1b in Figure 5-2).
2. In addition to the DVH curve, the corresponding 2D slices with isodose curves
overlaid will display as the user selects a point on the DVH curves.

57

3. Additionally, the quantified volume of dose that is overlapped into the rectum
critical structure volume is displayed along with the results as shown in the DVH
curve.
4. The corresponding 3D dose distribution is displayed together with the prostate
and critical structures. In this case, the 3D dose distribution (see the white arrow)
shows some dose outside of the prostate volume indicating possible dose to
neighboring critical structures.
58

Figure 5-2 Features of the ITPN. (1a) Navigation along single DVH curve navigation
along x-axis and y-axis using arrow buttons; (1b) Multiple DVH curves
navigation; (2) Display 2D CT Images with contours and isodose curve overlaid;
(3) Display quantitative measurement of overdose volume in critical structures;
and (4) Display 3D surface of chosen critical structure, dose distribution;.
(Orange-bladder; Green-rectum; Pink-prostate; Brown-isodose surface).

59

5.2.3 ITPN Mock-up GUI
Figure 5-3 shows the prototype graphical user interface (GUI) consisted of four regions:
(A) DVH display, (B) 2D image display with contours and isodose curves overlaid, (C)
quantitative measurements display, and (D) 3D rendering image display.

Figure 5-3 Prototype GUI of ITPN with four regions: (A) DVH display region; (B) 2D
image display with contours and isodose curves overlaid; (C) Quantitative
measurements display; and (D) 3D rendering image display;
5.3 Summary
The ITPN allows a user to intelligently navigate between DVH curves, containing 3D
dose distribution, and corresponding 2D dose distribution on the CT images along with
quantified area and volume data of overdosed regions for better evaluation of complex
60

treatment plans such as the hypofractionation protocol clinical example. The knowledge
developed for ITPN is also utilized as part of the data mining tool discussed in Chapter 6.
It is an option that allows the user to interact and re-evaluate the plan resulting from the
data mining search results.

61

Chapter 6 Data Mining Using Knowledge Driven Search Engine
6.1 Significance of Data Mining using Knowledge Driven Seach Engine
Data mining is the process of extracting hidden patterns from data. As more data is
gathered in current medical practice, data mining becomes an important tool to transform
this data into information or knowledge to improve treatment for future patients. In
proton therapy, the knowledge is extracted from two sources: the previous existing
treatment plans and the outcomes results. By searching, mining and correlating the
knowledge with standard criteria and patient information, one can utilize this knowledge
and mining results to improve treatment for new patients.
In this research, the data mining tool together with a search engine allows clinicians to
find the correlation between treatment plans parameters, such as dose distributions as
displayed in DVH curves, with clinical outcomes data, such as radiation toxicity of
existing hypofractionation protocol patient data to improve treatment plan for new
patient. The following sections will first describe the knowledge development and then
the subsequent knowledge-driven search engine with knowledge integration of treatment
plans and outcomes data.
6.2 Knowledge of DICOM RT and DICOM RT-ION Data
The knowledge of the DICOM RT and DICOM RT-ION objects includes the DVH
curves, correlation between 3D dose and 2D dose, and quantitative results of the received
dose on critical structures. This knowledge is extracted using the Data Quantitative
Module, discussed in Section 3.5.2, and stored in knowledge database.
62

6.2.1 DVH Data
The DVH curves and dose distributions are important indicators from the treatment plan
that show the impact of dose to the prostate and particular critical structures. Since the
clinical application is grading radiation toxicity, the dose distribution of each treatment
plan for each patient’s rectum, bladder, and the tumor need to be extracted from the RT
Objects. Figure 6-1 shows sample knowledge extracted from the DVH curve and dose
objects.

Figure 6-1 Dose and Volume Knowledge extracted from DICOM RT objects
The corresponding dose and percent volume will be identified to compare with the
current allowed limits for prostate hypofractionation prototol treatment plan at LLUMC.
The limits are listed in Table 6-1. As an example, the first row for rectum section in this
table reads that no more than 15% of rectum volume receives the dose of 60 Gy.

63

Table 6-1 Dose limit for critical structures
OARs Volumes Dose (Gy)
Rectum < 15% 60
< 25% 55
< 35% 50
< 50% 45
Bladder < 15% 60
< 25% 55
< 35% 51
< 50% 48
6.2.2 3D Dose and 2D Dose Correlation
The 3D dose information of a treatment plan is presented as a cumulative DVH. Each
point on a cumulative DVH curve of each structure shows the relationship of total dose
(x-axis value) received by a percentage of volume. For example, the point (x=42.50,
y=14.49), displayed on Figure 6-2, means that the dose 42.50 Gy is delivered to
maximum 14.49% volume of a structure.

Figure 6-2 Example of dose volume histogram (DVH) curves in a prostate treatment plan
(red-prostate; yellow-bladder; green-rectum.) The green O shows that the dose of
42.50 Gy is delivered to maximum 14.49% volume of the rectum.
64

The 2D dose information is stored in DICOM RT Dose objects. Each CT image, used to
plan the treatment, has an equivalent dose image stored in RT-Dose file(s), referenced by
the frame of reference and the z-index of patient position. The value of each pixel in a
dose image is the dose delivered to that anatomical location. Traditionally, a value ‘y’ on
y-axis on DVH curve is obtained by counting all the pixels, having the same value ‘x’ on
x-axis on each RT dose images, and translating the total number pixels to volume units.
The ITPN in this research allows the user to find a point on the DVH curve and extract
automatically all CT images that contain higher dose regions of an organ. The algorithm
for this knowledge discovery is as following:
1. Choose an ‘x’ value on a DVH curve, i.e. x = 42.5 Gy on Figure 6-1.
2. Search all RT-Dose Image and select only RT images containing any pixel value
higher than ‘x’
3. Use the frame of reference and patient position to identify CT Images
corresponding to RT Images of the previous step.
The value of this knowledge is to help clinician identify the higher dose locations and
change the treatment plan to minimize this dose.
6.2.3 Quantification of Received Dose on Organs at Risk
In practice of radiation therapy, the RT Dose image is normally visualized as isodose
curves overlaid on CT images. Quantification of intersecting areas or volumes between a
structure contour and isodose curves can provide knowledge for the user to evaluate the
effect of dose in the treatment plan. This area can be calculated using the Image
65

Processing Module, described in Section 3.5.3, once the CT image is identified for a
particular structure and level of isodose curve is requested. This measurement gives the
quantitative results of the 3D and 2D correlation in the previous section.
6.3 Knowledge of Outcomes Data
As mentioned in Section 2.5, the side effects of radiotherapy are quantified to radiation
toxicity grades in five categories (see Figure 6-3) according to the adverse events defined
in the Common Terminology Criteria for Adverse Events v3.0 (CTCAE). An adverse
event (AE) is any unfavorable and unintended sign (including an abnormal laboratory
finding), symptom, or disease temporally associated with the use of a medical treatment
or procedure that may or may not be considered related to the medical treatment or
procedure. Grade refers to the severity of the AE: Grade 1 Mild AE; Grade 2 Moderate
AE; Grade 3 Severe AE; Grade 4 Life-threatening or disabling AE; and Grade 5 Death
related to AE.
In this research, the AE of proton beam on both rectum and bladder are considered to
calculate the total toxicity grades. Figure 6-3 shows an example of toxicity grade
knowledge obtained in five categories, gastrointestinal, urinary, constitutional, skin, and
sexual function, of one prostate cancer patient treated with PT hypofractionation
protocol.
66

Figure 6-3 Example of Toxicity Knowledge. The weekly toxicity grade is the worst
(highest) grade in five categories. The total worst toxicity grade is the worst
grade of four weekly grades. In this example, the worst toxicity grade is 2.
Currently, at LLUMC, outcomes data stored in follow-up data forms are manually
recorded and entered on an Excel spreadsheet. Furthermore, the clinical staff who are
responsible to collect these treatment outcomes and assess the results needs to read
through the CTCAE v.3 standard in either paper form or digital PDF files to identify
outcomes such as toxicity grades based on the follow-up data. This process is not only
tedious and time consuming but also prone to mistakes due to human error. Furthermore,
user cannot query or relate patient data and toxicity grade with treatment plans to do
further analysis. Therefore, a toxicity grade computation module is developed to extract
radiation toxicity knowledge from the follow-up forms.
6.3.1 Toxicity Grade Computation
The implementation of toxicity grade computation module utilizes the CTCAE v.3 as a
guideline to create the ctcaemap table, as shown in the top Figure 6-4, for querying the
toxicity grade according to the AEType, AE and severity of these AEs. Figure 6-4 also
shows examples of information stored in this table. The AE, AEType and severity are
supplied from the electronic follow-up data form. For each category, the toxicity grade is
67

computed based on simple database table and query approach. The worst toxicity grade 2
in urinary, shown in Figure 6-3, is then identified as the highest grade of the five
categories in all four weeks.

Figure 6-4 Knowledge for Toxicity Grade. Top: ctcaemap table structure contains the
information published in CTCAE v.3 to evaluate radiation toxicity grade.
Bottom: examples of knowledge data stored in ctcaemap table in ePR database.
6.4 The Integration of Knowledge and the Development of the Search
Engine
Figure 6-5 shows the database schema example of the knowledge base combining the
toxicity information and treatment plan data. All the Class Objects in Figure 6-5 are
linked with Class Objects Patient. This database schema is a combination of knowledge
extracted from outcomes data and quantitative data extracted from DICOM data, of
which an example is discussed in Section 3.5.2.
68

Figure 6-5 Knowledge base database schema example combining toxicity and treatment
plan information
The search engine is developed based on database query which relates all or part of
knowledge stored in these tables in response to user query command. For example, the
search engine will extract the worst toxicity grade and correlate that with attributes in
Class Object DVH, Isodose Curve and Critical Structures, resulting in a list of patients
who matches the search criteria of those parameters. The result from this search will
allow knowledge discovery for outcomes analysis and treatment improvement.
6.5 Summary
Knowledge is defined and quantified based on the expert’s ability, in this case, either the
oncologist or physicist, to utilize data and other criteria in evaluating, grouping, and
defining certain clinical characteristics that are extracted from the standardized DICOM-
RT objects and outcome data stored within the ePR. In this chapter, I have discussed the
69

knowledge development used for data mining and the knowledge integration used in the
search engine. The search engine provides clinicians the ability to mine knowledge
extracted in DICOM data of treatment plans and outcomes. A clinical example
illustrating the search engine functionality and knowledge discovery capability is
discussed in Chapter 9.

70

Chapter 7 System Implementation
7.1 System Implementation
7.1.1 Web Technology
The ePR system has been developed with current open source Web technology, image
processing techniques and DCMTK-DICOM library. With the advantages of Web server-
client architecture, including Apache web server [25], MySQL Database [26], PHP
(Hypertext Preprocessor) scripting language [27] for the server side and a combination of
JavaScript, and activeX libraries for the client side, the patient data can be visualized
through web browsers and distributed across multiple locations by the local area network
and Internet.
The WampServer version 2.0 [28], a package of Apache web server 2.2, PHP 5.2 and
MySQL 5.1, is utilized for development and distribution on a server running Windows
operation system (OS). The development depends heavily on PHP scripts on server side,
jQuery JavaScript library [29], and Cascade Style Sheet (CSS) for graphical user
interface on web browsers.
7.1.2 Database
The database serving as the backbone of ePR system has been developed using the
Community version 5.1 of MySQL. Figure 7-1 shows the latest version of the database of
the ePR Server with a total of 68 tables: 8 tables for user management, 30 tables for
DICOM data, and 18 tables for DICOM knowledge extraction, and 12 tables for clinical
data and knowledge discovery support.
71

Figure 7-1 Screenshot of database tables of the ePR System for prostate cancer patients;
(A) Tables of DICOM data, (B) Tables of user management, (C) Tables of
DICOM extraction knowledge, and (D) Tables of clinical data and knowledge
discovery.

72

7.1.3 Graphical user interface (GUI)
The GUI of the ePR has been developed to support user interactions with the all the data
collected in the system for knowledge discovery. The three major GUI implementations
include:
• The initial data and follow-up electronic forms which allow users to input, view
and modify clinical data. (Figures 7-5, 8-3, 8-4 and 9-3)
• Intelligent Treatment Plan Navigator which allows users to view and navigate the
DVH curves, corresponding CT images, isodose curves and organs’ contours, all
obtained from DICOM objects. (Figures 7-6, 9-10 and 9-11)
• Search Engine which allows users to mine the data in the ePR system for
knowledge discovery. (Figures 9-6 and 9-7)
These GUIs will be explained in conjunction with step-by-step workflow in the next two
chapters.
7.2 Laboratory Evaluation of the ePR System
The proton therapy ePR system for prostate cancer patient was developed at the Image
Processing and Informatics Lab (IPILab), University of Southern California. The
evaluations followed the specifications in Chapter 1. The compliance and security
requirements are satisfied since the web-based ePR system was developed based on
HTTPS and DICOM standard, discussed in Chapter 3. Other specification requirements
are discussed in the below subsections.

73

7.2.1 Hardware & Software
Figure 7-2 shows the hardware testing configuration at IPILab. All the software
components mentioned in Section 7.1.1 were installed on the ePR System, running
Windows Server 2008, and checked to run properly. The Treatment Planning System
(TPS) Simulator, running Windows XP, with the Odyssey treatment planning software
non-clinical version 4.2 installed, was used to test the data flow of DICOM objects. The
Client Workstation (WS), running Windows 7, was used to verify data distribution
through the web interface and functionalities of the electronic forms, ITPN, and Search
Engine.
Client WS ePR Server TPS Simulator

Figure 7-2 Testing configuration at IPILab
7.2.2 Communication & Data
7.2.2.1 DICOM Communication and DICOM Objects of Treatment Plans
The DICOM-RT and DICOM RT-ION objects are pushed from the TPS to the ePR
system. CT Images was imported to the ePR using the Modality Simulator, shown in
Figure 4-1. Figure 7-3 shows 41 patient folders named with anonymized Patient ID and
subsequent folders. These DICOM data are stored in “Patient_data” of the ePR server.
More than 8500 DICOM objects from 41 datasets were successfully imported to the ePR
Server.
74

Figure 7-3 Verification of DICOM Data
7.2.2.2 Web Communication & Non-DICOM Data
There are two types of non-DICOM data: clinical data from the initial and follow-up data
forms, and the DVH data in text file. The non-DICOM data in initial data and follow up
data were entered manually through the web GUIs. The DVH data were imported to the
ePR using a script running on the Data Gateway. The non-DICOM data was verified at
the database. Figure 7-4 shows the data was successfully imported in the database.
75

Figure 7-4 Non-DICOM data was verified at the ePR server database
7.2.2.3 Data Distribution
This evaluation was performed at the client WS which had different web browsers such
as Chrome, Firefox, and Internet Explorer installed. The data were verified that they were
distributed across the network and displayed correctly on each web browser. As an
example, Figure 7-5 shows the screenshot of non-DICOM initial data displayed on a
Firefox browser. This data was verified by comparing to the original data collected from
LLUMC. Figure 7-6 displays the screenshots of the ITPN with DICOM data displayed.
76

Figure 7-5 Initial data displayed on Firefox browser
77

Figure 7-6 Screenshot of the ITPN with DICOM data displayed on Firefox browser
7.2.3 Features: ITPN & Search Engine
To ensure the ITPN functions correctly, the knowledge extracted from DICOM objects,
such as the reconstructed organ contours, isodose curves and DVH curves, displayed on
the right of Figure 7-6, were compared against the data displayed on the TPS. This test
was performed on all 41 patient datasets.
To verify the Search Engine performance, the results from the Search Engine was
compared against the search results performed manually on Excel spreadsheet.

78

7.3 Summary
The ePR system was validated to support major current web browsers, such as Firefox,
Internet Explorer, and Chrome. The entire 41 patient datasets were imported to the ePR
system. Each patient data was verified at the client side to ensure the data was displayed
correctly: 40 out of 41 patient datasets was successfully viewed on the web browser at the
client WS. The one patient dataset, which failed to display all data, had a different Patient
ID in the CT images and DICOM RT objects exported from the TPS; therefore, the ePR
treated this patient data as two incomplete datasets to display. This problem will be
addressed in future work to reconcile patient information.
In general, the system evaluation was performed to allow knowledge discovery to
improve the treatment plan of the new patient. An example with step-by-step workflow is
discussed in Chapter 9.

79

Chapter 8 Data and System Integration for Improving Patient
Outcome Study
8.1 Introduction
As stated before in previous chapters, the clinical proton therapy workflow is comprised
of four main steps: Consultation, Treatment Planning, Treatment Delivery and Follow-
up. In each workflow phase, different datasets are collected to perform further analysis in
order to improve the treatment methodology and quality of care.
Similar to most treatment protocols, the goal of the new hypofractionation treatment
protocols is to accurately deliver proton beam dose to the tumor in the prostate gland
while maximizing tumor control and minimizing toxicity side effects to adjacent critical
structures, such as the bladder and rectum. The outcome may vary depending on many
parameters, including but not limited to:
• Patient information and demographics,
• Medical history and patient’s pathologic conditions,
• Treatment protocol: total dose, dose per fraction and treatment schedule.
These data reside in the Initial Data Form, DICOM objects and Follow-up Data Forms,
which are explained previously in Figure 2.1 and Chapter 4. In order to evaluate this new
hypofractionation treatment protocol for prostate cancer, researchers need to collect all
relevant clinical data, images and treatment data. Currently, they are challenged with
collecting data in paper formats at different locations and in different clinical systems,
which is tedious and time consuming to collect and analyze. In addition, the datasets are
prone to be missing and have incomplete data fields. Therefore, there is a need for
80

digitizing the data forms and integrating all the necessary information into one integrated
database in an ePR system for performing patient outcomes analysis. For each step of the
existing dataflow, discussed in the next section, it normally takes several days to
complete.
The ePR System in this research is an effective system that provides a centralized archive
for all data of prostate cancer patients treated with proton therapy. The first contribution
of my thesis research is the improvement and efficiency in data collection and integration
to facilitate patient outcomes analysis. This significance will be explained in the
following sections by comparing dataflow in the current clinical environment as well as
once the ePR is deployed.
8.2 Existing Dataflow for Collecting Hypofractionation Patient Data
Although similar to the general clinical workflow for proton therapy, the workflow for
evaluating a new research treatment protocol has a few additional steps. In this thesis
research, we will use hypofractionation treatment protocol workflow as an example of
workflow for evaluation of a new treatment protocol. Figure 8-1 shows the existing
dataflow of collecting data for the hypofractionation treatment protocol at LLUMC. The
far left of the figure 8.1, in blue, shows the summary of general proton therapy workflow.
The data generated in each of the workflow steps are in red filled boxes of Input Data.
The dataflow is as follows:
1. In the consultation step, the Initial Data Form is filled by hand by physicians and
given to the clerical nurse who delivers the Initial Data Form in paper format to
81

the research department. These forms are normally stored in department files. The
delivery usually takes a few days.
2. In the treatment planning step, the DICOM objects, including CT and DICOM-
RT, residing in the TPS Servers for the physicist and oncologist who plan the
treatment. The CT and treatment plans in proprietary format are archived
periodically to the PT Department Archive. No DICOM-RT object is stored in
this Archive.
3. In the Follow-up step, an assigned nurse will call patients to fill out the Follow-up
Data Forms, and deliver these forms to the research department with the same
time frame length as the Initial Data Form of a few days.
4. Researchers or clinicians access treatment plans and clinical data for outcomes
analysis. The existing treatment plans can be loaded from the TPS Server at the
TPS workstation. If the plans are already archived, a service request needs to be
made to the IT team to load the plans from the department archive, which takes 1-
2 days. The clinical data forms are retrieved manually from the file cabinet.
The output of these four steps is the aggregation of information from treatment plans and
clinical data forms extracted, transformed and standardized and entered into a datasheet.
For example, the radiation toxicity grade is calculated manually from the answers in the
follow-up forms. The process of extracting information from treatment plans and
calculating toxicity grades usually takes a few minutes per treatment plan per form per
patient and could be subject to human error.
82

Figure 8-1 Existing proton therapy dataflow in the research of hypofractionation protocol
for selected patients at Loma Linda University Medical Center. The arrows at the
bottom depict the time required for each step and information extraction.
8.3 Improved Dataflow for Collecting Hypofractionation Patient Data with
ePR System
Figure 8-2 shows the dataflow when the ePR is used to evaluate a new treatment
protocol. As in Figure 8-1, the clinical example for evaluation of the new hypo-
fractionation treatment protocol is used. The steps of the improved dataflow are
displayed in comparison with the existing dataflow (see Figure 8-1):
1. In the Consultation step, the Initial Data Form is electronically filled in and
transferred instantly to the ePR System once the physician submits the form.
Figure 8.3 shows the electronic GUI of the Initial Data Form.
83

2. In the Treatment Planning step, the DICOM objects are generated at the TPS
Server and exported to the ePR Server using the DICOM communication
standard.
3. In the Follow-up phase, the Follow-up Data Forms are also filled electronically by
the nurse. The information is immediately sent to the ePR system. Figure 8.4
shows the electronic GUI of the Initial Data Form.
4. The researchers/clinicians can immediately access the clinical data and treatment
plans from the ePR system at any client workstation on the network using the
web-based browser graphical user interface (GUI).
The ePR automatically extracts, transform and standardizes data to valuable and useful
knowledge, e.g. radiation toxicity grade, for outcomes analysis as soon as the data is
received in the ePR. The human effort in this step is minimized; therefore, this step could
be eliminated from dataflow with the ePR implemented in the workflow.
84

Figure 8-2 New improved hypofractionation dataflow in research with ePR deployed.
The ouput is the aggregated data in the database of the ePR system (item 5) which
can be easily accessed by researchers/ clinicians at the web-client workstation. The
arrows at the bottom depict the time required for each step which is instantly.
8.4 Comparison of Dataflow without and with ePR Implementation
Table 8-1 shows the difference in each data flow steps between the current and once the
ePR is deployed. There are five advantages in the dataflow with ePR deployed: (1) the
electronic initial data form are a direct replacement to the paper forms, so human data
handling is limited by the use of these electronic forms delivered through network to the
ePR server; (2) CT and DICOM RT objects are exported to the ePR server directly from
data source using DICOM Export command; (3) the paper format of follow-up forms are
replaced by the electronic forms which are directly delivered to ePR in the same manner
as electronic initial data forms; (4) data retrieval is performed directly from the ePR and
85

data is delivered automatically to the user through local network; data standardization and
transformation are handled by the ePR system with limited human intervention; and (5)
all data of the patient is stored in one ePR folder in one system.
Table 8-1 Differences between hypofractionation dataflow with and without ePR system
Steps Data Existing Dataflow Advantages of Dataflow
with ePR
1 Initial Data
Form
The form is filled by hand.
Nurse delivered paper form
to research department.
The form is filled online
electronically, shown in
Figure 8.3.
2 CT & DICOM
RT for treatment
plan

These DICOM objects reside
in TPS and saved in PT
Department Archive.
Using DICOM services,
these DICOM objects
residing in TPS are pushed
to ePR Server for storage.
3 Follow-up Form The form is filled by hand.
This paper form is delivered
to research department by
the nurses.
The form is filled online
electronically, shown in
Figure 8.4.
4 Forms and
Treatment Plans
- The paper forms are
manually retrieved from the
file cabinet
- The DICOM objects are
retrieved from TPS or PT
Department Archive.
Both forms and treatment
plans can be retrieved and
loaded at researchers and
clinicians workstations.
5 Final Datasets
for analysis
The data in forms and
treatment plans are manually
extracted, transformed and
standardized to useful data
for outcomes analysis by the
researchers/clinicians.
The data is automatically
converted to final datasets
for outcome analysis by
the ePR system.

86

Figure 8-3 Screenshot of Initial Data Form filled in electronically inside white
rectangles.

Figure 8-4 Screenshot of Follow-up Form filled in electronically inside white rectangles.

87

8.5 Improvement and Efficiency in Data Collection and Integration with
ePR System
This ePR system consolidates all PT patient records in one patient folder which improves
the efficiency in data collection and integration that is the first contribution of my
research. Table 8-1 shows the time improvement and data completeness of dataflow with
ePR deployed in comparison to the current data flow. In steps 1, 3, 4, and item 5, the time
of data collecting and retrieval are reduced from several days to instantly collecting and
immediately available. The total data collection and retrieval time depends only on
network speed and data size, all other delays in human handling and manually retrieval
are eliminated with the implementation of the ePR system. Furthermore, human error
problem will be minimized with the use of ePR for data transformation and
standardization to perform outcomes analysis.
Table 8-2 Time improvement of workflow and data completeness
Steps/Item Tasks Time in Existing Dataflow Time in New
Dataflow with ePR
Step 1 Collect Initial Data
Form and deliver to
research department
2-3 days
(prone to missing forms
and/or incomplete data fields)
Instantly*
Step 2 Archive DICOM data No change
Step 3 Collect Follow-up
Data form and deliver
to research department
2-3 days
(prone to missing forms
and/or incomplete data fields)
Instantly*
Step 4 Retrieve data forms
and DICOM objects
Few hours to few days Instantly*
Step 5 Standardize and
transform data to
create final data sets,
e.g. calculate radiation
toxicity grade
Few minutes to few hours per
patient (This time depends on
how many forms the patient
has. This process is prone to
human error)
Zero
(This step is
eliminated from the
dataflow with ePR
installed.)
(* Network speed and data size dependent.)
88

In general, by comparing two right columns of Table 8-2, the time needed in each step
and item of the dataflow is reduced from days (see third column from the left) to real-
time (see last column). This time reduction in the ePR dataflow demonstrates a solid time
improvement in steps of data collection and integration for outcomes analysis in the ePR
based treatment protocol. Furthermore, an expert user can access all related clinical data
of one patient, including pre-treatment information in Initial Data Forms, treatment plans,
and post-treatment follow-up information with side effects, from a single one-stop source
ePR platform, ultimately providing the clinician with an efficient tool to gain insight and
facilitate knowledge discovery. The knowledge discovery using the ePR system is the
second contribution of my thesis research which will be discussed in the next chapter.

89

Chapter 9 Knowledge Discovery for Decision Support in Treating
Prostate Cancer Patient with Hypofractionation Protocol
Proton Therapy
9.1 Introduction
The second contribution of my thesis research is the utilization of knowledge in collected
and integrated data to provide decision support in treating prostate cancer patient with a
new hypofractionation treatment protocol in proton therapy. This decision support
capability of the ePR system with data mining and navigation tools gives the physicists
and oncologists the ability to optimize the treatment plan of a new patient, predict
outcome of the treatment plan in comparison to treatment plans of previous patients with
the same protocol and make necessary changes to improve the result. In this research,
after the changes are made, the prostate cancer patient will have the option to follow the
existing clinically accepted dose escalation treatment protocol (see Section 2.4.1), or a
refined hypofractionation protocol or can even choose an alternative treatment approach
outside the conventional realm of radiation therapy. This added value of the ePR system
provides tremendous advantages to the new patient because physicists and oncologists
will have a better understanding and control over the treatment planning to deliver the
best possible plan which the current workflow would not be able to do so. For example, if
patients elect for the hypofractionation treatment protocol, decision support tools can help
physicians to determine whether this protocol is suitable for them based on prior
knowledge obtained from the existing patients who had already undergone the
hypofractionation protocol. An example of how the ePR can improve the clinical
workflow with decision support will be explained in greater detail in the following
sections by describing the existing hypofractionation workflow, followed by the
90

workflow once the ePR system is installed, and concluded by a clinical example showing
the step-by-step workflow utilizing the ePR and its decision support tools.
9.2 Existing Hypofractionation Protocol Workflow
Similar to the workflow of general proton therapy, discussed in Section 2.2, the
hypofractionation treatment workflow has four main workflow steps: Consultation,
Treatment Planning, Treatment Delivery and Follow-up, depicted as steps 1-4 in Figure
9-1. However, two additional steps, 1a and 2a, are added for quality assurance (QA)
purposes. These QA steps are necessary since hypofractionation is not yet a clinically
accepted protocol for proton therapy practice on prostate cancer and still in an evaluation
process. In step 1a, the information to determine the eligibility and ineligibility of a
patient to be treated using the hypofractionation protocol are collected using the Initial
Data Form. The patient criteria are summarized in Table 9-1.
These patient criteria ensure that the patient is in low and intermediate risk groups for
prostate cancer [30]; and developed side effects are only due to proton therapy
hypofractionation treatment protocol.
The dose constraints assessment, in step 2a, ensures the patients following the
hypofractionation treatment protocol will have minimized risks of developing adverse
events. These dose constraints are mentioned earlier in Section 6.2.1.

91

Figure 9-1 Existing hypofractionation workflow consisted of 4 main workflow steps.
Steps 1-4 are current general proton therapy clinical workflow. Two additional steps
are added in the hypofractionation research workflow, steps 1a and 2a, to ensure the
patient is in low and immediate group risks.
Table 9-1 Patient criteria for hypofractionation protocol treatment
Criteria for Eligibility Conditions for Ineligibility
- Historically confirmed adenocarcinoma
- History and Physical Examination,
including digital rectal exam, within eight
weeks prior to registration
- Clinical stage T1-T2C* as per the AJCC**,
6
th
edition
- PSA < 10 ng/ml
- KPS > 90
- Age > 18
- Patient must sign study-specific informed
consent.
- Prior or concurrent invasive malignancy in
last 5 years
- Evidence of distant metastasis
- Regional lymph node involvement
- Previous radical surgery prostate cancer
- Previous treated with pelvis irradiation,
prostate -brachytherapy , or bilateral
orchiectomy
- Previous or concurrent chemotherapy for
prostate cancer
- History of Ulcerative Colitis or Crohn’s
Disease
- Concurrent Coumadin administration
*

Clinical stage (TStage): size and/or extent of the primary tumor in low and intermediate risk groups
** The American Joint Committee on Cancer
92

9.3 Enhanced Hypofractionation Protocol Workflow with ePR System
Once the ePR is deployed, the workflow of the hypofractionation treatment is enhanced
with 3 additional steps, 2b, c, and d, depicted as yellow boxes in Figure 9-2. In step 2,
after the dose constraints are approved, the clinicians can review existing patients’
treatment plans, steps 2b and 2c, and make a decision whether the patient should continue
with the hypofractionation protocol, step 2d, based on the data available for review and
comparison from the ePR. The three steps are described in detail below:
• Step 2b: The clinician is guided by a knowledge-driven Search Engine, discussed
in Chapter 6. The conditions of the query command includes but not limited to
demographics, prior treatment conditions, such as pre-treatment PSA level,
number of positive biopsies cores, pathology cancer stage (Gleason score) and
clinical cancer stage, percentage volume of critical structures covered by specific
dose, and side effects interfering activities of daily living (ADL) (toxicity grade ≥
3). The side effects are used to compute radiation toxicity grade which in turn
helps physicists and oncologists to predict the outcome of a new treatment from
previous similar patients in the database.
• Step 2c: After the search results are returned, the clinician can review previous
patients’ treatment plans. The interactive functions of the Intelligent Treatment
Plan Navigator, described in Chapter 5, are used in this step to navigate along the
DVH curve and determine high dose regions in bladder and rectum in CT images.
• Step 2d: The physicists working with the oncologist can determine if the patient
should continue with the hypofractionation protocol. Three scenarios might
happen:
93

o Yes: The plan is executed.
o No: The patient is advised to be treated with an existing dose escalation
protocol or try a non- proton therapy treatment. He will have a choice to
follow the existing treatment if the toxicity is tolerable to him.
o The treatment can be modified to reduce negative effects. The workflow
step will then go back to the beginning of step 2 for another iteration.

Figure 9-2 Enhanced hypofractionation workflow with ePR system. Blue boxes: Existing
hypofractionation protocol workflow. Yellow boxes: additional workflow steps with
ePR implemented. Steps 2b, c, and d: Additional steps to utilize ePR system for
decision support.
The added value of the enhanced workflow is to allow the practice of knowledge-based
medicine in the new hypofractionation protocol treatment, in which the physicists and
oncologists will have decision support tools to facilitate an optimal treatment to be
delivered to the patients. The knowledge used in this workflow was explained in Sections
6.2 and 6.3.
94

9.4 Clinical Example of Utilizing Knowledge Discovery for Decision
Support
9.4.1 Description of Hypothetical Patient
Patient Demo S. (Pt. Demo) is one hypothetical patient (see Appendix A) which was
created by an expert specialized on proton therapy hypofractionation protocol treatment
at the James M. Slater, M.D. Proton Treatment and Research Center, Loma Linda, CA
(LLUMC) for demonstration purpose. His patient chart shows that this 60 year-old
patient was diagnosed with clinical stage T1c, entry PSA measurement of 5.2 ng/ml, and
Gleason score of 6, and treated with hypofractionation protocol with 60 Gy total dose and
3Gy dose per fraction; and he developed rectal toxicity of grade 3.
Pt. Demo was created and used in a simulated clinical environment to test the knowledge
discovery in the ePR for decision support. The next section describes the use of the
navigation and data mining tools on how to utilize Pt. Demo and other existing
hypofractionation treatment plans to make a decision on treatment of a new patient with
matched pretreatment data, tumor size and shape.
9.4.2 Clinical Example: Patient Criteria Assessment using the ePR
The following description is a clinical example of utilizing knowledge within the ePR for
decision support. Patient New (Pt. New), 60 years old, who was diagnosed with prostate
cancer and decides to treat with proton therapy at LLUMC. The patient would like a
shorter treatment time so he could go back home early. Pt. New’s clinical information is
registered in the ePR, shown in red boxes and circles in Figure 9-3, to be determined
whether he is eligible for hypofractionation treatment.
95

This is one benefit of the ePR system, my first contribution described in Chapter 8, since
rules can be made to guide the decision based on these clinical information in searching
for previous patients’ treatment plans in the ePR system. For example, the typical search
will use important parameters, such as: age, clinical stage, Gleason score, and entry PSA
measurement.
The result of the eligibility assessment is shown in Figure 9-4. In this case, he satisfies
the patient criteria for hypofractionation protocol treatment. The oncologist at LLUMC
introduces him to a new method that could save him 4 weeks of treatment at Loma Linda
with 60Gy of total dose, in comparison to 9 weeks with the current clinically accepted
proton therapy dose escalation protocol with a higher dose of 81Gy. The patient is
concerned whether this treatment would give him more side effects later since the
treatment time is shorter but dose is administered sooner thus not allowing the patient to
heal in time. The oncologist tries to optimize the plan to determine if Pt. New is suitable
for this new hypofractionation protocol. The next section describes the decision making
process utilizing knowledge from the ePR system.
96

Figure 9-3 Screenshot of Patient New’s information, in red boxes and circles, registered
to the ePR to determine his eligibility for hypofractionation protocol.
97

Figure 9-4 Screenshot of Pt. New criteria assessment result that he satisfied the criteria.
9.4.3 Clinical Example: Knowledge Discovery Utilizing the ePR
This section describes an example of knowledge discovery utilizing the ePR system to
improve treatment of a new patient based on treatment plans from previous patients.
Figure 9-5 shows the step-by-step guideline of this clinical example in this section.

Figure 9-5 Step-by-step guideline of the clinical example
9.4.3.1 Data Mining
This section describes data mining steps, shown in left box of Figure 9-5. First, the
pretreatment data and age of Pt. New are used to determine search criteria of previous
patients with hypofractionation treatment plans in the ePR system. Table 9-2 shows Pt.
New information in comparison with search criteria which will be used to search for
matched patients. After that, the Search Engine was used to find the matched patients.

98

Table 9-2 Patient New information and treatment plan in comparison to search criteria
Patient New Search Criteria
Age 60 55-65*
Clinical Condition T1c T1c
Gleason Score 6 < 7*
Entry PSA
Measurement
5.2 ng/ml From 4.0 ng/ml and 6.0 ng/ml*
Total Dose 60Gy 60Gy
Dose per Fraction 3Gy 3Gy
(*Wider search range to expand results to more closely match patients)
Figure 9-6 shows the GUI of the Search Engine Query Command Page, where the
oncologist enters the querying parameters which are relevant to Pt. New. The result of
this search provides a list of 7 patients, shown in Figure 9-7.
99

Figure 9-6 Screenshot of knowledge-based search engine query input for Pt. New (see
red ellipses and boxes in Figure 9-3)

100

Figure 9-7 Screenshot of knowledge-based search results displaying previous patients
that have undergone new hypofractionation treatment protocol who have toxicity
grades ranging from 0 to 3
9.4.3.2 Selection of Hypofractionation Patients Matched Search Criteria
Referring to the middle box in Figure 9-5, this section describes the steps that clinicians
perform to choose a treatment plan from the search results in order to create a plan for Pt.
New. The treatment plan will be chosen based on tumor size/volume, toxicity grade and
tumor shape.
First, the tumor volume of Pt. New is 81cc; hence, two patients among the seven matched
this criterion of having similar prostate volumes, shown with red ellipses in Figure 9-7.
These two matched previous patients have different toxicity grades 3 and 0 but both are
utilized to help the physicist to evaluate the treatment.
Second, the patient who had lowest toxicity would be selected. Therefore, patient
Twenty-One W. (Pt. Twenty-One), who had a positive outcome of rectal toxicity of 0, is
the first patient to be utilized to create treatment plans for Pt. New.
101

Lastly, when the tumor size is taken into consideration, the plan of Pt. Twenty-One could
not be used for Pt. New because the shapes of the lower boundary of the prostate tumor
are quite different. Figure 9-8 shows different shape of Pt. New’s prostate (top), wrapping
on top of the rectum, and Pt. Twenty-One’s prostate (bottom). The oncologist rejects to
follow Pt. Twenty-One’s treatment plan because it would not provide good tumor
coverage to achieve optimal tumor control because the proton beam for Pt. New would be
radiated to the two hip joints which could create possible higher toxicity grade (see green
arrows in Figure 9-8).
102

Figure 9-8 Difference in the shapes of the prostates, red contours, of Pt. New (top) and
Pt. Twenty-One (bottom)
9.4.3.3 Modification of Treatment Plan to Suite the New Patient
However, the treatment plan of Pt. Demo with toxicity grade 3 was a concern. His
treatment is loaded to the TP Evaluation page to evaluate Pt. Demo treatment plan, first
task in the right box of Figure 9-5. The dose parameters from this treatment plan were
utilized to guide the physicist in Pt. Demo treatment to avoid a possible negative outcome
103

of high toxicity. This TP Evaluation page was implemented with ITPN functionality,
discussed in Chapter 5, so the physicist and oncologist could utilize knowledge
discovered from the DICOM-RT objects as well as the quantified data to evaluate
treatment plans. Also, from the treatment plan of Pt. Demo, the physicist learned that the
margin of beam around the prostate was 10mm. Figure 9-9 displays the beam
configuration captured from the TPS system, the beam margin around the prostate is in
the red box.

Figure 9-9 Screenshot of beam margin configuration of Pt. Demo showing the beam
margin of 10mm around prostate.
In addition, upon reviewing the existing treatment plan, selected CT slices were shown
areas of potential high dose to critical structures, i.e. rectum and bladder, using the ITPN
tool, as in Figure 9-10. Changes are made to the treatment plan of Pt. New to try to
reduce the quantitative amount of the rectum’s and bladder’s dose volume that would
receive higher proton dose, second task in the right box of Figure 9-5. These changes are
104

performed in the TPS and the results are sent to the ePR. In the modified plan of Pt. New,
the beam margin was adjusted to 7mm from the border of prostate. Table 9-3 shows the
dose volume difference in Pt. New’ treatment plan: the maximum volume percentages of
adjacent critical structures which receive 60Gy have been reduced from 6.4% to 1.9% for
rectum, and from 1.6% to 0% for bladder. Note that the volume percentages in both plans
are less than 15% dose limits for rectum and bladder, the first row for rectum section of
Table 6-1. This reduction promises a better outcome of Pt. Demo with radiation toxicity
less than grade 3.
105

Figure 9-10 Screenshot of ITPN shows overlapping region of the isodose curve 95%
(orange), target prostate (pink) and crtical organs: rectum (green) and bladder
(yellow). The isodose 95% covered entire prostate.
Table 9-3 Comparison of volume percentage of the tumor target and critical structures
between previous plan and modified plan of Pt. New
Prostate at 55Gy Rectum at 60Gy Bladder at 60Gy
Previous Plan > 95% 6.4% 1.6%
Modified Plan > 95% 1.9% 0%
Limit* n/a <15% <15%
(* See the first row for rectum section of Table 6-1)
106

Figure 9-11 displays the DVH curves of two plans on the ITPN GUI, which shows no
change in target volume and a decreasing of percentage rectum volume receiving 55Gy
from 25.7% to 18.18%.
After the treatment plan was finished with the new dose constraints and recalculated in
the proton therapy treatment planning system (TPS), the data is sent to the ePR for
evaluation and validation, last task in the right box of Figure 9-5.
107

Figure 9-11 Screenshots DVH curves on the ITPN of Pt. New shows 100% volume of
prostate receiving 55Gy but a decreasing volume percentage of rectum receiving
55Gy from 25.70% to 18.18% with the modified hypofractionation plan of 7mm
beam margin was used.
108

9.5 Summary
The practice of radiation therapy involves decision making where oncologists are
challenged with evaluating treatment options for cancer patients to ensure optimal
treatment for their disease. Traditionally, clinical experience and expert knowledge in
literature served as the basis for medical practice. However, experience and expertise
alone may not necessarily prevent a negative result and promise an effective outcome
[31]. Therefore, knowledge-based or evidence-based medicine is a new trend in current
clinical practice where disease management and decision making relies on clinical
evidence in combination with clinician’s experience and expertise, patients’ preferences,
and an easily accessible patient information and large clinical based treatment plans
database.
In the field of radiation oncology, this knowledge-based practice is manifested in the
development and utilization of protocols based on treatment recommendations for a
particular disease. In order to make such a recommendation, there is a rigorous process of
research and evaluation of the effectiveness of new treatment protocols. The process for
new treatment protocol evaluation requires the first step of acquiring relevant and
complete information of the patient and then evaluating this information to maximize the
positive effects in the treatment outcome. For example, if the research of proton therapy
hypofractionation protocol treatment proves an equivalent or better result compared to
existing dose escalation protocols, the shorter course of proton therapy would benefit not
only prostate cancer patients coming from different geographic locations, but also those
who need a high accuracy level of dose delivery as an alternative treatment option due to
the overall cost and time.
109

The standard practice is normally based on scientific evidence generated from population
studies. However, scientific evidence and standardized knowledge from large population
is not always effective in cancer management to individual patients with unique
anatomical structures, tumor size and shapes, and clinical, psychological and social
circumstances [31]. The Proton Therapy ePR system in this research provides clinicians a
new paradigm to utilize information in existing patient demographics and treatment data,
allows knowledge discovery of this data to help oncologists and physicists make optimal
decisions and provide personalized care for prostate cancer patients in proton therapy. All
the demographic, treatment plans and outcomes data are stored in an easily accessible
database. The example shown in this chapter illustrates the step-by-step reasoning and
utilization of data mining and navigation tools in the ePR to achieve knowledge
discovery and evidence-based treatment plan which is novel and a major contribution of
my research.

110

Chapter 10 Current Status and Future Direction
10.1 Current status
The development of the proton therapy ePR system has been completed. A
comprehensive testing and evaluation has been performed at IPILab on 41
hypofractionation prostate cancer patient cases collected from LLUMC. The system was
developed to support web-based medical imaging intensive applications with server-
client architecture. This architecture together with implementation of HTTPS and
DICOM communication allows different data types, such as treatment plans and patient
clinical data, to be archived, visualized and distributed across multiple locations by the
local area network and/or Internet. The PT ePR system is DICOM and HIPAA compliant.
Currently, the system is verified to have the following features:
• Handle text input on electronic forms for initial and follow-up data from web-
based GUIs,
• Display non-DICOM data with a web-browser at a client WS,
• Perform DICOM communications with the TPS to receive DICOM RT ION
objects,
• Use the ITPN to display and navigate between CT images and knowledge
extracted from DICOM RT ION objects, such as DVH curves, contours, isodose
curves, and
• Search for matched patient criteria using the knowledge driven Search Engine.
The proton therapy ePR system is ready for implementation for a clinical trial. Moreover,
the workflow analysis in Chapter 8 has proven the efficiency of the ePR system in data
111

collection and integration for evaluating new treatment protocols of proton therapy. As
LLUMC has approved the ePR for implementation at the proton therapy facility in the
clinical setting, it promises a new frontier for research in proton therapy and medical
imaging informatics. Data will be archived in a secure system and easily accessed and
retrieved by authorized clinicians and researchers. Beyond the efficiency for data
collection, the more data is collected, the more knowledge can be derived and discovered
to improve treatment plan for new patients, as illustrated in Chapter 9.
10.2 Future Research Direction
The development of the proton therapy ePR system can evolve to the next phase: clinical
implementation and evaluation at proton therapy facility of LLUMC. The clinical
implementation will include training, users’ feedback and performance testing when the
system acquires a larger cohort of patients and interacts with different type of users, such
as nurses, oncologists and physicists. In addition to the evaluations performed at IPLab,
discussed in Chapter 7, the following evaluations can be achieved:
• User feedback from GUI to evaluate the effectiveness of the ePR when interacting
with users,
• Time performance to evaluate the efficiency of the ePR in data distribution, and
• Stability and reliability of the ePR system in handling a larger cohort of patient
cases compared to the current 41 cases already collected as well as multiple user
feedback for system improvement from different locations.
112

In the beginning phase of clinical implementation, the staff at the clinical site needs to
perform double duties to maintain the existing workflow in parallel to start using the ePR
system so that the normal working workflow would not be interrupted.
In addition to clinical evaluation, the development of the ePR can continue to be
developed to address the issue of patient reconciliation and improve the ITPN to display
3D surfaces and data mining tool to find correlation between radiation toxicity and dose
received on each critical structure.
As mentioned earlier, a USC IRB (Institute Review Board) with study number HS-08-
00652 was applied and approved on March 02, 2009 to conduct research studies on
human subjects while respecting their rights and welfare. This approval will allow
collecting 300 datasets of prostate cancer patients treated with proton therapy
hypofractionation protocol. Besides the 41 cases already collected for this thesis research,
the new data collection can be performed using the ePR once the ePR is deployed at
LLUMC. The collected data can be used for future large-scale outcomes analyses which
are listed below:
• Efficiency of using the ePR for data collection, i.e. improvement in time
performance and missing data rate of the workflow with ePR system implemented
in comparison to existing workflow without ePR system.
• Effectiveness of using the ePR for knowledge discovery and data mining on larger
datasets of hypofractionation prostate cancer patients, i.e. any change of treatment
protocol and whether the modified treatment plans provide better outcomes such
as lower radiation toxicity.
113

Since the workflow and dataflow are similar among many radiotherapy treatment
protocols, the implementation of this proton therapy ePR system can be extended beyond
hypofractionation prostate cancer patients to other types of cancer and even to other
radiotherapy methods.
114

References
1. National Cancer Institute. Prostate Cancer Home Page. 2009 [cited 2009 June 25];
Available from: http://www.cancer.gov/cancertopics/types/prostate.
2. O’Dowd, G.J., Veltri, Robert W., Miller, Craig M., Strum, Stephen B., The Gleason
Score: A Significant Biologic Manifestation of Prostate Cancer Aggressiveness On
Biopsy. 2001, PCRI Insights. p. 1-5.
3. Liu, B.J., et al. A knowledge-based imaging informatics approach to managing patients
treated with proton beam therapy. in Medical Imaging 2007: PACS and Imaging
Informatics. 2007. San Diego, CA, USA: SPIE.
4. The National Association for Proton Therapy. PROTON THERAPY - The National
Association for Proton Therapy (NAPT) Proton Beam Therapy, Protons and Prostate
Cancer. [cited 2009 April 20]; Available from: http://www.proton-therapy.org/.
5. Slater, J.D., et al., Proton therapy for prostate cancer: the initial Loma Linda University
experience. International Journal of Radiation Oncology*Biology*Physics, 2004. 59(2):
p. 348-352.
6. Hanks, G.E., et al., PSA confirmation of cure at 10 years of T1B, T2, N0, M0 prostate
cancer patients treated in RTOG protocol 7706 with external beam irradiation. Int J
Radiat Oncol Biol Phys, 1994. 30(2): p. 289-92.
7. Brenner, D.J. and E.J. Hall, Fractionation and protraction for radiotherapy of prostate
carcinoma. International Journal of Radiation Oncology*Biology*Physics, 1999. 43(5):
p. 1095-1101.
8. Huang, H.K., Integration of HIS, RIS, PACS, and ePR, in PACS and Imaging
Informatics. 2004, John Wiley & Sons: Hokoben, New Jersey. p. 323.
9. Ballas, L.K., et al., Radiation therapy facilities in the United States. International Journal
of Radiation Oncology*Biology*Physics, 2006. 66(4): p. 1204-1211.
10. Zietman, A.L., W.U. Shipley, and H. Tsuji, Prostate, in Proton and Charged Particle
Radiotherapy, T.F.a.K. Delaney, Hanne M., Editor. 2008, Lippincott Williams &
Wilkins: Philadelphia. p. 213-222.
11. Law, M.Y.Y., A model of DICOM-based electronic patient record in radiation therapy.
Computerized medical imaging and graphics : the official journal of the Computerized
Medical Imaging Society, 2005. 29(2): p. 125-136.
12. Law, M.Y.Y. and B. Liu, Informatics in Radiology: DICOM-RT and Its Utilization in
Radiation Therapy. Radiographics, 2009. 29(3): p. 655-667.
13. Cooperberg, M.R., et al., The Contemporary Management of Prostate Cancer in The
United States: Lessons From The Cancer Of The Prostate Strategic Urologic Research
Endeavor (Capsure), A National Disease Registry. 2004. p. 1393-1401.
14. Kupelian, P.A., et al., Radical prostatectomy, external beam radiotherapy <72 Gy,
external beam radiotherapy >=72 Gy, permanent seed implantation, or combined
seeds/external beam radiotherapy for stage T1-T2 prostate cancer. International Journal
of Radiation Oncology*Biology*Physics, 2004. 58(1): p. 25-33.
115

15. Vicini, F., et al., Radiation dose for prostate cancer: is more better? Nat Clin Prac Oncol,
2006. 3(6): p. 298-299.
16. Zietman, A.L., et al., 10-Year Outcome for Men With Localized Prostate Cancer Treated
With External Radiation Therapy: Results of a Cohort Study. 2004. p. 210-214.
17. Delaney, T.F., Clinical Issues in Proton Radiotherapy, in Proton and Charged Particle
Radiotherapy, T.F.a.K. Delaney, Hanne M., Editor. 2008, Lippincott Williams &
Wilkins: Philadelphia. p. 108.
18. Zietman, A., et al., The Patterns of Care Survey of radiation therapy in localized prostate
cancer: Similarities between the practice nationally and in minority-rich areas.
International Journal of Radiation Oncology*Biology*Physics, 2001. 50(1): p. 75-80.
19. Zelefsky, M.J., et al., High-dose intensity modulated radiation therapy for prostate
cancer: early toxicity and biochemical outcome in 772 patients. International Journal of
Radiation Oncology*Biology*Physics, 2002. 53(5): p. 1111-1116.
20. Zietman, A.L., et al., Comparison of Conventional-Dose vs High-Dose Conformal
Radiation Therapy in Clinically Localized Adenocarcinoma of the Prostate: A
Randomized Controlled Trial. JAMA, 2005. 294(10): p. 1233-1239.
21. Slater, J.D., Schulte, Reinhard W., Proton-Beam vs Intensity-Modulated Radiation
Therapy: Too Soon for a Randomized Trial. Oncology, 2008. 22(7).
22. National Cancer Institute. Cancer Therapy Evaluation Program: Protocol Development.
2009 [cited 2009 June 25]; Available from:
http://ctep.cancer.gov/protocolDevelopment/electronic_applications/ctc.htm.
23. Liu, B.J., et al., Image-assisted knowledge discovery and decision support in radiation
therapy planning. Computerized Medical Imaging and Graphics, 2007. 31(4-5): p. 311-
321.
24. NEMA, DICOM Standard, in Information Object Definitions. 2009.
25. The Apache Software Foundation. Welcome! - The Apache HTTP Server Project. [cited
2010 April 18]; Available from: http://httpd.apache.org/.
26. Oracle Corporation and/or its affiliates. MySQL :: The world's most popular open source
database. [cited 2010 April 18]; Available from: http://www.mysql.com/.
27. The PHP Group. PHP: Hypertext Preprocessor. [cited 2010 April 18]; Available from:
http://php.net/.
28. Bourdon, R. Install PHP 5 Apache MySQL on Windows: WampServer. [cited 2010
April 18]; Available from: http://www.wampserver.com/en/.
29. The jQuery Project. jQuery: The Write Less, Do More, JavaScript Library. [cited 2010
April 18]; Available from: http://jquery.com/.
30. Chung, H.T., Prostate Cancer, in Radiation Oncology: An Evidence-based Approach,
L.W. Brady, et al., Editors. 2008, Springer-Verlag: Berlin Heidelberg. p. 302.
31. Lu, J.J. and L.W. Brady, Introduction, in Radiation Oncology: An Evidence-based
Approach, L.W. Brady, et al., Editors. 2008, Springer-Verlag: Berlin Heidelberg. p. VII-
XVI.

116

Alphabetical References
Ballas, L. K., E. B. Elkin, et al. (2006). "Radiation therapy facilities in the United States."
International Journal of Radiation Oncology*Biology*Physics 66(4): 1204-1211.
Bourdon, R. "Install PHP 5 Apache MySQL on Windows: WampServer." Retrieved April 18,
2010, from http://www.wampserver.com/en/.
Brenner, D. J. and E. J. Hall (1999). "Fractionation and protraction for radiotherapy of prostate
carcinoma." International Journal of Radiation Oncology*Biology*Physics 43(5): 1095-
1101.
Chung, H. T. (2008). Prostate Cancer. Radiation Oncology: An Evidence-based Approach. L. W.
Brady, H.-P. Heilman, M. Moils and C. Nieder. Berlin Heidelberg, Springer-Verlag: 302.
Cooperberg, M. R., J. M. Broering, et al. (2004). The Contemporary Management of Prostate
Cancer in The United States: Lessons From The Cancer Of The Prostate Strategic
Urologic Research Endeavor (Capsure), A National Disease Registry. 171: 1393-1401.
Delaney, T. F. (2008). Clinical Issues in Proton Radiotherapy. Proton and Charged Particle
Radiotherapy. T. F. a. K. Delaney, Hanne M. Philadelphia, Lippincott Williams &
Wilkins: 108.
Hanks, G. E., C. A. Perez, et al. (1994). "PSA confirmation of cure at 10 years of T1B, T2, N0,
M0 prostate cancer patients treated in RTOG protocol 7706 with external beam
irradiation." Int J Radiat Oncol Biol Phys 30(2): 289-292.
Huang, H. K. (2004). Integration of HIS, RIS, PACS, and ePR. PACS and Imaging Informatics.
Hokoben, New Jersey, John Wiley & Sons: 323.
Kupelian, P. A., L. Potters, et al. (2004). "Radical prostatectomy, external beam radiotherapy <72
Gy, external beam radiotherapy >=72 Gy, permanent seed implantation, or combined
seeds/external beam radiotherapy for stage T1-T2 prostate cancer." International Journal
of Radiation Oncology*Biology*Physics 58(1): 25-33.
Law, M. Y. Y. (2005). "A model of DICOM-based electronic patient record in radiation therapy."
Computerized medical imaging and graphics : the official journal of the Computerized
Medical Imaging Society 29(2): 125-136.
Law, M. Y. Y. and B. Liu (2009). "Informatics in Radiology: DICOM-RT and Its Utilization in
Radiation Therapy." Radiographics 29(3): 655-667.
Liu, B. J., H. K. Huang, et al. (2007). A knowledge-based imaging informatics approach to
managing patients treated with proton beam therapy. Medical Imaging 2007: PACS and
Imaging Informatics, San Diego, CA, USA, SPIE.
Liu, B. J., M. Y. Y. Law, et al. (2007). "Image-assisted knowledge discovery and decision
support in radiation therapy planning." Computerized Medical Imaging and Graphics
31(4-5): 311-321.
Lu, J. J. and L. W. Brady (2008). Introduction. Radiation Oncology: An Evidence-based
Approach. L. W. Brady, H.-P. Heilman, M. Moils and C. Nieder. Berlin Heidelberg,
Springer-Verlag: VII-XVI.
117

National Cancer Institute. (2009). "Cancer Therapy Evaluation Program: Protocol Development."
Retrieved June 25, 2009, from
http://ctep.cancer.gov/protocolDevelopment/electronic_applications/ctc.htm.
National Cancer Institute. (2009). "Prostate Cancer Home Page." Retrieved June 25, 2009, from
http://www.cancer.gov/cancertopics/types/prostate.
NEMA (2009). DICOM Standard. Information Object Definitions.
O’Dowd, G. J., Veltri, Robert W., Miller, Craig M., Strum, Stephen B. (2001). The Gleason
Score: A Significant Biologic Manifestation of Prostate Cancer Aggressiveness On
Biopsy, PCRI Insights. 4: 1-5.
Oracle Corporation and/or its affiliates. "MySQL :: The world's most popular open source
database." Retrieved April 18, 2010, from http://www.mysql.com/.
Slater, J. D., C. J. Rossi, et al. (2004). "Proton therapy for prostate cancer: the initial Loma Linda
University experience." International Journal of Radiation Oncology*Biology*Physics
59(2): 348-352.
Slater, J. D., Schulte, Reinhard W. (2008). "Proton-Beam vs Intensity-Modulated Radiation
Therapy: Too Soon for a Randomized Trial." Oncology 22(7).
The Apache Software Foundation. "Welcome! - The Apache HTTP Server Project." Retrieved
April 18, 2010, from http://httpd.apache.org/.
The jQuery Project. "jQuery: The Write Less, Do More, JavaScript Library." Retrieved April 18,
2010, from http://jquery.com/.
The National Association for Proton Therapy. "PROTON THERAPY - The National Association
for Proton Therapy (NAPT) Proton Beam Therapy, Protons and Prostate Cancer."
Retrieved April 20, 2009, from http://www.proton-therapy.org/.
The PHP Group. "PHP: Hypertext Preprocessor." Retrieved April 18, 2010, from http://php.net/.
Vicini, F., L. Kestin, et al. (2006). "Radiation dose for prostate cancer: is more better?" Nat Clin
Prac Oncol 3(6): 298-299.
Zelefsky, M. J., Z. Fuks, et al. (2002). "High-dose intensity modulated radiation therapy for
prostate cancer: early toxicity and biochemical outcome in 772 patients." International
Journal of Radiation Oncology*Biology*Physics 53(5): 1111-1116.
Zietman, A., J. Moughan, et al. (2001). "The Patterns of Care Survey of radiation therapy in
localized prostate cancer: Similarities between the practice nationally and in minority-
rich areas." International Journal of Radiation Oncology*Biology*Physics 50(1): 75-80.
Zietman, A. L., C. S. Chung, et al. (2004). 10-Year Outcome for Men With Localized Prostate
Cancer Treated With External Radiation Therapy: Results of a Cohort Study. 171: 210-
214.
Zietman, A. L., M. L. DeSilvio, et al. (2005). "Comparison of Conventional-Dose vs High-Dose
Conformal Radiation Therapy in Clinically Localized Adenocarcinoma of the Prostate: A
Randomized Controlled Trial." JAMA 294(10): 1233-1239.
Zietman, A. L., W. U. Shipley, et al. (2008). Prostate. Proton and Charged Particle Radiotherapy.
T. F. a. K. Delaney, Hanne M. Philadelphia, Lippincott Williams & Wilkins: 213-222.

118

Appendices
Appendix A Patient Demo S
Reinhard Schulte Hypofractionation Protocol – EPR Example March 2010
Example Case 2: Late Rectal Toxicity
PRESENT HISTORY
The patient is a 60 year old man who presented for consultation to discuss treatment
options for his early stage prostate cancer. The patient was diagnosed with clinical stage
T1c Nx Mx prostate cancer. Over the last three years, he presented with increasing prostate‐
specific antigen (PSA) levels as listed in Table 1. A transrectal ultrasound‐guided needle
biopsy of the prostate gland prostate in 11/2007 was negative for cancer but a renewed
biopsy with 20 biopsy cores in 05/2009 revealed adenocarcinoma in the right and left lobe
of his prostate, Digital rectal exam (DRE) was normal (clnical T stage T1c).
Table 1. Pre‐Treatment PSA
Date PSA (ng/ml)
08/06 2.2
02/09 5.6
02/07 5.2
HISTOPATHOLOGY
Histopathology showed a moderately differentiated adenocarcinoma of the prostate
Gleason score 3 + 3 = 6 in 30% of the tissue from the right lobe and 50% of the tissue from
the left lobe. Due to the low risk of lymph node and metastatic disease in T1c tumors with
Gleason score < 7 and PSA values <20 ng/ml, additional imaging studies were not
performed in this patient.
PAST HISTORY
The patient stated that he had no previous radiation nor chemotherapy and has not had a
diagnosis of malignant disease.
The patient has had obstructive urinary symptoms for the past 3‐4 years that have been
attributed to benign prostatic hypertrophy. He was not taking medication for this at the
time of treatment.
Page 1
119

Reinhard Schulte Hypofractionation Protocol – EPR Example March 2010
REVIEW OF SYSTEMS (relevant to prostate cancer)
The patient stated to have urinary urgency, urinary frequency, and a decrease in the urinary
stream. He was not on anti‐androgen therapy. He did not indicate to have erectile
dysfunction (ED).
His performance status was 100% (normal, no complaints, no evidence of disease). An age
and treatment status adjusted quality of life factor of 100 was assigned. The American
Urological Association (AUA) Symptom Index for BPH had a score of 20 (on a scale from 0 to
40 points).
PHYSICAL EXAMINATION (relevant to prostate cancer)
DRE of the prostate revealed no abnormality.
TREATMENT PROTOCOL
The patient agreed to be treated on a hypofractionation Phase I/II study protocol at Loma
Linda University Medical Center. He was treated with protons along to a final tumor dose of
60 Gy. The treatment consisted of 20 fractions and was given over 33 calendar days. No
additional systemic therapy (anti‐androgen or chemotherapy) was given.
DOSE‐VOLUME PARAMETERS (to be added by Anh)
EVALUATION DURING TREATMENT
ACUTE TOXICITY
GU Symptoms
Cystitis: The patient experienced a mild amount of cystitis (frequency with dysuria) toward
the end of treatment. Corrective measures were taken.
CTCAEv3 Grade: Cystitis ‐ Frequency with dysuria; macroscopic hematuria, grade 2.
CTCAEv3 Attribute (unrelated, unlikely, possible, probable, or definite): definite.
GI Symptoms: none
Constitutional Symptoms: none
Page 2
120

Reinhard Schulte Hypofractionation Protocol – EPR Example March 2010
FOLLOWUP
EVALUATION HISTORY
The patient was followed by his urologist every three‐four months for the first 2 years, and
then yearly for another 2 years. He then moved to another state and was lost to follow‐up.
PATIENT STATUS
Date of last contact: 05/2013
Status: Alive without signs of prostate cancer but with significant treatment related
morbidity (see below).
TUMOR FOLLOW‐UP
The PSA values obtained by the patient’s urologist are listed in Table 2. DREs performed by
the urologist were normal without evidence of new nodules suspicious for recurrence.Table
1. Pretreatment PSA.
Table 2. Post‐Treatment PSA
Date PSA (ng/ml)
04/10 3.5
08/10 2.5
01/11 1.4
09/11 0.5
08/12 0.4
05/13 0.5

Tumor status at last follow‐up: no evidence for recurrent disease.
LATE TOXICITY (ADVERSE EVENTS)
Genitourinary: none
Page 3
121

Reinhard Schulte Hypofractionation Protocol – EPR Example March 2010
Gastrointestinal
Proctitis: Beginning 34 months after radiation treatment the patient noticed occasional pain
during defecation. At 36 months following radiation treatment, lancinating rectal pain with
defecation began to develop. His symptoms progressed to include frequent rectal bleeding
(red blood on stool) and increasingly worse rectal pain with bowel movements, lasting up to
30 minutes. Hydrocortisone and mesalamine rectal suppositories offered no relief at that
point. Ten months after completion of radiation therapy, the patient was unable to work on
a regular basis because of severe pain on defecation, with sitting or driving. He underwent
rectal examination under anesthesia. Areas of superficial ulceration were seen associated
with scarring along the anterior rectal mucosa. The posterior rectal mucosa was apparently
normal. There was no evidence of external hemorrhoidal thrombosis. A botulinum toxin
injection to the internal anal sphincter was performed was performed at that time but
resulted in no relief. His pain was severe enough to require OxyContin (Purdue Pharma LP,
Stamford, CT) 40 mg twice a day and diazepam. He still had considerable tenesmus, but no
bleeding. His physicians recommended a diverting colostomy, which was eventually
performed 38 months after treatment.
CTCAEv3 Grade: Proctitis ‐ Stool incontinence or other symptoms interfering with activities
of daily life (ADL); operative intervention indicated; grade 3.
CTCAEv3 Attribute (unrelated, unlikely, possible, probable, or definite): definite.
Constitutional:
Insomnia: With increasing pain symptoms due to proctitis that patient could had difficulties
falling asleep and frequently woke up at night.
CTCAEv3 Grade: Insomnia ‐ Frequent difficulty sleeping interfering with ADL, grade 3.
CTCAEv3 Attribute (unrelated, unlikely, possible, probable, or definite): definite.
Sexual:
Libido: The patient reported that his rectal pain and the pain medication he required
interfered with his interest in sexual activity to a point that affected his relationship with his
wife.
CTCAEv3 Grade: Libido ‐ Decrease in interest and adversely affecting relationship;
intervention indicated, grade 2.
CTCAEv3 Attribute (unrelated, unlikely, possible, probable, or definite): definite.
Page 4
122

Appendix B Follow-up Questionnaires
Anjali Davis, RN, Version 1.0 Sep. 18, 2009
Reinhard Schulte, MD

LOMA LINDA UNIVERSITY MEDICAL CENTER
DEPARTMENT OF RADIATION MEDICINE
Prostate Cancer Hypo fractionation Protocol
Questionnaire
Side Effect Profile Year 1 (CTC AE Version 3.0)
Category:
Gastrointestinal:
1. Have you had to alter any dietary habits after your Proton treatments for Prostate
Cancer?
2. Have you had any change in bowel habits after the treatments ended? Is it associated
with meals?
3. Have you noticed any pelvic pain/discomfort after the treatments ended? If so what
measures have you taken to relieve it?
4. Have you lost any weight after the treatments ended?
5. Have you had any rectal bleeding? If so, how much and how often?
6. Have you had any rectal discomfort with bowel movements?
7. Have you noticed any incontinence of bowel movements? If so what measures are you
taking for it?
Genitourinary:
1. Have you noticed any change in urinary frequency since the treatments ended? If so
what is the change?
2. Have you noticed any discomfort with urination? If so, have you taken any measures to
relieve the discomfort?
3. Have you noticed any abnormalities with the urine flow after the treatments ended?
4. Have you had any incidence of retention of urine? If so what measures were taken to
relieve it?
5. Have you noticed any blood in the urine since your treatments ended?
6. Have you noticed any urinary incontinence or dribbling since your treatments ended? If
so what measures have you taken for it?
7. Do you wake up at night to urinate? If so how many times? Are you taking anything
for it?
8. How many times did you wake up before and during the treatments?
Sexual function:
1. Have you noticed any erectile dysfunction, after the treatments ended? If so are you
taking anything for it?
2. Have you noticed any ejaculatory dysfunction, after the treatments ended?
Page 1
123

Anjali Davis, RN, Version 1.0 Sep. 18, 2009
Reinhard Schulte, MD

Category
Constitutional or other (specify):
1. Have you noticed any other changes since your treatments ended?
2. Have you sought any medical help?

Page 2

Abstract (if available)

Abstract Radiotherapy (RT) is an image-intensive specialty in medicine that utilizes radiation for treatment of cancer. Similar to x-ray beam treatment, proton therapy (PT) utilizes high energy particle proton beam to kill cancer cells at the target region. Due to the physical properties of the proton beam, PT delivers dose with higher accuracy than conventional radiotherapy with x-ray beams and no exit dose. In PT, patient data are scattered in different systems within the oncology department which hinders the research on many topics such as whether PT with higher dose or different delivery schedules is more beneficial to reducing toxicity of prostate cancer patients. This research aims to introduce a data mining method and a treatment plan navigator utilizing the infrastructure and data repository of a PT electronic patient record (ePR) system that can ultimately help to minimize radiation toxicity in prostate cancer patients and improve patient outcomes in future proton therapy treatment.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

An electronic patient record (ePR) system for image-assisted minimally invasive spinal surgery

Asset Metadata

Creator Le, Anh Hong Tu (author)

Core Title Mining an ePR system using a treatment plan navigator for radiation toxicity to evaluate proton therapy treatment protocol for prostate cancer

School Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Biomedical Engineering

Publication Date 07/26/2010

Defense Date 04/28/2010

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag data mining,electronic patient record,knowledge discovery,OAI-PMH Harvest,prostate cancer,proton,radiation toxicity

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Huang, Han K. (committee chair), Liu, Brent J. (committee chair), D'Argenio, David Z. (committee member), Law, Maria Y. Y. (committee member), McNitt-Gray, Jill L. (committee member), Shung, Kirk K. (committee member)

Creator Email anhhle@usc.edu,anhle09@yahoo.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-m3212

Unique identifier UC1313931

Identifier etd-Le-3757 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-354062 (legacy record id),usctheses-m3212 (legacy record id)

Legacy Identifier etd-Le-3757.pdf

Dmrecord 354062

Document Type Dissertation

Rights Le, Anh Hong Tu

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Repository Name Libraries, University of Southern California

Repository Location Los Angeles, California

Repository Email cisadmin@lib.usc.edu