University of Southern California Dissertations and Theses

Characterization of the brain in early childhood

(USC Thesis Other)

Characterization of the brain in early childhood

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

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

May 2018

Dissertation Committee:
Research Assistant Professor Natasha Lepore, Ph.D. (Advisor and Chair)
Professor Michael Khoo, Ph.D. (Co-chair)
Assistant Professor Judy Pa, Ph.D.
Professor Meng Law, M.D.
Research Assistant Professor Anand Joshi, Ph.D.

Copyright 2018 Niharika Gajawelli

CHARACTERIZATION OF THE BRAIN
IN EARLY CHILDHOOD
by
Niharika R. Gajawelli
ii

Abstract
The characterization of healthy brain and cranial anatomical growth during early childhood
is a vital step in our understanding of normal and pathological neurodevelopment. The
brain and cranium develops rapidly in the first 5 years of life, and describing this process
in vivo through medical imaging would allow comparison tools to aid in disease diagnosis.
This thesis focuses more specifically on the anatomical development of three key brain and
head structures that have so far been understudied in early childhood: the central sulcus
(CS), the neurocranium, and the corpus callosum (CC). Below a brief abstract is presented
for these three structures respectively.

Sulcal growth begins at the fetal stage and continues evolving through the first years of life.
The CS, which begins developing at 20 weeks gestational age, has been shown to vary in
depth over the range of the sulcus. The CS is located adjacent to the precentral gyrus which
plays an important role in motor function, yet, despite this significance, normal
development of the CS has not been studied. This thesis explores CS depth changes in early
childhood using MRI.

Along with the brain, the neurocranium also grows rapidly in the first 2 years of life to
accommodate brain growth. Neurocranium growth patterns have rarely been studied except
for global values such as circumference and head volume. Neurocranium thickness, in
particular, is an important feature to investigate especially to predict the predisposition of
the neurocranium to disease or fracture. In this thesis, we present a pipeline to compute
neurocranium thickness using the Laplace-Beltrami operator on volumetric meshes
generated from processing MRI data in early childhood.

One of the most studied structures in adults and aging, the CC is also susceptible to change
in early childhood. Cognitive and behavioral deficits in neurodevelopmental disorders have
been correlated with structural changes in the CC. However, studies on healthy
development of the CC using structural MRI data, especially between the ages of 1 and 5
iii

are scarce. To this end, a combined measure of surface multivariate tensor based
morphometry (mTBM) and medial axial distance (MAD) was used in this thesis to detect
sets of local geometric shape differences between the different subject groups. This
combined metric has demonstrated strong statistical power over voxel based TBM or
individual measures of MAD and mTBM respectively in a number of pediatric and adult
neuroimaging studies.

iv

Acknowledgements
This work was possible due to the support of a large number of people. Although I may
miss a few of you here, be sure that you remain together with me in all my endeavors,
happiness and sadness.
First, I am sincerely grateful to my first advisor, the late Dr. Manbir Singh who gave me
the opportunity to begin my PhD research and join the extraordinary team at the department
of biomedical engineering at USC. He really cared for his students and was there with us
even during the late evening MRI scans. The untimely death of Dr. Singh led to a vacuum
for all of us but his memories will always stay with us. Mrs. Singh and Kabir, thank you
for sharing him and his memories with us and continuing to support us through our PhDs.
We are lucky to have met him and you both. I am also glad that I fulfilled my promise to
Dr. Singh to continue and complete the PhD program.
I am truly grateful to Dr. Natasha Lepore who took over the responsibility of my PhD and
guided me through the research process, and launching me into my projects, leading to my
final dissertation submission. I have learned a great deal from you and have been
empowered as a woman researcher thanks to you. Thank you for being my mentor, friend,
as well as my PhD advisor. You have been an inspiration to me. I have become more
confident through the I-Corps program, overcoming my shyness in talking with people and
this would not have been possible without your encouragement and support.
I would like to thank Dr. Meng Law for his continued support through the duration of my
PhD, and for his encouragement in publishing my first paper. I am also grateful to him for
ensuring me and my colleagues had funding to keep going, without which this PhD would
not be possible, Thank you for providing the chance to shadow you and the other
radiologists at USC and for the discussions to do better research relevant in the clinic.
Thank you, Dr. Judy Pa, for your guidance which helps me continue work on resting state
fMRI work. I appreciate our weekly discussions on fMRI and have been inspired by your
enthusiasm and dedication to the field of neuroimaging and neuroscience. I feel lucky to
v

have been able to work with not just one but two strong women advisors, and feel
empowered as a woman to continue my career in the field of neuroimaging.
I will always be indebted to Dr. Krishna Nayak, who helped my take a first step in this field
by giving me an opportunity to work on research before I even joined the PhD program.
Additionally, I would like to extend my gratitude toward all excellent faculty and
committee members in the BME department, Dr. Michael Khoo, who kindly agreed to be
my committee co-chair, Dr. John Wood, whose comments I find extremely insightful and
Dr. Anand Joshi, whose work I think is extremely innovative. Discussions with all of you
regarding my research have made my final dissertation better. I would also like to thank
Dr. Norberto Gzywacz and Dr. Helena Chui, for your leadership and looking into our well-
being. Thank you Mischal Diasanta and Karen Johnson for all your support through the
years, and making sure I have everything in order to graduate. You both are awesome!
I would also like to thank Dr. Vidya Rajagopalan for her insightful comments during
research discussions and sharing her research ideas! Dr. Mary Nelson for providing clinical
perspective on pediatric data, and Dr. Marvin Nelson for engaging in discussion relating to
results interpretation. Helene Nadeau for helping me get started on some of my projects at
CHLA and for her hospitality in Montreal. Thank you to Julia Castro and Ruth Rizzo at
CHLA for making all the grant submissions smooth and for helping us progress in research.
I would also like to thank Michael Kromnick and Andrea Belz, who advised me through
the I-Corps program and my collaborators, Sean Deoni, Olivier Coulon and Yalin Wang
for their support.
I think the path to PhD. is more of a path to self-learning and discovery. I have met so many
people in this wonderful journey. Some only for fleeting moments, but have left a lasting
impression. I would like to list out the names of all the people who have been there for me
at different points of time, sometimes for fun, sometimes for serious discussions and
sometimes just for comfort. The PhD journey was more enjoyable because of all of you.
vi

I am thankful for the members and close friends of my former lab, most of whom also came
to CHLA. All of us were part of one team and group and remain so till the day as students
of the late Dr. Manbir Singh. I particularly enjoyed our evening MRI scans and late night
dinners. Darryl Hwang, thank you for being a wonderful friend and encouraging me to
complete my dissertation. Sinchai Tsao, thank you for bringing me into the PhD program,
I really appreciate it. Bryce Wilkins, thank you for all the discussions on DTI at CHLA, I
learned a lot from you. Nam Lee, we only got a little more than a year together but I was
inspired by your enthusiasm for research.
To my lab members, former and current, at CHLA, I couldn’t have done this without you
all (in no particular order). Natacha Paquette, thanks for keeping me calm through the
stressful times, I love the insight you bring to our work. Julie Coloigner, I really enjoyed
working with you while you were at CHLA, and your determination for research has
inspired me. I also got more excited about fMRI through your research. Roza Vlasova and
Daniella Sacchetto, for the interesting collaborations, Yi Lao, for being an awesome friend
and lab mate and helping me in the final stages of my PhD, even though I kept bugging
you many a time. Laura Martinez, for the interesting conversations in the lab, Yaqiong
Chai, for the encouragement and for being a fellow PhD candidate with me in this journey.
Jeff Tanedo, thank you for being an awesome friend I can always count on and for the
wonderful discussions. I am so glad to have met you through this process. I would also like
to thank the following people for helping me in some way or another during my time at
USC: Clarissa James, Natalie Ramsy, Carlos Salazar, Andrea Ezis, Siddhant Sawardekar,
Mighten Yip, Peter Michels, and Samantha Ma. All the research assistants at CHLA, while
I may not name you all, I sincerely appreciate the effort you all put into research.
To my current and former members at the Pa lab, thank you Lisette Isenberg, Ashwin
Sarkhare, Zach Hobel, Jocelyn Argueta, Joey Contreras, Aryan Madani, and Chris
Patterson for the interesting discussions. I have been enlightened through discussions with
you and am glad to be part of this lab.
vii

All of this however, could not have been achieved without the support of my family
members. My loving husband Harish, who has been a constant source of support,
throughout all the difficult times, encouraging, teaching, and motivating me all the way. I
will forever be grateful to you. I would like to thank both my parents Nageshwar Rao and
Indira for having the confidence in me and giving me this wonderful opportunity to study
in USC. My in laws, Krishna Sridhara and Usha Sridhara for their love, affection and
encouragement during the process of my research.
I would also like to thank my awesome friends. My inspiration Divya Varadarajan, who is
my partner in crime, sharing the happy as well as difficult experiences with me in the PhD
journey, and Srikanth Nori, who puts up with me hogging Divya’s time. Tintisha Sagar,
my longtime friend, you inspire me, thank you for being there and seeing me through this
process. Chitresh Bhushan, for being a good friend and helping me in my initial stages of
PhD, Srinivas Yerramalli, Gopi Neela, Karthik Shanmugam and Sanjay Purushotham for
interesting evening discussions during dinnertime, and Ruchi Deshpande and Sneha Verma
for sharing the PhD journey with me.
Kumaran Shanmugam, thank you for always boosting my ego, Rakesh Mallela and Rahul
Gundapaneni for always lightening the mood in frustrating times and my dance group
members for being the forever inspiring and strong women that I admire.

viii

Contents
Abstract ............................................................................................................................... ii
Acknowledgements ............................................................................................................ iv
Contents ........................................................................................................................... viii
List of Abbreviations ......................................................................................................... xi
List of Figures ................................................................................................................... xii
Chapter 1. Introduction ....................................................................................................... 1
1.1 Motivation ............................................................................................................ 2
1.2 Dissertation Focus and Main Contributions ......................................................... 3
1.3 Organization of the Dissertation .......................................................................... 5
Chapter 2. Brain and neurocranium development in early childhood ................................ 6
2.1 Brain development ............................................................................................... 6
2.2 Central Sulcus Development ................................................................................ 8
2.3 Homunculus ......................................................................................................... 9
2.4 Neurocranium development ............................................................................... 10
2.5 Corpus callosum development ........................................................................... 13
2.6 Magnetic Resonance Imaging of the brain in early childhood ........................... 13
Chapter 3. Development of the central sulcus in young children .................................... 16
3.1 Introduction ........................................................................................................ 16
3.2 Method ............................................................................................................... 18
Data ........................................................................................................................... 18
Processing ................................................................................................................. 18
3.3 Results ................................................................................................................ 21
ix

3.4 Discussion .......................................................................................................... 26
3.5 Conclusion .......................................................................................................... 30
Chapter 4. Neurocranium thickness mapping in early childhood ..................................... 32
Abstract ......................................................................................................................... 32
4.1 Introduction ........................................................................................................ 33
4.2 Method ............................................................................................................... 35
Data ........................................................................................................................... 35
Pre-Processing........................................................................................................... 36
Thickness Calculation ............................................................................................... 38
Surface registration ................................................................................................... 40
4.3 Results ................................................................................................................ 40
4.4 Discussion .......................................................................................................... 44
4.5 Conclusion .......................................................................................................... 47
Chapter 5. Corpus callosum development in children ...................................................... 48
5.1 Introduction ........................................................................................................ 48
5.2 Method ............................................................................................................... 51
Data ........................................................................................................................... 51
Processing pipeline ................................................................................................... 52
Surface registration and multivariate tensor based morphometry (mTBM) ............. 53
Thickness computation ............................................................................................. 54
Group statistics.......................................................................................................... 54
5.3 Results ................................................................................................................ 55
5.4 Discussion .......................................................................................................... 57
x

5.5 Conclusion .......................................................................................................... 59
Chapter 6. Conclusions and Future Work ......................................................................... 60
6.1 Central sulcus ..................................................................................................... 60
6.2 Neurocranium ..................................................................................................... 60
6.3 Corpus callosum ................................................................................................. 60
6.4 Publications ........................................................................................................ 61
Journal publications .................................................................................................. 61
To be submitted......................................................................................................... 61
Conference papers ..................................................................................................... 62
Abstracts ................................................................................................................... 63
References ......................................................................................................................... 64

xi

List of Abbreviations
ADHD Attention Deficit Hyperactivity Disorder
BIC Bayesian Information Criterion
BMP Bone Morphogenetic Protein
CC Corpus Callosum
CDC Center for Disease Control
CS Central Sulcus
DTI Diffusion Tensor Imaging
GM Gray Matter
MAD Medial Axis Distance
MADMTBM Combined measure of MAD and mTBM
MRI Magnetic Resonance Imaging
MTBM Multivariate Tensor Based Morphometry
MWF Myelin Water Fraction
PPFM 'Pli de Passage Fronto-parietal Moyen'
WM White Matter

xii

List of Figures
Figure 1. Motor Homunculus (adapted from Penfield, W. & Rasmussen, T., The cerebral
cortex of man: a clinical study of localization of function)
50
............................................ 10
Figure 2. Skull of newborn shown with cranial bones and fontanelles. Adapted from Gray's
Anatomy (public domain)
52
............................................................................................. 12
Figure 3. Adult skull with closed sutures shown. Adapted from Gray's Anatomy (public
domain)
52
.......................................................................................................................... 12
Figure 4. Cross-sectional representation of change in brain MRI contrast of GM, WM,
CSF contrast over time...................................................................................................... 14
Figure 5. Cortical sulci recognition and spatial normalization. Adapted from
10,45,85
....... 19
Figure 6. The sulcal depth profiles of the left and right CS for 1, 2, 3 and 4+ year old groups.
Top and Bottom: Left CS and right CS depth curves respectively. .................................. 22
Figure 7. Normalized mean depth curves for all subjects (red) and normalized logarithmic
regression coefficients (blue) for the left CS (top) and right CS (bottom). The higher values
indicate greater change. The blue diamonds indicate significance after Bonferroni
correction. ......................................................................................................................... 23
Figure 8. Logarithmic curves fitted to position 71 for the left CS (top) and position 32 for
the right CS (bottom). These positions correspond to the areas of greatest change in the
regression .......................................................................................................................... 24
Figure 9. Mann-Whitney test results comparing two groups with 6 month intervals for the
left CS (top) and right CS (bottom). The red dots indicate regions with significant
difference. ......................................................................................................................... 25
Figure 10. Example of skull segmentation shown on an 18 month old brain ................... 34
xiii

Figure 11. Pre-processing pipeline ................................................................................... 37
Figure 12. Distribution of subjects in cohort .................................................................... 38
Figure 13. (a) Example of neurocranium extracted using FSL. The outer region is the
superior patch and the inner table is defined as the inferior patch. (b) Tetrahedral mesh
created from the extracted neurocranium using the Iso2mesh toolbox. (c) Zoomed in figure
of the tetrahedral mesh enclosed in the square. Streamlines are later generated from the
outer to the inner patches to compute thickness. .............................................................. 39
Figure 14 Neurocranium thickness in 3 age groups. (a) 12 months, (b) 24 months, and (c)
36 months. Color bar indicates thickness values .............................................................. 41
Figure 15. Group differences between each consecutive group. Results were corrected
using permutation testing using 10000 permutations and a significance threshold of 0.05.
The color bar indicates the p-value. .................................................................................. 42
Figure 16. Results of significant difference (p=0.05) after non-parametric t-tests and
multiple comparison correction using permutation testing (10000 permutations) comparing
12 months and 36 month groups. Biggest difference is seen around the lambdoid suture.
The color bar indicates the p value. .................................................................................. 43
Figure 17. Results of regression showing change over time from 6 months to 24 months
after FDR correction to significance level of p=0.05. Red regions indicate the coefficients
showing biggest change over time. Regression coefficients shown in (a1), and
corresponding p-values are shown in (a2). Color bar indicates [max (p-value) – p-value]
for visualization. Neurocranium thickness change over time for the maximum intensity
vertex in the black dotted circle for children between ages of 6 months and 24 months
shown in (b). ..................................................................................................................... 44
Figure 18. Corpus Callosum outlined in blue is easily visible on a midsagittal slice of the
brain .................................................................................................................................. 49
xiv

Figure 19. CC segmentations for different age groups .................................................... 52
Figure 20. Example of a tetrahedral mesh generated on the segmentation of a CC ......... 53
Figure 21. Results showing a. the determinant of J (left top), b. MTBM (left bottom), c.
MAD (right top), and d. MADMTBM (right bottom) between the 12m vs 24m groups . 56
Figure 22. Results showing a. the determinant of J (left top), b. MTBM (left bottom), c.
MAD (right top), and d. MADMTBM (right bottom) between the 24m vs 36m groups . 56
Figure 23. Results showing a. the determinant of J (left top), b. MTBM (left bottom), c.
MAD (right top), and d. MADMTBM (right bottom) between the 36m vs 48m+ groups 57
Figure 24. Linear regression results showing regions of greatest change in all subject data
........................................................................................................................................... 57

1

Chapter 1. Introduction
The brain expands rapidly in early childhood, reaching 95% of its final volume by age 6.
During this time, it is particularly vulnerable to environmental influences, and this is also
the time when many genetically mediated developmental disorders begin to influence brain
growth. Abnormal neurodevelopment affects a growing number of children every year in
the United States, but often is first detected only when delays in developmental milestones
or dysfunction start to manifest themselves
1
. For example, more than 13% of children in
the U.S. are estimated to suffer from conditions such as attention deficit hyperactivity
disorder (ADHD) or autism
2
, but the former is only typically detected at school ages, and
the latter at 2+ years old. Characterization of early healthy neuroanatomical development
is essential to detect or treat neurodevelopmental pathologies at an earlier stage, in order to
minimize their effects.
As a safe and non-invasive modality, in-vivo imaging with magnetic resonance imaging
(MRI) is ideally suited to characterize brain structure and growth. In pediatric populations,
the challenge of understanding in-vivo neuroanatomy is confounded by the lack of normal
control data and analysis tools. As a result, compared to later stages of childhood, there are
few studies that investigate brain development during the earliest and fastest changing
years (0 to 5 years old).
Increased efforts are now being made on pediatric data acquisition through the creation of
large public and private databases of healthy pediatric brain MRI, such as the NIH MRI
Study of Normal Brain Development
3
, the Pediatric Imaging, Neurocognition and
Genetics Study (http://pingstudy.ucsd.edu), the University of North Carolina at Chapel
Hill
4
, and the Advanced Baby Imaging Lab database (www.babyimaginglab.com).
However, even once a dataset is available, there are several important challenges in
characterizing brain development in children. First, in early infancy, MRI T1-weighted
image intensities are the reverse of adult brains. This is followed by a period of poor
differentiation between gray and white matter, until at around 12 months of age, MRI
intensity patterns finally begin to resemble those of adults. Both the contrast and ratio
2

between white and gray matter change rapidly, making it difficult to compare MRI images
over a spectrum of different ages during early stages of life
5
. These evolving signal
intensities arise primarily from changes in relaxation time caused by the reduction in water
content in both gray and white matter over time, and from increases in myelination with
age
6,7
. Additionally, pediatric MRI data are typically collected with lower resolution to
compensate for movement, and even when higher resolution images available, many details
are not visible due to the smaller brain size. Such differences between children and adult
brain images make it harder to use methods developed for adults’ MRI processing and
analysis. Hence there is a significant need to adapt or develop methods specific to pediatric
data.
1.1 Motivation
The motivation of this work stems from the need to understand brain growth and changes
at a more fundamental level in order characterize normal healthy development. Normal
development of the brain is particularly important to classify abnormal development.
Specifically in early childhood, rapid changes in the brain motivate us to not only
investigate gross measures such as overall brain or subcortical structure volumes but also
to scrutinize localized brain changes, such as individual sulcal, gyral or subcortical
structure growth, to detect the more subtle components of growth. This can lead to earlier
disease detection and earlier intervention before substantial damage happens in the brain.
This thesis focuses on three key structures of the head and brain: the central sulcus,
neurocranium, and corpus callosum.
The central sulcus (CS) dives the frontal and parietal lobes
8
and it is located adjacent to the
primary motor and the somatosensory cortices. The CS is useful as a major landmark of
the brain since it develops at around 20-23 weeks gestation age
8,9
. The CS plays an
important role in motor function development and is associated with learning and
plasticity
10,11
. CS length and depth have also been implicated in developmental disorders
such as Williams syndrome
12
and ADHD
11
, respectively. Investigation of CS shape
development will provide insight into how this evolution tracks age specific traits.
3

The neurocranium and brain develop simultaneously and influence each other in early life
13
.
It has been shown that the same gene family influences development of both the brain and
the neurocranium
13
. Neurocranial malformations are often a symptom of
neurodevelopmental disorders, including genetic conditions. Similarly, neurocranial
growth restrictions impact the development of the brain. As a result, infants with cranial
abnormalities often suffer from cognitive impairment, elevated intracranial pressure and
motor disabilities
14,15
. Additionally, neurocranium thickness also tells us how malleable
the skull is with respect to fracture.
The corpus callosum (CC) is the primary pathway of interhemispheric information
transfer
16
and involved in interconnecting the major subdivisions of the cerebral cortex
from high-level associative areas. Perhaps with reason to its central location, the CC is
implicated in various diseases including autism
17
, Alzheimer’s disease (AD)
18
, and
traumatic brain injury (TBI)
19
. Prior studies of the CC in children above age 4 have revealed
trends for more pronounced growth in posterior versus anterior callosal sections
20,21
,
however, strong anterior growth between age 3 and 6 has been shown in one study
22
.
Little is known about the age related changes or growth trajectories of these structures, due
to the lack of normative in-vivo data and quantitative imaging tools to analyze it. These
major obstacles need to be addressed for the early detection and optimal treatment of
conditions of infancy and early childhood.
1.2 Dissertation Focus and Main Contributions
This dissertation focuses on pediatric brain and neurocranium development using measures
derived from shape analysis. As discussed above, efforts toward characterizing the normal
developing brain and neurocranium are still in the early stages and quantitative features
acquired through shape analysis will complement current literature.
One of the largest pediatric brain MRI databases in the world was created at the Advanced
Baby Imaging Lab (http://www.babyimaginglab.com) by our collaborator, Dr. Sean Deoni.
4

It now comprises more than 500 cross-sectional and longitudinal brain scans in children
aged 0 -5, which we will be using for this dissertation. The data used here consists of T1-
weighted anatomical brain scans of children between the ages of 6 and 60 months. Different
subsets of the data were used for the different projects. The details are listed in the
corresponding chapters.
a. Central sulcus: The central sulcus is involved in motor function and is associated
to plasticity changes in the brain. CS development between ages of 12 months to
60 months was investigated here. We used the BrainVisa software package
available online to extract the sulci and compare development in a large pool of
subjects. We found that the shape of the CS depth resembles the shape of the adult
CS even at 12 months of age, but features such as the PPFM become more
prominent as the child grows. Additionally, we found the greatest changes in the
brain depth happen in regions that are more heritable.

b. Neurocranium: Determination of neurocranium thicknes with respect to age may
yield important information on brain and head vulnerability to trauma.
Neurocranium thickness changes between the ages of 6 months and 36 months was
investigated as part of this study. We developed a pipeline to analyze the
neurocranium for the first time using FSL skull extraction on MRI data, tetrahedral
mesh generation and computing neurocranium thickness calculation by solving the
Laplace-Beltrami equation to investigate differences at a regional scale. We found
that the neurocranium thickness changes the most in the posterior part of the
neurocranium, consistent with modeling from computed tomography studies
23
.

c. Corpus Callosum: Little is known about the morphological development of the CC
in early childhood. Hence investigating shape (surface + thickness) changes of the
corpus callosum with respect to age is a key first step to characterizing CC
development. Here, surface based changes of the CC through ages 12 months to 60
months were investigated. A combined measure of medial axial distance and
surface multivariate tensor based morphomety was used to investigate local
5

changes across of the CC. We found that the body of the CC undergoes the greatest
changes between the ages of 12 and 36 months, and that the genu and splenium
undergo their most rapid growth between the ages of 24 and 48 months.
1.3 Organization of the Dissertation
Chapter 2 presents a high level overview of the brain development in early childhood. We
specifically provide background regarding the central sulcus, neurocranium and corpus
callosum, pertaining to this dissertation.
Chapter 3 presents the development of the central sulcus in early childhood. We illustrate
age related differences and discuss the relationship between the region of the central sulcus
and function.
Chapter 4 presents the development of the neurocranium. We evaluate neurocranial
thickness changes in children between 6 months to 36 months of age using a new surface
morphometry pipeline.
Chapter 5 investigates the change in the corpus callosum with respect to age using surface
multivariate tensor based morphometry.
Chapter 6 is an appendix of additional studies done.

6

Chapter 2. Brain and neurocranium development in early
childhood
The cranium and the brain undergoes a period of rapid growth in early childhood. The
cranium expands from about 25% of its adult size at birth, to 65% and about 90% of its full
size by the first year of life
24
and by 4-5 years of age, respectively
25,26
, and the brain to
about 90% of its final brain volume by age 6. This chapter deals with the biology of the
neurocranium and the brain, and acts as a reference for the physical interpretation of the
results presented in later chapters.
2.1 Brain development
The diverse functions of the brain depend on a complex series of growth stages beginning
in the womb. Within the first 10 weeks of gestation, during the embryonic stage of
development, undifferentiated neural progenitor cells first from the ectoderm layer of the
embryo form the neural plate, which serves as a foundation for the nervous system. At
around 3 weeks gestation
27
, the neural plate begins to close and forms the neural tube,
which results in the spinal cord as well as the brain
28
. This cell differentiation produces
immature neurons, which then migrate to their positions in the brain, extending axons
toward their target cells, thus creating synaptic connections between the growing axon and
the target cells
28
. One of the factors influencing cell differentiation in the neural plate are
inducing factors or signaling molecules provided by other cells. For example, a suppressive
signal of the bone morphogenetic protein (BMP) promotes the ectoderm to differentiate
into the epidermis. The inhibition of BMP signaling appears to involve the expression of
the Sox gene family
28
. Similarly, gene families such as the notch, hedgehog, Wnt, TGF B
and FGF are shown to be involved in differentiation of the brain and the neurocranium
(discussed further in Chapter 4).
These synapses are responsible for transmission of information through electrical and
chemical signals, facilitating different patterns of connectivity in the brain. While most of
7

the neurons are formed before birth, synapses keep developing during infancy, as brain
development continues postnatally
28
.
GM development in neonates was found to be fastest immediately after birth at a whole
brain growth rate of 1% per day, slowing to 0.4% per day at the end of 3 months, through
longitudinal MRI analysis
29
. Postnatally, numerous factors including synaptogenesis and
proliferation, dendritic and axonal growth along with increased neurophil arborization, and
neuronal migration from the ventricular (or proliferative) zone, among others contribute to
the region volume increase in the human brain in the first 3 months of life
29
. These
processes occur at the same time as synaptic pruning and apoptosis
,
processes that are likely
to reduce volume
30–32
. Such competing processes likely contribute to the inverted U-
shaped developmental course
33,34
of the GM volume.
It is also known that cortical growth rate is region dependent. For example, in early
childhood, cortical thickness increased by 16.2% from 3 to 6 months and less than 0.1 %
from 9 to 12 months in the first 18 months of life
35
. It was also shown that cortical surface
area has been shown to increase more in the visual, auditory, sensory cortices compared to
the motor and association cortices in the 1
st
year of life
36
.
Through the lifespan, brain volumes peak at different times in different lobes, shown by
the inverted U-shaped developmental trajectory of the GM
31,34
. It has been shown that in
girls, the peak value of frontal lobe gray matter is reached at 11.0 years in girls and the
peak value of temporal lobe cortical gray matter is reached at 16.7 years, while in boys the
ages are 12.1 years and 16.2 respectively
37
.
Underlying the cerebral cortex is the white matter (WM), consisting of myelinated neuronal
axons that relay the signals to and from other neurons. Myelin is a sheath that surrounds
the axon and acts as insulation to speed up transmission of information along the axon.
Myelin consists of 70% lipid and 30% protein, with a high concentration of cholesterol and
phospholipid
28
. The human brain has a low concentration of myelin at birth, and myelin
water fraction (MWF) studies have shown that myelination trajectories are sigmoidal
38
.
While brain region dependent, most myelin trajectories start with a lagged growth period
8

between 90-150 days, then exponentially grow until about 400 days of age, after which
they become logarithmic
38
. Similar trajectories have also been shown in DTI
7
.
Other structures of the brain include subcortical gray and white matter structures such as
the corpus callosum, hippocampus, amygdala, thalamus, and putamen. Subcortical volume,
early childhood, the hippocampus and caudate volumes have been shown to increase by
13% and 19%, respectively between the age of 1 and 2
39
. Additionally, it has been shown
that similar to the cortex, the caudate nucleus also follows an inverted U-shape
developmental trajectory, peaking at age 10.5 in females and age 14 in males
40
.
2.2 Central Sulcus Development
Cortical folding in the brain is an important process for optimizing brain wiring and
functional organization
41
. Multiple hypothesis exist on how cortical folds are formed. The
most widely cited hypothesis regarding cortical folding include the differential tangential
expansion and the axonal tension hypothesis. The former states that cortical folding
happens as a result of difference in the growth rate between upper and lower cortical layers
and this generates stress sufficient enough to induce cortical surface buckling
42,43
. The
latter states that cortical folds are caused by tension on the surface of the brain from the
pulling forces of populations of axons that are connected together
44
. The crests of these
folds are called gyri and the grooves are called sulci.
The CS, which is the most constant landmark on the brain surface, emerges between 20 to
22 weeks gestational age and separates the motor and somatosensory cortices. The CS
originates above the sylvian fissure and traverses the brain surface to the interhemispheric
fissure in healthy adults. While individual differences do exist, the CS has demonstrated
similar patterns or features amongst individuals
45
. The CS appears as a sinusoidal shape
with three dominant curves. The middle bend of these curves was named by Broca as the
pli de passage moyen or PPFM (also known as the precentral knob), a representation of the
hand motor and sensory area in the pre- and postcentral gyri
46–48
. While cortical regions
representing the whole-hand motor and sensory function were identified in the primary
sensorimotor cortex, it was also shown to be localized to the PPFM, in addition to the finger
9

and thumb sensory and motor regions
48
. This hand motor function in addition to tongue
and lower face sensory areas on the CS are all reliable landmarks for identifying function
46
.
They have also routinely been identified in neurosurgery and correlated in with image
guidance results
46
.
2.3 Homunculus
Motor and sensory functions for different parts of the body were first identified on the CS
by Penfield and Boldrey (1937)
49
. A neurological map, or homunculus, was developed
through cortical stimulations of 126 patients undergoing surgery. Cortical tissue in primary
motor and somatosensory cortices are organized according to specific function, however,
not necessarily proportional to the physical mass in our bodies. For example, the motor
homunculus shows that a large part of the cortex corresponds to the hand and fingers,
compared to the legs. Regions such as hands, fingers and lips have more complex
functionality, and more motor and sensory connections and are therefore represented over
a larger area. From top to bottom, the homunculus represents the genitals and toes, leg
areas, the upper torso, hand and finger function, jaw and tongue areas. The motor
homunculus describes brain areas dedicated to motor processing located in the precentral
gyrus, whereas the sensory homunculus describes brain areas of the post central gryus
dedicated to sensory information processing. The homunculus is presented here as an
important resource for interpretation of results in Chapter 3, where we focus on the central
sulcus, located in-between the pre and postcentral gyri.
10

Figure 1. Motor Homunculus (adapted from Penfield, W. & Rasmussen, T., The cerebral cortex of
man: a clinical study of localization of function)
50

2.4 Neurocranium development
Neurocranial bones grow by a combination of sutural growth, bone remodeling and
displacement by the expanding brain
14
. In utero, prenatal brain development shows a sharp
rise around the 20th week of gestation
51
. The brain and cranium grow rapidly through the
2
nd
half of the fetal period and throughout infancy.
The neurocranium at birth consists of thin bony plates of ossified bone tissue, which are
connected by the soft connective tissues of the sutures and fontanelles as shown in Figure
2. This flexible construction permits the neurocranium to deform and expand to
accommodate the rapidly growing brain in early childhood. The most striking difference
between the adult neurocranium shown in Figure 3 is the incomplete ossification of the
bones in the newborn neurocranium. The bones progressively ossify as the child grows,
11

leading to the closure of the fontanelles by age 1. The frontal or metopic suture closes first,
between 3 to 9 months of age and the coronal, sagittal and lambdoid sutures close between
22 – 39 months of age.
53

There exist common underlying genetic processes involved in development of the cortex
and the neurocranium. Some studies
29,39,54
that have looked separately at brain volume and
cranial size, have shown similar growth trajectories during infancy and early childhood and
indicated correlation of these measurements. From the point of view of genetics, several
gene families have been identified as simultaneously involved in both cranial and brain
development, as summarized in
13
, including the Notch, Hedgehog, Wnt, TGFb and FGF
families. For example, in the brain, the Notch pathway is involved in neural stem cell
proliferation, differentiation and apoptosis, and neural crest cell induction, while on the
neurocranial side, it is implicated in cell differentiation
55–57
. Additionally, tensile strain on
the neurocranium from brain growth is thought to force the neurocranium to reshape and
adapt to the new brain shape. This occurs because these forces deform the membrane or
cytoskeleton of the cells, which in turn affect cell processes, such as signaling and
proliferation
58,59
.
12

Figure 2. Skull of newborn shown with cranial bones and fontanelles. Adapted from Gray's Anatomy
(public domain)
52

Figure 3. Adult skull with closed sutures shown. Adapted from Gray's Anatomy (public domain)
52

13

2.5 Corpus callosum development
The corpus callosum (CC) is the largest WM fiber bundle connecting the two hemispheres
of the brain. The CC contains axons of varying diameters through its structure. DTI studies
of the brain have shown that CC fibers connect to specific brain regions (e.g. fibers from
the splenium project to the occipital lobe). The WM contrast of the CC is due to myelin.
WM myelination rates vary between structures in the brain. For example, the body of the
CC begins myelination prenatally with some microscopic myelination at term
9
. Postnatally,
myelination in the CC starts to be visible at around 4 months of age, beginning in the
splenium as interhemispheric connections of the visual and visual association areas
myelinate allowing binocular vision and visual accommodation
9,21,60
. Myelin content in all
parts of the CC increase rapidly until about 18 months of age and gradually thereafter until
5 years, following a Gompertz distribution
38
. In the limited number of studies that exist in
the age range between 1 and 5, it was shown that the CC develops in a non-linear fashion
and the CC area increases, regardless of gender, following a similar trajectory as the cortex,
especially in the first few years of life
61
.
2.6 Magnetic Resonance Imaging of the brain in early childhood
Gray and white matter structures, along with the neurocranium can be visualized using T1-
weighted structural magnetic resonance imaging (MRI) scans. A T1-weighted MRI, or
structural MRI scan, measures the rate at which the water molecules or protons in the tissue,
excited by the RF pulse, relaxes back to equilibrium. The tissue contrast is characterized
by the T1 time constant, which differs in value for GM, WM, fat, CSF and bone, hence
producing a contrast we can visualize in MRI scans.
However, this T1 time constant values differ between young children and adults. A few
previous studies that measured T1-relaxation times in children report a prolonged T1-
relaxation time in infants, which correspond to the high water content in the newborn
brain
6,7
. This high water content causes poor differentiation between gray and white matter
on the MRI scan. The T1-relaxation times decrease sharply with age, especially in the first
year of life, corresponding to the decline in water content and increase in myelination.
14

While all the major WM tracts were shown to be well defined by age 1, a contrast similar
to that of adults was not seen until early adolescence
16,18
. Figure 4 shows the change in
contrast over time from 3 months to 36 months in cross-sectional data at about the same
axial slice position. The lack of contrast between WM and GM in the 3 and 6 months age
range can clearly be seen here.

Figure 4. Cross-sectional representation of change in brain MRI contrast of GM, WM, CSF contrast over
time
Through cortical volume based studies conducted in the past
31,33,39,64,65
, it has been shown
that cortical volumes follow an inverted U-shaped developmental trajectory. While the
rates of growth differ between brain regions, one study investigating brain growth between
ages 1 and 5 has shown a predominantly logarithmic trajectory of deveopment
66
. In another
study on development, between the ages of 3 and 30 years, analysis of 1274 MRI scans
from 674 healthy, showed differences in cortical volume, surface area, cortical thickness,
gyrification, and convex hull area between genders over time
65
. It was shown that surface
area changes rather than cortical thickness changes contribute to the change in cortical
volume between males and females
65
.
15

While brain development is a complex process, with many aspects that cannot be explain
in a mere chapter, some important aspects of brain development using structural MRI scans,
relevant to this dissertation were presented here.
16

Chapter 3. Development of the central sulcus in young
children
Niharika Gajawelli
a,b
, Sean Deoni
c,d
, Natalie Ramsy
b
, Holly Dirks, Douglas Dean, Jonathan
d
,
O’Muircheartaigh
d
, Marvin D. Nelson
e,f
, Natasha Lepore
a,b,f
*, and Olivier Coulon
g
*
[* equal senior author contribution]
a
CIBORG Lab, Department of Radiology, Children's Hospital Los Angeles, CA, USA

b
Department of Biomedical Engineering, University of Southern California, CA, USA

c
Department of Pediatric Radiology Research, Children's Hospital Colorado, CO, USA

d
Department of Biomedical Engineering, Brown University, RI, USA

e
Department of Radiology, University of Southern California, CA, USA

f
Department of Radiology, Children's Hospital Los Angeles, CA, USA
g
CNRS Research Director, Aix-Marseille University, Marseille, France

Abstract— The human brain grows rapidly in early childhood, reaching 95% of the final
volume by age 6. Understanding brain growth in childhood is important both to answer
neuroscience questions about anatomical changes in development, and as a comparison
metric for neurological disorders. Maps describing healthy neuroanatomical development
are an instrumental tool in early diagnosis, monitoring and intervention for neurological
diseases. Although some lifespan and pediatric studies have explored changes in cortical
volumes and surface areas, research investigating the individual sulci are still in its early
stages. In this paper, we examine the development of the central sulcus in children aged 12
to 60 months by delineating this structure from structural magnetic resonance images. The
central sulcus is one of the earliest sulci to develop at the fetal stage and is implicated in
diseases such as Attention Deficit Hyperactive Disorder and Williams syndrome. We show
developmental trajectories of the central sulcus with respect to age and discuss the regions
that indicate the greatest changes as well as the nature of these changes. We observe that
while the central sulcus had a similar shape as that of an adult even at 12 months, its depth
increases with age, mirroring that of surface area changes in the precentral cortex.
3.1 Introduction
Normal development of the brain is of key interest in both neuroscience and medicine. In
particular, cortical folding, which begins at about 20 weeks gestational age
9
, keeps
17

evolving through childhood
67
. Hence, understanding sulcal and gyral growth can provide
valuable insight into brain development
35
. Cortical parameters such as sulcal depth can also
be a marker of functional specificities or developmental pathologies
45,68
. For example,
subtle abnormalities can be indicative of diseases
69,70,71,12
. For example, the central sulcus
(CS) was found to be shorter in length compared to normal controls, in Williams syndrome,
a rare genetic disorder
12,71
. Additionally, in children with Attention Deficit Hyperactive
Disorder (ADHD), the average and maximum depths of the left CS were found to be
significantly larger than in healthy children and hyperactivity-impulsivity scores were
shown to positively correlate with between-group differences in the mid-section of the CSs
bilaterally
11
. Therefore, in order to gain insight into why and how the CS morphometry
differs between populations, it is important to establish a baseline of normal cortical folds
maturation.
The CS is a prominent sulci that begins development in the fetal stage, dividing the parietal
and frontal lobes. Due to its location between the motor and somatosensory cortices, the
CS is highly involved in motor function. For example, it was shown that piano players had
greater local variability in the somatotopic hand area of the right CS compared to controls,
suggestive of plastic changes in the morphology of the CS due to long term motor skill
training
72
. The motor and somatosensory homunculli
50
, can be particularly useful in
determining the functional changes associated with a specific region of the CS.
The CS depth is an interesting feature to study due to its variation through the length of the
CS, and in particular, several studies have associated it with handednesss
45,73,74
. The CS
contains a distinct fold that connects the pre and postcentral gyri, termed by Broca as the
pli de passage moyen or PPFM
46,47
. Previous studies have shown that hand motor function
is localized to the PPFM
46
. The CS regions, closer to the sylvian fissure correspond to the
tongue and lower face sensory function. High heritability was shown in regions of the CS
depth curve corresponding to the hand and mouth areas, which lay on both sides of the
PPFM
75
. However, the PPFM itself was not shown to be highly heritable
75
. Therefore, this
area can be a key feature to examine in relation to functional motor scores as it may more
likely be associated with plasticity, driven by environmental factors. Due to the rapid
18

cerebral change and environmental influences that contribute to the learning of new skills
in young children, the CS is a promising biological trait to investigate in development.
In this paper, we focus specifically on the changes that occur in the CS depth in early
childhood between the ages of 1 and 5. Through the analysis of CS depth, a developmental
trajectory is created. While previous studies have used gross measures on this sulcus, none
to date have zoomed in on local development along the length of the CS. This type of
analysis may be helpful in understanding the specificity of changes that occur and may
have the potential to be used as an anatomical biomarker for disease
12,71

3.2 Method
Data
We selected brain volumes from 130 subjects ranging in age from 270 to 1900 days
(roughly corresponding to 12 months to 60 months) from the Advanced Baby Imaging Lab
database (www.babyimaginglab.com). The data used consisted of high resolution T1 MP-
RAGE MRI scans (1.4-1.8mm
3
) of healthy normal children with the following inclusion
criteria: singleton birth between 37 and 42 weeks gestation with no abnormalities on fetal
ultrasound and no reported history of neurological events or disorders in the infant. Data
acquisition details can be found in
76,77
. Each subject or their guardian was informed of the
goals of the study and signed a formal consent. The study was approved by the Institutional
Review Board of Brown University and informed consent was obtained from the guardians
of all participants. All data was de-identified before pre-processing. Note that the
handedness of the subjects are not assessed in this study due to lack of handedness
information at early ages.
Processing
Brain volumes were first skull-stripped using the BrainSuite
78
software and linearly
registered with 6 degrees of freedom to an age matched template using FSL FLIRT
79,80
. A
geodesic intensity correction algorithm
81
was then applied for correcting the bias field of
the brains and resampled to a 1x1x1mm
3
space. The geodesic bias field correction was
used to improve the contrast between gray and white matter, which in turn led to a more
19

accurate automatic gray/white tissue segmentation using BrainVisa morphologist
pipeline
82
. The histogram of the gray and white matter was manually tuned to achieve an
accurate brain mask and hemispheric split. Once the tissue segmentation and surface
modeling were complete, the BrainVisa pipeline produced graphs containing cortical sulci
meshes
83
. Using these graphs, we manually chose the CS. We then applied the sulcal
parameterization pipeline in BrainVisa to the left and right central sulci of each subject
84
,
which resulted in a depth measure at each position along the sulcus. The sulcal
parameterization process detects the dorsal and ventral extremities of the CS and computes
a smooth isometric parameterization for the sulcus. This gives us a relative position
between the two extremities for each point along the sulcus. The depth at each position
between [0, 100] is computed by measuring the length of the corresponding isoparametric
lines as shown in Figure 5.

Figure 5. Cortical sulci recognition and spatial normalization. Adapted from
10,45,85

In previous morphological statistical studies, the resulting sulcal depth profiles have shown
good inter-subject reproducibility and therefore can be used for group studies. For
calculation of the mean sulcal depth, and group comparison of sulcal depth, the data was
20

grouped into 6 month intervals for age groups from 12 to 24 months and 12 month intervals
from 24 months to 48 months. This type of interval was used to retain statistical power as
well as capture the changes in the younger age groups. Additionally, as the number of
subjects were much smaller (6 and 7 subjects at 48 and 60 months, respectively), and since
we expect the growth to slow down compared to the younger ages, according to cortical
surface area and volume changes studied in the past
66
, subjects corresponding to ages 48
and 60 months were grouped together. The ages and number of subjects are shown in Table
1 below. One subject who fell into the 42 month age range was only included in the
regression analysis.
Age Male Female Total
12 month 18 18 36
18 month 16 9 25
24 month 13 13 26
36 month 17 10 27
48 + 60 months 11 5 16
Table 1. Table showing the number of subjects in the different age groups
The mean and standard deviation in sulcal depth for each group (12, 24, 36 and 48+ month
groups) was then computed. The depth curves were smoothed with a moving average span
of 3. The local minima, which represent the precentral knob or PPFM
45
, of the sulcal depth
curves for each subject were detected and noted. This local minima is typically located in
the inferior two thirds of the CS depth curve. Additionally, in a number of cases, such as
when the distribution was not bimodal (see Discussion), there were multiple local minima.
Therefore, we also inspected these points visually to confirm subjects that didn’t have a
bimodal distribution. Because the CS depth is our feature of interest and has been related
to functional localization
46,86
and functional specificities such as handedness
10,45
, we aim
at characterizing depth developmental trajectories at every position of the CS. To determine
age-based trajectories, linear regression was done using MATLAB to investigate the CS
changes at each of the 100 positions within the entire age group. The following equations
showing the linear, logarithmic and quadratic models that were tested:
21

𝐿𝑖𝑛𝑒𝑎𝑟 : 𝑠𝑢𝑙𝑐𝑎𝑙 𝑑𝑒𝑝𝑡 ℎ ~ 𝐵 1 ∗ 𝑎𝑔𝑒 + 𝐵 2
𝑄𝑢𝑎𝑟𝑎𝑡𝑖𝑐 : 𝑠𝑢𝑙𝑐𝑎𝑙 𝑑𝑒𝑝𝑡 ℎ ~ 𝐵 1 ∗ 𝑎𝑔𝑒 2
+ 𝐵 2 ∗ 𝑎𝑔𝑒 + 𝐵 3
𝐿𝑜𝑔𝑎𝑟𝑖𝑡 ℎ𝑚𝑖𝑐 : 𝑠𝑢𝑙𝑐𝑎𝑙 𝑑𝑒𝑝𝑡 ℎ ~ 𝐵 1 ∗ ln(𝑎𝑔𝑒 ) + 𝐵 2

The Bayesian Information Criterion (BIC) was computed for each model. The BIC values
indicated the best fit as the logarithmic model for most of the 100 positions (87 points in
the left hemisphere and 64 points in the right hemisphere), and therefore was used in for
linear regression analysis in this paper. Finally, the p-values were corrected using
Bonferroni correction.
To compare the differences in CS depth between the 12 month and 24 month, 24 month
and 36 month, 36 month and 48+ month groups Mann Whitney tests were conducted at
each of the 100 positions, and then corrected for multiple comparisons via permutation
testing using all the sulcal depth points along the CS. The Mann Whitney test is used to
avoid assumption of a Gaussian distribution of the sulcal depth. Those group comparisons
aim at identifying the positions at which there is a significant depth change from one group
to another.
3.3 Results
The mean sulcal depth curves of the left and right CS in age groups 12, 24. 36, and 48+
month old groups are shown in Figure 6. In all groups shown and in both hemispheres,
there is a steep increase in the depth profiles from the superior part of the CS (position 1)
to around position 32, followed by a small dip around position 50 with a gradual increase
in the depth profile until the next peak at around position 65, after which the curves decline
steeply. These three features have been previously identified respectively as the superior
peak, pli de passage fronto-parietal moyen (PPFM), and inferior peak
45
. The dip
corresponding to the PPFM is most prominent in the older children.
Figure 7 shows the correlation coefficients after running regression with the logarithmic fit
(blue) using the entire age cohort (270 – 1900 days), along with mean depth curve (red)
over the entire age range. The results were normalized for visualization purposes. The blue
diamonds show the most significant regions after Bonferroni correction. The point with the
22

greatest change, at position 71 and was the most significant for the left hemisphere and
position 32 was most significant for the right hemisphere. These two points that showed
most significance were plotted against age and fitted with a logarithmic fit curve as shown
in Figure 8. The curves show greater increase in the earlier ages but the slope tapers off
and becomes more constant towards older age groups.

Figure 6. The sulcal depth profiles of the left and right CS for 1, 2, 3 and 4+ year old groups. Top and
Bottom: Left CS and right CS depth curves respectively.
23

Figure 7. Normalized mean depth curves for all subjects (red) and normalized logarithmic regression
coefficients (blue) for the left CS (top) and right CS (bottom). The higher values indicate greater change.
The blue diamonds indicate significance after Bonferroni correction.

24

Figure 8. Logarithmic curves fitted to position 71 for the left CS (top) and position 32 for the right CS
(bottom). These positions correspond to the areas of greatest change in the regression
Figure 9 shows the results of the Mann-Whitney tests done to compare the various groups
for the left and right CS respectively. The red dots indicate significant regions on the CS
between the two groups. The greatest difference is seen in between the 18m and 24m
groups in the left CS and right CS between positions 20 and 40, and also between the 12m
and 18m group in the right CS. The positions between 60 and 80 are significantly different
between 18 and 24 months for the left CS and between 24 and 36 months for the right CS.
25

Figure 9. Mann-Whitney test results comparing two groups with 6 month intervals for the left CS (top) and
right CS (bottom). The red dots indicate regions with significant difference.
The results of the Mann-Whitney tests in Figure 9 show that the greatest differences
between age groups are observed before 36 months, with the greatest difference seen in the
left and right CS as follows:

26

12m v 18m 18m v 24m 24m v 36m
Left CS positions 24-45, 65-80 positions 5-10,70-78
Right CS positions 85-95 positions 10-30 positions 38-45, 62-75
Table 2. Table showing positions where Mann-Whitney test results were significant and the corresponding
age groups
We also investigated the relationship between age and the L1 location of the sulcal profile
85
,
an output from BrainVisa, which is indicative of the PPFM position, in subjects with
clearly identifiable bimodal distributions in both the left and right hemispheres. This
resulted in a smaller data size of 50 subjects. Using logarithmic regression, we found a
main effect of age on the location of the L1 position in both, left and right hemispheres.
The uncorrected p-values 0.0011437 for the left hemisphere and 0.043457 for the right
hemisphere.
Additionally, we looked into the relationship between the ratio of the PPFM and the
maximum depth of the CS with respect to age, to explore if age is correlated with the
changes in the PPFM. We found that this ratio was not correlated with age or gender.
3.4 Discussion
Our result show that the CS depth curves have an overall similar shape to that of an adult,
and the PPFM is visible even at 12 months of age. Additionally, linear regression results
show that the CS depth progression mirrors that of surface area changes
66
. Finally, group
differences between age groups show that different regions of the CS evolve at different
times, the most noticeable changes happening before 48 months. The overall shape of the
CS depth profile in all age groups is similar to that of adults
45
. However, as can be observed
from Figure 6, as age progresses there is a steady increase in depth as well as an improved
resemblance to the adult profile. The former result is consistent with existing literature
indicating a rise in brain volume and thickness in the first two years of life
39,87
. Expansion
in surface area of the left precentral gyrus and in volume in the right precentral gyrus have
been demonstrated in a previous study investigating cortical growth in children
66
, and the
mean CS sulcal depth growth profile found here is consistent with these observations.
27

The region around position 50 is known as the “precentral hand knob” or 'Pli de Passage
Fronto-parietal Moyen' (PPFM), a buried gyrus connecting the frontal and parietal lobes
through the CS. From Figure 6, the dip in that region becomes more prominent with age.
The PPFM is an anatomical landmark to hand motor and sensory function
46,47,86
. Cortical
stimulation studies have confirmed the existence of a whole hand motor and sensory
functional area at the anatomical PPFM
88
in addition to individual finger and thumb
function found more inferiorly along the precentral and postcentral gyri. Furthermore,
hand motor function in the same precentral knob like structure was also found in fMRI
studies
89
. In our dataset, the mean PPFM location for all brain volumes for the left
hemisphere was at position 50.3 and at 47.1 for the right hemisphere.
The linear regression results in Figure 7 show the greatest change in between positions 65-
80 for the left CS and positions 30-50 for the right CS. The somatosensory homunculus
indicates that the region around 65-80 on the left CS is associated with the mouth and
tongue region. Due to the age range we are investigating, this strongly suggests that it is
language related and what we see here indeed might be a cortical change linked to the
emergence of language.
The different rates of change between the left and right CS may stem from asymmetry
between the hemispheres. Left and right CS asymmetry has been shown in studies of
handedness
90–92
. One of the causes of the difference in the linear regression results could
be due to emergence of handedness between ages of 18 months and 30 months, however,
this needs to be investigated further.
Previous literature using quantitative genetic analysis in the CS of chimpanzees has shown
significant heritability in the overall CS surface area and depth
93
. The CS depth was shown
to be more heritable in the superior regions, the corresponding significant region shown in
our Figure 7 for the right CS, and less in the inferior and central portions. In a different
study investigating the CS in humans, more heritability was shown in the CS in areas that
associated with the hand and mouth areas
75
, corresponding to the inferior part of the CS,
which corresponds to significant regression points in the CS in Figure 7. While according
28

to Figure 7, the regions of significance in the left and right CS differed, the PPFM area,
which was not considered heritable in previous studies
75,93
, did not show significance in
growth. This might imply that the PPFM is less influenced by genetics and may have more
plasticity. Future studies involving investigation of the PPFM depth and various cognitive
or motor scores may advance our understanding of plasticity involvement in the growth of
the CS.
In a study of regional brain volumes segmented using the FreeSurfer software package
94
in
age 12 months to 60 months, it was shown that changes in surface area in the left precentral
cortex followed a logarithmic curve with significance
66
. The curve for the left CS in Figure
8 mirrors the result shown in that study. While the changes in volume of the left precentral
cortex were not shown, the changes in the right precentral cortex showed a trend that had
a slight decay in slope after about 1800 days of age. Therefore, sulcal depth appears to
follow the curve trends of surface area compared to volume. This is intuitive to interpret
since sulcal depth is measured from the outer cortex of the brain to the white matter and
surface area folds in along the depth.
From Table 2, while not perfectly matched, we see a lot of overlapping regions, especially
between 18 and 30 months of age. According to the somatosensory and motor homunculus,
the overlapping significant positions in the two hemispheres correspond to the upper body
areas such as the hand/fingers (positions 20-35) and face/mouth/jaw regions (position 65-
80). The significance of these positions in relation to development may be understood
through the Center for Disease Control (CDC) developmental milestone guidelines
(https://www.cdc.gov/ncbddd/actearly/pdf/checklists/all_checklists.pdf). There is a great
deal of learning and communication development by the time the child is three. For
example, a child at 12 months maybe able to make sounds to try to say words, while at 18
months he or she will be able to say some words, and at 24 months be able to say sentences
with multiple words as well as repeat words overheard in conversation. Another change
associated with development is a change in diet, as the variety and quantity of food
consumed change. The significant CS positions closer to 90 in the right CS may be
associated with the jaw area and swallowing. Such changes in environment and
29

development may manifest as significant differences in the CS. We also see significant
difference in the in the CS corresponding to the upper body areas between the 18 month
and 24 month groups in both hemispheres. Milestone differences correspond to a child
being able to eat with a spoon at 18 months, while a 24 months old child might build things
with blocks, indicating more dexterous movement. Handedness also begins to emerge at
this age.
However, aside from direct motor function associated observations, speech also
continuously evolves during early childhood, and the changes we observe in this study
corresponding to the hand motor area in the 18 month to 24 month in the left CS and 24
month to36 month range in the right CS may be linked to its development. Previous studies
have reported hand motor excitability in the corticospinal pathways during speech
95–97
.
This excitability was shown to exist in the hand motor area but was not significant in the
primary motor area of the leg
96
. This is thought to be due to an evolutionary link between
the cortical hand area and language development and is associated to the mirror neurons
that are active during both observation and execution of tasks. Speech is also typically
accompanied by involuntary gestures. For instance, it was found that blind subjects who
knew that they were speaking to blind opponents were also found to be gesturing during
speech
98
, representing a strong connection between speech and hand motor area
96
.
Finally, it has been shown that in the adult CS, there are three types of distributions of the
CS; unimodal, bimodal and trimodal
45
. Bimodal distributions, which have a clear inferior
and superior peaks with a dip corresponding to the PPFM, were the most common.
However, 3.6% of the left CS and 16.4% of the right CS were shown to be trimodal, while
1.8% of the left CS and 3.6% of the right CS were shown to be unimodal
45
. In our dataset,
while bimodal was the most common distribution at 77% for the left CS and 62% for the
right CS, 13.7% of the left CS and 29% of the right CS was unimodal, and 10% of the left
CS and 9% of the right CS was trimodal. In this study, we included all distributions for
calculation of the mean sulcal depth curves and linear regression, as well as group
comparisons as we did not want to bias our data to subjects with only bimodal distributions.
Having said that, the same analysis was done using only the CS with bimodal distributions
30

but the results followed a similar pattern. This can be attributed to the fact that the majority
of data in our population sample was of bimodal subjects.
For the linear regression, we also used brain hemispheric volume instead of age as the
dependent variable in our regression model. However, this did not show as strong a
relationship as with age. While age and brain volume showed a reasonable correlation of
0.6, the fact that the PPFM location did not correlate with hemispheric volume may indicate
that the CS development reflects more the functional skills such as those listed in the
developmental milestones, rather than simply brain volume (i.e. a younger child who
happens to have a larger brain has not achieved the same developmental milestones as an
older child, even if the brain volume is smaller). Future studies involving correlation of
functional scores such as motor and language scores with the PPFM will facilitate this
understanding better.
A limitation in our current study is the uneven number of subjects in individual age groups.
Our dataset includes more subjects in the younger ages compared to the older ones and
therefore our sample sizes are not evenly distributed. This makes it difficult to account for
individual variability in the data such as head size. Additionally, due to most of our data
being cross-sectional, it was not possible to declare if and when the CS distributions change,
for example from unimodal to bimodal. However, considering the small percentage of
adults who have unimodal distributions, the CS shape may well change over time to
become more bimodal. A follow-up with longitudinal data will enable us to understand the
change in the CS distribution more accurately.
3.5 Conclusion
The CS may play an important role in understanding development, and the developmental
trajectories more specifically have the potential to play a role as anatomical biomarker for
tracking diseases such as ADHD and Williams syndrome. In the future, due to the
variations in individual children in development, more subjects will be assessed to improve
the accuracy of the trajectory. Additionally, sulcal asymmetry will also be investigated.
31

Longitudinal MRI data, genetic information and handedness will provide more insight into
characterization of the CS.
32

Chapter 4. Neurocranium thickness mapping in early
childhood

Abstract
The neurocranium changes rapidly in early childhood to accommodate the developing
brain. Developmental disorders and environmental factors such as sleep position may lead
to abnormal development of the neurocranium
99,100
. It is important to understand how the
healthy neurocranium develops in order to provide a baseline for early detection of
developmental disorders. However, the neurocranium is not yet well studied in early
childhood, due to lack of available imaging databases. In hospitals, CT is typically used to
image the neurocranium when a pathology is suspected, but the presence of ionizing
radiation makes it harder to construct databases of healthy subjects. In this study, instead,
we use a large dataset of MRI data from healthy normal children in the age range of 12
months to 2 years. After isolating the neurocranium from the MRI, we used conformal
geometry based analysis pipeline to detect local thickness changes over this age span. This
growth model of the neurocranium with shape changes will help us understand cranial bone
and suture development with respect to the brain, which will in turn inform better treatment
strategies for neurocranial disorders.
Keywords: Neurocranium, brain, neonatal/pediatric imaging
Niharika Gajawelli
a,b
, Sean Deoni
c,d
, Jie Shi
e
, Holly Dirks
d
, Marius Linguraru
h
, Marvin D. Nelson
f,g
,
Vidya Rajagopalan
a
, Yalin Wang
e
, Natasha Lepore
a,b,f
a
CIBORG Lab, Department of Radiology, Children's Hospital Los Angeles, CA, USA

b
Department of Biomedical Engineering, University of Southern California, CA, USA

c
Department of Pediatric Radiology Research, Children's Hospital Colorado, CO, USA

d
Department of Biomedical Engineering, Brown University, RI, USA

e
Department of Computer Science, Arizona State University, AZ, USA

f
Department of Radiology, University of Southern California, CA, USA

g
Department of Radiology, Children's Hospital Los Angeles, CA, USA
h
Sheikh Zayed Institute for Pediatric Surgical Innovation, Children’s National Health System, Washington
DC
33

4.1 Introduction
The neurocranium develops rapidly in early childhood. It consists of tiny bony plates of
ossified bone tissue, which are connected by the soft connective tissue of the sutures and
fontanelles
13,23
. This flexible construction permits it to deform and expand to accommodate
the rapidly growing brain in early childhood. The 3 major sutures fuse as the child grows,
the metopic suture closing first
101
by age 1 while the coronal and lambdoid suture close
before age 3. The cranium expands rapidly from 25% of its adult size at birth to 90% of its
adult size by age 4-5
102,25
. The brain, meanwhile reaches 95% of its final volume by the
age of 6.
Abnormalities in the shape of the neurocranium in childhood can happen as a result of
abnormal development or injury. Craniosynostosis, which is the premature closing of the
sutures happens in about 1 in 2000 to 2500 live births
103
, and may be caused by genetic
conditions such as Apert, Crouzon, and Muenke syndromes, or can be non-syndromic.
Untreated craniosynostosis has been linked to significantly lower mean mental and motor
scores compared to normative populations
15
and in one study, to language acquisition
delays
104
. Deformational plagiocephaly has seen a recent increase in prevalence in early
infancy, estimated from 5% to 48%
100
in response to the recommendation that babies
should stay in supine position to prevent sudden infant death syndrome. Studies have
shown some association between motor function delays and deformational plagiocephaly
99
.
While non-synostonic deformities improve with age, craniosynostosis becomes worse and
may lead to adverse developmental outcomes if left undiagnosed or untreated within the
first years of life
15,105
. Neurocranium fractures, the leading symptom of injury in infants
due to motor accidents, falls and abuse
106
may also lead to skull abnormalities. A growth
map of the neurocranium may help in distinguishing between accidental and inflicted head
trauma in children. According to
107
, while the adult cranial suture and cranial bone have a
similar stiffness, the pediatric cranial suture deforms 30 times more than cranial bone
before failure and 243 times more than the adult cranial bone. Due to this, fracture may
cause the neurocranium a major shape change, impacting the brain as well. It is thus
imperative for clinicians to accurately diagnose cranial deformities early, and accurately
34

and monitor progress of the brain and neurocranium development. The lack of normative
data and quantitative imaging tools to analyze neurocranial development are major
obstacles to the early detection and optimal treatment of conditions of infancy.
Subsequently, treatment approaches and their success vary widely among institutions and
clinicians, and are usually based on subjective criteria.
CT images are typically used to diagnose abnormalities of the neurocranium. While studies
show that CT is useful pediatric patients, they have demonstrated a small but significant
increase in the risk of long term adverse outcomes such as developing brain cancer or
leukemia
21
,
108,109
, which suggests that the usage of alternative imaging procedures without
ionizing radiation should be used whenever possible
109
. It is, therefore, inappropriate to
subject normal healthy children to CT to collect data to build a growth map of shape
changes of the neurocranium from CT images. Fortunately, the neurocranium can also be
delineated on MRI and an example neurocranium segmentation for an 18 month old brain
is provided in Figure 10.

Figure 10. Example of skull segmentation shown on an 18 month old brain
MRI has also recently shown value in diagnosis of craniosynostosis by using a novel
sequence to enhance the bone-soft tissue boundary,
110
as well as to investigate brain
abnormalities associated with craniosynostosis prenatally
111
.

In the current study, we map normative, neurocranial thickness in children between 6 and
36 months of age using a large, preexisting database of MRI scans of healthy children
35

ranging in age from 0-2 years old, acquired at the Baby Imaging Lab
(http://www.babyimaginglab.com), We focus on thickness because it is an important factor
in predicting susceptibility to injury, as well in diseases of the neurocranium such as
thalassemia and Paget’s disease. Thickness measurements also have potential for
applications in electroencephalogram measurements, that can be erroneous due to variation
in neurocranium and scalp thickness
112,113
.
In this paper, we use neurocranium masks derived from high resolution MRI brain scans
in young children and demonstrate a pipeline for the novel application of investigating
neurocranium thickness on MRI data. We first generate tetrahedral meshes allowing sub-
voxel resolution analysis of the thin neurocranium surfaces. We then solve the Laplace
equation at each vertex on the neurocranium surface, creating a harmonic field, through
which we calculate thickness using streamlines. The weighted-SPHARM approach is used
to register surfaces, allowing group comparison. We add to the body of studies involving
CT data in ages under 2 to map neurocranial thickness.
4.2 Method
Data
The dataset used includes 55 T1 MP-RAGE MRI scans (1.4-1.8mm
3
) of healthy normal
children between the ages of 0-3 years of age at the time of MRI scanning. The inclusion
criteria is as follows: singleton, full term (37-42 weeks at birth) with no abnormalities on
fetal ultrasound and no reported history of neurological events or disorders in the infant.
Data acquisition details can be found in
76,77
. The study was approved by the Institutional
Review Board of Brown University and informed consent was obtained from the guardians
of all participants. All data was de-identified before pre-processing. Note that the
handedness of the subjects are not assessed in this study due to lack of handedness
information at early ages. The scans at approximate ages of 6 months, 9 months, 12 months,
18 months, 24 months, and 36 months randomly selected from the Baby imaging database.
All data were acquired on a Siemens 3T Tim Trio scanner equipped with a 12 channel head
RF array. To minimize intra-scan motion, children were asleep and swaddled with a
pediatric MedVac vacuum immobilization bag (CFI Medical Solutions, USA) and foam
36

cushions. Scanner noise was reduced by lessening the peak gradient amplitudes and slew-
rates, and using a noise-insulating scanner bore insert (Quiet Barrier HD Composite,
UltraBarrier, USA).
Pre-Processing
The data was pre-processed using the pipeline (shown in Figure 11 ) and described as
follows. First, we used the FSL BET software to remove the neurocranium of the T1-
weighted anatomical dataset in order to ensure accurate registration to the template. While
we are interested in investigating the neurocranium, we needed to register the data for
comparison of groups. Without removing the neurocranium initially, the high intensity
signal from scalp outside the brain caused the registration to be imprecise. The
neurocranium stripped data was then resampled to a 1x1x1mm
3
resolution for consistency
throughout processing. The resampled image was bias corrected using the N4 ANTs bias
correction tool
114
. The result was then linearly registered to an age matched template, which
was previously generated from the same dataset and also resampled to 1x1x1mm
3

resolution, using FSL FLIRT
79,80
with 6 degrees of freedom. This transformation from
FSL FLIRT was saved and then applied to the original dataset with the neurocranium. The
resulting registered image was then bias field corrected using the N4 ANTS bias correction
tool. Finally, we applied the FSL BET
115,116
tool to extract the outline of the inner and outer
neurocraniums. Each mask was visually inspected to ensure accuracy and to eliminate
datasets with overlapping voxels between the inner and outer neurocranium or including
areas that were not part of the neurocranium.
37

Figure 11. Pre-processing pipeline
After visual inspection, the datasets with the clearest delineation of the neurocranium mask
were chosen for this study specifically. The dataset consists of 6 subjects in the 6 month
cohort, 8 subjects in the 9 month cohort, 9 subjects for the 12 month cohort, 8 subjects for
the 18 month cohort and 10 subjects for the 24 month cohort as shown in Figure 12. The
extracted neurocranium does not include the craniofacial skeleton.
38

Figure 12. Distribution of subjects in cohort
Thickness Calculation
The neurocranium thickness calculation used in this paper involved a similar approach to
that used for the corpus callosum in
117,118
. We used the Iso2mesh toolbox
119
to create
volumetric tetrahedral meshes for the inner and outer neurocranium masks generated by
FSL BET
116
. Visual representation of an example of the extracted neurocranium and
tetrahedral mesh are shown in Figure 13.
The thickness was calculated from the inner boundary (inferior patch) to the vertices on
the outer surface mesh (superior patch). To calculate the thickness, the string energy on the
mesh was first defined. Using vi and vj as vertices on the mesh and the edge connecting the
two defined as [vi vj], a piece-wise linear function f:V R was defined, where the set of all
vertices on the mesh is V. If each edge [vi vj] is assigned a string constant k(vi vj), the string
energy on the mesh is defined as
𝐸 (𝑓 ) = < 𝑓 , 𝑓 > = ∑ 𝑘 (𝑣𝑖 , 𝑣𝑗
[𝑣𝑖 ,𝑣𝑗 ]∊𝐾 )(𝑓 (𝑣𝑖 ) − 𝑓 (𝑣𝑗 ))(𝑓 (𝑣𝑖 ) − 𝑓 (𝑣𝑗 ) (1)
,where K is the set of all edges on the mesh. Here k(vi vj) is called the discrete harmonic
energy
120
. The volumetric Laplace-Beltami operator on a vertex vi of a tetrahedral mesh is
∆𝒇𝑳𝑩 (𝒗𝒊 ) = ∑ 𝑘 (𝑣𝑖 , 𝑣𝑗
[𝑣𝑖 ,𝑣𝑗 ]∊𝐾 )(𝑓 (𝑣𝑖 ) − 𝑓 (𝑣𝑗 )) (2)
39

∆𝒇𝑳𝑩 (𝒗𝒊 )=0 when f is minimized and satisfies Laplace’s equation. We define the values
of the function f on the superior boundary (B1) as 1 and the inferior boundary (B0) as 0.
The function f is minimized for all other internal vertices as described in the boundary
conditions below.
∆𝑓 𝐿𝐵 (𝑣𝑖 ) = 0 ∀𝑝 ∉ (𝐵 0 ∪ 𝐵 1)
𝑓 (𝑣𝑖 ) = 0 ∀𝑝 𝐵 0,
𝑓 (𝑣𝑖 ) = 1 ∀𝑝 𝐵 1

A volumetric harmonic field was then built between the two patches by solving a linear
system defined with the Laplacian matrix. Streamlines
121,122
were constructed to connect
the two surfaces by solving the equation below. Here u(s) is a parametric curve with arc
length s and x is a point on the surface patch we start from.
𝑢 ′
(𝑠 ) = ±
∇f(u(s))
|∇𝑓 (𝑢 (𝑠 ))|
, 𝑢 (0) = 𝑥 (3)
The thickness is defined as the total arc length of the streamline that traverses the
neurocranium from superior to inferior patches
117,120,121
.

Figure 13. (a) Example of neurocranium extracted using FSL. The outer region is the superior patch and the
inner table is defined as the inferior patch. (b) Tetrahedral mesh created from the extracted neurocranium
using the Iso2mesh toolbox. (c) Zoomed in figure of the tetrahedral mesh enclosed in the square.
Streamlines are later generated from the outer to the inner patches to compute thickness.
40

Surface registration
Next, we used the weighted-SPHARM approach described in
123,124
to register the thickness
of each individual subject to a common template. This enabled us to get a surface with the
same number of vertices for each subject, allowing us to compare neurocranial thickness
between groups at various regions. Each neurocranium’s inferior and superior surfaces
were mapped to a sphere using an area preserving, surface flattening algorithm, described
in (http://brainimaging.waisman.wisc.edu/~chung/midus/). Spherical hamonics was used
in this study due to the closeness in shape to the neurocranium. We utilized a template
sphere with 40962 vertices to compute the spherical harmonic confidents in order to get a
SPHARM representation of the surfaces. Finally, we applied two sample t-tests for
thickness comparison between the groups as follows: a. 9m vs 12m, b.12m vs 18m, c. 18m
vs 24m, and d. 12m vs 24m. Finally, we conducted a regression analysis to understand the
changes in neurocranium thickness with respect to age. Due to the small number of
subjects in the 6 month age group, we did not use this in the group comparison, however,
this dataset was included in the regression analysis to investigate thickness change with
respect to age. The reason the 6 month data was included in the regression was because the
age in days was used and there was not need to group data as was needed with the group
comparisons.
4.3 Results
Figure 14 shows the neurocranium thickness at a.12 months, b. 24 months and c. 36 months,
where the color bar denote the thickness of the neurocranium, The color bar goes from dark
blue to red, the red indicating the thicker regions. Our results show consistent increase in
the thickness of the neurocranium in the posterior part of the brain towards the lambdoid
suture, indicated with the red arrows. The inferior region of the neurocranium (z-value <
95), including regions closer to the air cavities and areas closer or below the cerebellum
were disregarded in this study due to the lack of accuracy and variability in the data. This
region has been set to 0 in the results below.
41

Figure 14 Neurocranium thickness in 3 age groups. (a) 12 months, (b) 24 months, and (c) 36 months. Color
bar indicates thickness values
Figure 15 represents the p-values showing significant change in thickness of the
neurocranium between the ages of a. 9 and 12 months, b. 12 and 18 months, c. 18 and 24
months and d. 24 and 36 months acquired by conducting non-parametric t-tests and
correcting for multiple comparisons using permutation testing with 10000 permutations to
significance value of p=0.05. Figure 15 above shows the regions of the neurocranium that
vary the most across comparison of the two groups. Our results show the largest difference
between the ages of 12 and 18 months for the posterior part of the head, which as with
thickness, seems to overlap with part of the lambdoid suture. Using the same parameters
for group comparison, Figure 15 shows the difference between the neurocranium thickness
at 12 months and 36 months.
Figure 17 is linear regression showing neurocranium thickness change with respect to age
starting at 6 months until 36 months. The regions in red show the largest change. We used
regression analysis in MATLAB to map the change in neurocranial thickness over time,
with focus on the posterior point on the neurocranium that showed the largest change.
Figure 17 (a1) and (a2), which illustrates the regression coefficients and the p-values of the
neurocranium thickness change with respect to the child’s age (in days) from 6 months to
36 months, again show the largest change with respect to age is in the posterior part of the
brain. Here, again our results show that the largest change is localized to areas
corresponding to the lambdoid suture area of the neurocranium. Finally, using the most
significant vertex, we plotted the progression of neurocranium thickness change again age
in children between the ages of 6 months and 24 months in Figure 17 (b). Here our results
42

show a linear trend. The 36 month data was excluded for this figure as the t-tests showed
little difference in neurocranium thickness between the age groups of 24 and 36 months.

Figure 15. Group differences between each consecutive group. Results were corrected using permutation
testing using 10000 permutations and a significance threshold of 0.05. The color bar indicates the p-value.
43

Figure 16. Results of significant difference (p=0.05) after non-parametric t-tests and multiple comparison
correction using permutation testing (10000 permutations) comparing 12 months and 36 month groups.
Biggest difference is seen around the lambdoid suture. The color bar indicates the p value.
44

Figure 17. Results of regression showing change over time from 6 months to 24 months after FDR
correction to significance level of p=0.05. Red regions indicate the coefficients showing biggest change
over time. Regression coefficients shown in (a1), and corresponding p-values are shown in (a2). Color bar
indicates [max (p-value) – p-value] for visualization. Neurocranium thickness change over time for the
maximum intensity vertex in the black dotted circle for children between ages of 6 months and 24 months
shown in (b).
4.4 Discussion
Our results give us insight into the development of the neurocranium, as maps shows a
clear increase in neurocranial thickness in the posterior part of the head, likely overlapping
with parts of the lambdoid suture. While most studies focus on global measures such as
neurocranial volume or head circumference, our study, using non-invasive MRI adds to the
45

body of previous studies by providing information on the local changes of the
neurocranium.
Neurocranial growth has also been investigated using non-invasive 3D optical surface
scanner data in children of 6-12 months of age
125
. They showed that the head grows more
in length (2.84%), by measurement in the head circumference increase. At the same time,
the total cranial volume increased by 18.76%
125
.
In
23
, authors developed a statistical head model for children aged 0-3, allowing them to
predict neurocranium geometry. Parameters such as age and head circumference were used
to predict features such as neurocranium size/shape, suture geometry and neurocranium
thickness. There, the authors showed a greater thickness of the neurocranium in the
occipital region in ages 1.5 years and older. In addition, the model also showed that the
lambdoid suture closes between the ages of 1 and 1.5 and neurocranium thickness increased
in regions corresponding fontanelle after it closes.
In Figure 15 (b)Figure 15, the most rapid increase in thickness was near the posterior
fontanelle and along the lambdoid suture between the ages of 12 and 18 months. Thickness
differences found in Figure 15 (b), agree with the neurocranium growth model generated
from CT data
23
, where increased thickness in the posterior neurocranium and regions
around the lambdoid suture. Hence it can be reasoned that the neurocranium thickness
change in the posterior region of the neurocranium might be associated to the closing of
the lambdoid suture at around 18 months. Additionally, we also see a thickness increase in
the parietal bones surrounding the sagittal suture between these age groups, consistent with
marginal increase in thickness shown in the CT data generated statistical model
23
.
The largest difference between the 18 and 24 months old cohorts (Figure 15 (c)) was seen
in the anterior part of the head and in the bones surrounding the occipital and temporal
lobes. The concentration of points in the temporal region corresponds to a region close near
the squamosal suture that divides the parietal and temporal bones.
46

Finally, while comparison between 24 and 36 months old groups (Figure 15 (d)) showed
little significant change in the occipital bone, significant differences were seen in the
anterior part of the head in the frontal bone and on the temporal bone, similar to locations
seen in Figure 15 (b). This indicates that most of the neurocranium thickness changes in
the posterior and occipital bone area happens prior to 2 years, but areas on the frontal bone
and temporal bone may continue to evolve even after age 2.
Figure 16 shows the thickness difference between 12 and 36 months; the biggest difference
around the posterior fontanel and regions corresponding to the lambdoid sutures. Again,
this may be attributed to neurocranium growth after posterior fontanel fuses.
During infancy, between 0-2 years of age, the brain increases rapidly in surface area and
cortical thickness increases. It has been shown that in the first year of life, the brain surface
area expands by about 1.80 times of its original area, with the highest expansion in regions
including the occipital lobe
36
. Additionally, occipital cortex thickness also increased
rapidly between the ages 0 to 9 months
35
. Such growth in the brain drive neurocranium
changes.
Normal healthy development of the neurocranium and its relationship to brain development
is also relevant in understanding deformational plagiocephaly, which occurs due to the
continuous placement of an infant in one position. Some recent studies have shown that
children with deformational plagiocephaly have motor delays in early childhood
99
.
Although this is not a causal relationship, and it is possible that deformational
plagiocephaly occurs in children who have developmental delays, it will be beneficial for
physicians assessing these children to have a growth model to of the normally developing
neurocranium to compare to as well as to recognize the relationship between neurocranium
and brain development in early childhood to prevent developmental delays from persisting.
A limitation in our study is that we could reliably extract the neurocranium as
neurocranium from a small number of subjects as the neurocranium is quite thin in early
childhood and difficult to automatically extract from MRI data. To overcome this problem,
future studies investigating the neurocranium may benefit from new MRI sequences to
47

enhance the bone-soft tissue boundary, such as described in
110
. The neurocranium models
in
23
also show changes in the anterior part of the neurocranium, near the coronal suture,
that we were not able to observe in our group comparisons. This may be due to the fact that
the thickness change in the anterior part of the neurocranium was lower compared to the
posterior part, as shown by the neurocranium thickness model in
23
. This model also shows
a small area of the sagittal suture closed after age 2, and this might be the biological reason
thickness increase in the anterior region is not seen in the age range used in this study.
4.5 Conclusion
In this study, we investigated the growth of the neurocranium across ages 6 to 36 months.
Our results showed the biggest change in the posterior part of the neurocranium, where the
lambdoid suture lies, between 12 and 18 months, corroborating results of the geometric
head model created from CT data
23
. Therefore, it is important in future studies to
investigate younger age groups as well as age groups in smaller intervals to build a more
detailed timeline of neurocranium growth.
The neurocranium growth model will improve over time by inclusion of more subjects to
become a reliable template of neurocranium growth. We show in this pilot study, that using
MRI head scans, neurocranium thickness growth can be mapped out to show changes in
areas that correspond to previous CT studies. Future work will include a correlation of the
neurocranium thickness and parameters acquired from the brain, such as cortical thickness,
which will allow us to acquire a more complete picture of the development of the brain and
neurocranium. The association between plagiocephaly and neurocranium development,
particularly in light in of the fact that our results show consistent group differences in the
back of the neurocranium, will also be investigated.

48

Chapter 5. Corpus callosum development in children
Niharika Gajawelli
a,b
, Sean Deoni
c,d
, Carlos Salazar
b
, Natasha Paquette
a
, Natalie Ramsy
b
, Holly Dirks
d
,
Douglas Dean, Jonathan
d
, O’Muircheartaigh
d
, Marvin D. Nelson
e,f
, Yalin Wang
g
, Natasha Lepore
b,f

a
CIBORG Lab, Department of Radiology, Children's Hospital Los Angeles, CA, USA

b
Department of Biomedical Engineering, University of Southern California, CA, USA

c
Department of Pediatric Radiology Research, Children's Hospital Colorado, CO, USA

d
Department of Biomedical Engineering, Brown University, RI, USA
e
Department of Radiology, University of Southern California, CA, USA

fDepartment of Radiology, Children's Hospital Los Angeles, CA, USA
g
Department of Computer Science, Arizona State University, AZ, USA

5.1 Introduction
The corpus callosum (CC) is the largest white matter fiber bundle in the brain and integrates
motor, sensory and cognitive processes between the two hemispheres of the brain. The CC
has many homotopic projections, connecting the homologous areas between the two
hemispheric regions
126
. The anterior callosal fibers connecting the frontal lobes transfer
motor information. The callosum fibers, connect the posterior parietal, temporal and
occipital lobes in both hemispheres and fuse multimodality sensory information.
Specifically, auditory information integration is associated with the isthmus of the CC and
visual information integration is associated with the splenium
61,117
.
Increasing number of studies report that cognitive and behavioral deficits in
neurodevelopmental disorders correlate with subtle structural changes in the CC. These
diseases include autism
60,127
, schizophrenia
21,128,129
and attention-deficit disorder. Studies
have shown sex differences in the sub-regions of the CC
130,131
and have associated sex
differences with morphometric abnormalities in neuropsychiatric disorders.
49

Figure 18. Corpus Callosum outlined in blue is easily visible on a midsagittal slice of the brain
Anatomically, the midsagittal slice of the CC can be divided into multiple regions, such as
the genu, (the anterior part of the CC), the rostrum (the inferior part of the CC after the
anterior bend), the body, and the splenium (the posterior part of the CC). It has been shown
that different areas of the CC projects axons to different areas of the brain
132
. The CC can
be seen clearly on a midsagittal MRI slice and is shown in Figure 18. The WM contrast of
the CC is produced due to the myelin sheath that wraps around the axonal fibers of the CC.
During early brain development, the CC myelinates at different rates postnatally as
examined in a study done of autopsied infants
9
. In this study, it was shown that the rostrum
myelinated later than other callosal regions, with grossly visible myelin of 50% of cases at
12 months age and myelination continuing after the 2
nd
year. The body of the CC, with
fibers connecting the midportions of the posterior frontal and parietal lobes of the brain, on
the other hand showed microscopic myelination already at term in 25% of cases.
Myelination occurred rapidly after 5 months of age, and 90% of cases were grossly
myelinated by 13 months of age. The splenium, containing fibers connecting the temporal
and occipital lobes, had grossly visible myelination in 90% of the cases by 21 months
9
.
50

Other studies on CC development have shown callosal size changes or thickness changes
through analysis of the midsagittal slice in two dimensions. For example, Giedd et al.
showed that the midsagittal CC areas increased linearly from ages 4 -18 in healthy children
and that posterior and mid-regions of the CC showed larger age-related changes than
anterior regions
130
. In another study that included a younger cohort of subjects, from 1
month of age to adults, it was also shown that the developmental trajectory was non-linear
61
.
CC area increases were noted, following a similar trajectory as the cortex, in the first few
years of life regardless of gender
61
.
Apart from structural T1-weighted scans, the CC can also be investigated using diffusion
tensor imaging (DTI), which is most commonly used for WM studies. Through measures
of fractional anisotropy, DTI can show the extent of diffusion in the WM tracts such as the
CC
126,132
. However, despite accessibility of modalities to investigate the this structure,
currently, little is known on structural normal pediatric development of the CC in early
childhood, especially in children under age 5, due to the lack of largescale systematic
neuroimaging studies in early childhood. The exception being studies done on myelin water
fraction (MWF) showing increasing degree of myelination in children from 3 months to 60
months
38,133
.
In the limited number of existing anatomical MRI studies
21,61,130
that specifically examined
the heathy development of the CC, the midsagittal slice size was investigated. While such
analysis offers important information regarding CC development, the CC extends out
laterally from the midsagittal slices. Hence, three dimensional surface based measures,
combined with more traditional measures such as thickness, will add value in mapping out
CC development more comprehensively. For example, CC differences have also been
shown in congenitally blind, late blind (onset > 8years), and sighted subjects in a previous
study using surface based morphometry analysis
117,120
. It was shown that visual
deprivation occurring after the visual cerebral architecture was developed still impacted
the morphological development of the splenium of the CC
117,120
. However, the impact is
lessened when blindness was acquired late in development, possibly because already
myelinated regions might be more resistant to change
117,120
.
51

In this paper, we investigate the CC development in children aged 12 to 60 months using
combined thickness and surface-based morphometry measures to extract localized
differences that might be associated with early brain development.
5.2 Method
Data
Our dataset comprised brain volumes from 50 healthy subjects ranging from 365 to 1900
days of age (roughly corresponding to 12 months to 60 months) and randomly selected
from the Advanced Baby Imaging Lab database (www.babyimaginglab.com). Group
comparisons were done using 4 age groups of 12 months, 24 months, 36 months, and 48+60
months (combined due to lack of subjects in each group). Gender was not investigated in
this study.
The distribution of subjects is shown in the Table 3. The data used consisted of high
resolution 3 dimensional T1 MP-RAGE MRI scans of the brain (1.4-1.8mm
3
). Inclusion
criteria for the participants were: singleton birth between 37 and 42 weeks gestation with
no abnormalities on fetal ultrasound and no reported history of neurological events or
disorders in the infant. Data acquisition details can be found in
76,77
. Prior to the MRI
acquisition, each subject or their guardian was informed of the goals of the study and signed
a formal consent. The study was approved by the Institutional Review Board of Brown
University and informed consent was obtained from the guardians of all participants. All
data was de-identified before pre-processing. Note that the handedness of the subjects are
not assessed in this study due to scarcity of handedness information at early ages. However,
in future longitudinal studies, handedness information will be included retroactively.
Age Number of subjects
12 months 10
24 months 12
36 months 16
48+ 60 months 12
Table 3. Distribution of subjects in different age groups
52

Processing pipeline
The T1-weighted MP-RAGE scans were first skull-stripped and then rigidly registered
using 6 degrees of freedom to an age-matched template. The CC was then manually
segmented using the ITK-snap toolkit (http://www.itksnap.org)
134
. Initial segmentations
were advised by an expert in pediatric radiology and average intra-rater agreement was
0.8421. Sagittal segmented example images of the CC are shown in Figure 19. The lateral
boundaries were traced until the CC started to radiate and merge with the white matter.

Figure 19. CC segmentations for different age groups
After segmentation, 3-dimensional tetrahedral meshes representing the CC were created
using an in house conformal mapping program. This program is based on an adaptively
sized tetrahedral mesh molding method
135
. An example of a mesh is shown in Figure 20.
53

Figure 20. Example of a tetrahedral mesh generated on the segmentation of a CC
Surface registration and multivariate tensor based morphometry (mTBM)
Since the CC is long and thin in shape, subcortical surface parameterization algorithms
136

such as spherical harmonics
137
might lead to large distortions in the registration. Therefore,
instead of a sphere, the CCs are modeled as surfaces topologically equivalent to cylinders
with two open boundaries using the holomorphic one-form based method
138
. Constrained
harmonic mapping is used to achieve one-to-one correspondence between the cylindrical
plane and each anatomical surface. In mTBM, the registration between each subject and
the template yields a displacement field. At each vertex, a Jacobian matrix
𝐽 =
𝑑𝜙𝑢 𝑑𝑢
𝑑𝜙𝑢 𝑑𝑢

𝑑𝜙𝑣 𝑑𝑢
𝑑𝜙𝑣 𝑑𝑣

, can then be computed from the deformation field and parameters from the
Jacobian matrix can be used as metrics for subsequent group comparisons.
The determinant of J (det J) is an important univariate measure of local area changes and
their directions (i.e. shrinkage or expansion), however, this gives us an incomplete picture
of the CC morphological changes with age. Multivariate measures, such as mTBM have
been shown to give more statistical power compared to univariate measures
139–141
.
54

Therefore, here we used the deformation tensor, used in mTBM, defined by 𝑆 = √𝐽𝑇𝐽 1/2
,
also known as the Cauchy-Green tensor. On the each vertex of the tetrahedral mesh, the
Jacobian is computed by [w3 – w1, w2 – w1][v3 – v1, v2 – v1]
−1
, where [v1,v2,v3] is a
triangle on the subject mesh being mapped on a triangle on a template mesh [w1, w2, w3].
Since 𝑆 = √𝐽𝑇𝐽 1/2
, gives us 3 elements at each vertex, log(S) can be computed to get a
3x1 feature vector to use as a multivariate measure
142
.
Thickness computation
Radial distance is one of the most commonly used morphometry measure on surface
data
141,143
. Specifically, radial distance calculated in 3D structure informs us on the entire
surface CC and as opposed to solely the midsagittal plane. The radial distance is calculated
as the distance from the medial axis to each vertex
141
and the medial axis is found using
the mid-point of the iso-parametric curves, which are perpendicular to the computed
conformal grid. While MTBM is sensitive to changes such as shear along the surface
tangent direction, the thickness calculated allows us to investigate change along the surface
normal direction. Combining both of these measures, we construct a 4x1 multivariate
feature vector
120,141
.
Group statistics
For group comparisons, the Hotelling’s T
2
test, which is the multivariate generalization of
the Student’s t-test
144
, was applied on sets of values in the Log-Euclidean space of the
deformation tensors. The Mahalanobis distance M, defined below, was used to measure the
mean vectors differences between age groups.
𝑀 =
𝑁𝑠𝑁𝑇 𝑁𝑠 +𝑁𝑇
(𝑆 − 𝑇 )
′
∑ −1 (𝑆 − 𝑇 ) (1)
Here Ns and NT are the number of subjects in two different groups, S and T are the mean
vector of each group, and sigma is the combined covariance matrix of these two groups
142
.
To eliminate the assumption for a normal distribution, we used a permutation test,
randomly assigning subjects to different groups and then compared the true labels to the
55

distributions generated
120,139
. A second permutation test was run to correct for multiple
comparisons
120,139,142,145
. The permutations were repeated 5000 times in both cases.
Group statistics were calculated between 12 and 24 months, 24 and 36 months, 36 and
48+60 month group cohorts. The 48 and 60 month groups were combined due to the small
number of subjects in each of those individual groups. Finally, a regression model was
applied to model the change in MAD of group through this entire age group.
5.3 Results
The global map significance of univariate and multivariate comparisons for each age
comparisons is displayed in Figure 21 for the 12m-24m comparison, Figure 22 for the
24m-36m comparison, and Figure 23 for the 36m-48+m comparison. In each figures, a.
represents the determinant of the Jacobian matrix (left top), b. represents mTBM results
(left bottom), c. represents the medial axis distance (MAD), (right top), and d. represents
the combined measure of thickness and mTBM (MADMTBM) results, (right bottom). The
red values show greater significance. Figure 24 shows the results of linear regression
analysis.
The greatest area of significance between groups can be seen at the genu and splenium of
the CC, both at the junction of the body. We can see difference as young as 12-24m,
indicating that the CC undergoes significant growth after the first years of life and changes
in growth continue to be significant in the older age groups (36m-48+m). In addition, while
changes in the genu of the CC remain significant across all age groups, changes observed
in the splenium of the CC is more restricted in older age groups (36m-48+m) compared
with younger age groups comparison (12m-24m). Of note, visual inspection of the
univariate and multivariate measures reveal that the combined MAD+MTBM measure has
the greatest sensitivity to localized changes, thus allowing a better localization of the
morphological changes. Linear regression shows that the area of biggest change is in the
genu.
56

Figure 21. Results showing a. the determinant of J (left top), b. MTBM (left bottom), c. MAD (right top),
and d. MADMTBM (right bottom) between the 12m vs 24m groups

Figure 22. Results showing a. the determinant of J (left top), b. MTBM (left bottom), c. MAD (right top),
and d. MADMTBM (right bottom) between the 24m vs 36m groups
57

Figure 23. Results showing a. the determinant of J (left top), b. MTBM (left bottom), c. MAD (right top),
and d. MADMTBM (right bottom) between the 36m vs 48m+ groups

Figure 24. Linear regression results showing regions of greatest change in all subject data
5.4 Discussion
In DTI tractography studies of WM fiber projections from the CC, it has been shown that
the fibers from the genu project to the prefrontal lobe, fibers from body project to the
premotor and supplementary motor areas, the primary motor cortex, primary sensory
regions, and the remaining part of the CC, including the splenium, projects fibers to the
parietal, occipital, and temporal lobes
132
. Hence, the group differences found in this study
stem from cortical and white matter tract development in these particular regions. For
example, improvement in motor functioning or planification in toddler and young children
may be associated with morphological changes within the CC and the genu.
58

In this study, we applied combined univariate and multivariate measures of the
morphological changes of the CC in early development. This is the first time a surface
based morphometry analysis has been done in children between the ages of 12 to 60 months.
Our results show consistent changes in the CC from 12 months to 60 months, with the
greatest changes occurring in the genu and splenium regions of the CC across all age groups.
Furthermore, as can be observed in the figures above, all measures gave consistent results.
Nevertheless, MADMTBM showed the most exhaustive changes between the groups
suggesting higher detection power of local surface changes of the CC.
Interestingly, between 12 and 24 months of ages, all the comparisons done shows
consistent differences in the splenium of the CC. These results are consistent with a post-
mortem study that showed the splenium was myelinated in 90% of autopsied cases in the
study by 21 months of age
146
. The white matter fibers from the splenium connect to regions
of the temporal and occipital lobes as shown in previous DTI studies
132
. The significant
difference we observed in this region might reflect the early changes associated with the
visual system development known to mature early that in childhood.
In contrast, between 24 and 36 months of age the most prominent difference can be seen
in in the genu and rostrum area of the CC (Figure 22). While the splenium region are known
to develop early, our results suggests that the rostrum myelinates later in childhood, with
its development continuing past the 2
nd
year. Furthermore, between 36 and 48+ months of
age (Figure 23), continuing growth of the CC can be observed, with significant differences
persisting in the genu and the splenium, although more restrain than the 24 and 36m group
comparison. Nonetheless, our results further suggest developmental changes on the body
of the CC occurring during that age range.
The regression result of Figure 24 indicates the rapid rate of growth in the genu of the CC
through the age group studied. Our results are consistent with one of the few previous
studies including our age range of interest, where tensor maps, indicating radial tissue
expansion rates, were used to determine a subject’s growth between 3 to 6 years
22
. In this
age range, the anterior region of the CC, including the genu, showed the largest growth.
59

The anterior region of the CC is involved in executive functioning, which include skills
such as mental vigilance, organization, regulating behavior and action planning. Thus, the
rapid development of the CC observed in our study, may be associated in the emergence
of the early executive functions in the developing brain.
Myelination studies done in all WM, has shown exponential increase in the myelin water
fraction (MWF) in the first 500 days in all parts of the CC, after which myelination
increased more gradually
38
. The significant changes we see in our results (Figure 21, Figure
22) in the genu and splenium of the CC may reflect this exponential increase. However,
while increase in myelinated white matter (MWF) was also seen in the body, we did not
found significant changes in the major part of the body of the CC, indicating that the
changes captured by the surface based mTBM are different from those measured by MWF.
Additionally, since the body of the CC is heavily myelinated by 12 months of age
9
the age
range used in this study may not capture changes that happen in the body of the CC.
5.5 Conclusion
In this paper, we showed that using combined measures of multivariate MTBM and MAD,
we can identify localized changes in the CC, more pronounced in the genu and splenium,
between the ages of 1 and 5. Future studies will include creation of developmental
trajectories as well as include more ages to get a full range of subjects.

60

Chapter 6. Conclusions and Future Work
Understanding normal development of pediatric brain structures and developmental
trajectories is indispensable. This dissertation investigated three key structures of pediatric
brain development - the central sulcus, the neurocranium, and the corpus callosum.
6.1 Central sulcus
We investigated the CS depth in children between the ages of 1 and 5. We found that CS
depth changes mirror the brain surface area changes.
Future work: Cognitive scores such as motor and language scores can help us understand
function and correlate them to morphological changes. The PPFM is an area of plasticity
and its investigation with respect to cognitive scores may prove useful. Asymmetry and
gender differences may be explored with growing data bases.
6.2 Neurocranium
Neurocranium thickness was examined through use of MRI in children between 6 months
to 36 months of age with a new surface morphometry pipeline. Thickness changes were
found in the areas that correspond to the lambdoid suture and occipital fontanel.
Future work: Brain and neurocranium development are influenced by the same genes.
Investigation of co-development of the regional brain changes and neurocranium may
provide great insight.
6.3 Corpus callosum
Surface mTBM was used for the first time to examine changes of the CC in early childhood
and differences in the genu region was prominent, corresponding to early executive
function.
Future work: Correlation of the cognitive scores with CC growth is the logical next step.
Growth differences in other subcortical structures brain regions such as the hippocampus,
putamen and thalamus. Correlating those with developmental scores may prove useful to
understand function.
61

Finally, the overarching goal could be creating age- and sex-matched healthy normal
standards in pediatric brain structure including morphometry 3D maps of the whole brain
and of subcortical structures, head circumference, total grey and white matter volume,
ventricular cerebrospinal fluid volume. With these templates and measures in place,
research into creating comparative standards with a patient population paves way for early
medical intervention and cognitive therapies
6.4 Publications
Journal publications
1. N. Gajawelli, Y. Lao, M. Apuzzo, R. Romano, C. Liu, S. Tsao, D. Hwang, B. Wilkins, N.
Lepore, and M. Law. "Neuroimaging changes in the brain in contact versus noncontact
sport athletes using diffusion tensor imaging." World neurosurgery 80, no. 6 (2013)
2. B. Wilkins, N. Lee, N. Gajawelli, M. Law, and N. Leporé. "Fiber estimation and
tractography in diffusion MRI: Development of simulated brain images and comparison of
multi-fiber analysis methods at clinical b-values." NeuroImage (2014).
3. Lao, Y., Nguyen, B., Tsao, S., Gajawelli, N., Law, M., Chui, H., Weiner, M., Wang, Y.
and Leporé, N., 2017. A T1 and DTI fused 3D corpus callosum analysis in MCI subjects
with high and low cardiovascular risk profile. NeuroImage: Clinical, 14, pp.298-307.
4. Tsao, S, Gajawelli, N., Zhou, J., Shi, J., Ye, J., Wang, Y., Leporé, N., "Feature selective
temporal prediction of Alzheimer's disease progression using hippocampus surface
morphometry." Brain and Behavior (2017).
To be submitted
1. Niharika Gajawelli, Natalie Ramsy, Sean Deoni, Holly Dirks, Douglas Dean, James
OMuircheartaigh, Marvin D Nelson, Olivier Coulon* and Natasha
Lepore*, ”Developmental trajectory of the central sulcus in young children”, in preparation
for submission to Human Brain Mapping (2018).
2. Niharika Gajawelli, Sean Deoni, Jie Shi, Holly Dirks, Marvin D Nelson, Yalin Wang,
Natasha Lepore, ”Mapping Skull Thickness in Early Childhood”, in preparation for
submission to Neuroimage (2018).
3. Niharika Gajawelli, Sinchai Tsao, Michael Kromnick, Marvin Nelson, Natasha
Lepore, ”International trends in adoption of post-processing in clinical medical imaging”,
in preparation of submission to Journal of the American College of Radiology (2018).
62

4. Niharika Gajawelli, Carlos Salazar, Natalie Ramsy, Sean Deoni, Jie Shi, Holly Dirks,
Marvin D Nelson, Yalin Wang, Natasha Lepore, ”Corpus Callosum changes in early
childhood”, in preparation for submission to Neuroimage Clinical (2018).
Conference papers
1. Gajawelli, Niharika, Sean Deoni, Holly Dirks, Douglas Dean, Jonathan O'Muircheartaigh,
Yalin Wang, Marvin D. Nelson, Olivier Coulon, and Natasha Lepore. "Central sulcus
development in early childhood." In Engineering in Medicine and Biology Society (EMBC),
2017 39th Annual International Conference of the IEEE, pp. 161-164. IEEE, 2017.
2. Gajawelli, Niharika, Sean Deoni, Jie Shi, Liang Xu, Holly Dirks, Douglas Dean, Jonathan
O'Muircheartaigh et al. "Changes in neurocranium thickness in early childhood." In 11th
International Symposium on Medical Information Processing and Analysis (SIPAIM 2015),
pp. 96810I-96810I. International Society for Optics and Photonics, 2015.
3. Gajawelli, Niharika, Sean Deoni, Holly Dirks, Douglas Dean, Jonathan O'Muircheartaigh,
Siddhant Sawardekar, Andrea Ezis et al. "Characterization of the central sulcus in the brain
in early childhood." In Engineering in Medicine and Biology Society (EMBC), 2015 37th
Annual International Conference of the IEEE, pp. 149-152. IEEE, 2015.
4. James, C., Lepore, F., Collignon, O., Gajawelli, N., Lepore, N., & Coulon, O. (2017, July).
Central sulcus depth and sulcal profile differences between congenitally blind and sighted
individuals. In Engineering in Medicine and Biology Society (EMBC), 2017 39th Annual
International Conference of the IEEE (pp. 3008-3011). IEEE.
5. N. Gajawelli, S. Tsao, M. Yip, M. Law, N. Lepore ICA-based Multi-Fiber Tractography
for Brain Tumor Surgical Planning MICCAI 2014 DTI Challenge
6. S. Tsao*, N. Gajawelli*, P.A. Michel, D. Hwang, Y. Lao, F. Yepes, V. Rajagopalan, M.
Law, N. Lepore ICA-based Multi-Fiber DWI Tractography in Neurosurgical Planning.
MICCAI 2013 DTI Challenge (* equal contribution)
7. Y. Lao, M. Law, J. Shi, N. Gajawelli, L. Haas, Y. Wang, and N. Leporé. "A T1 and DTI
fused 3D corpus callosum analysis in pre-vs. post-season contact sports players." In
SPIE,2015.
8. D. H. Hwang, S. Tsao, N. Gajawelli, M. Law, and N. Lepore. "TractRender: a new
generalized 3D medical image visualization and output platform." SPIE, 2015
63

9. Y. Lao, N. Gajawelli, L. Hass, B. Wilkins, D. Hwang, S. Tsao, Y. Wang, M. Law, and N.
Leporé. "3D pre-versus post-season comparisons of surface and relative pose of the corpus
callosum in contact sport athletes." SPIE 2014
10. S. Tsao, S. J. Ma, P. A. Michels, N. Gajawelli, M. Law, H. Chui, and N. Lepore. "The
power of hybrid/fusion imaging metrics in future PACS systems: a case study into the
white matter hyperintensity prenumbra using FLAIR and diffusion MR." In SPIE 2014
11. S. Tsao, N. Gajawelli, J. Zhou, J. Shi, J. Ye, Y. Wang, and N. Lepore. "Evaluating the
predictive power of multivariate tensor-based morphometry in Alzheimer's disease
progression via convex fused sparse group Lasso." SPIE 2014
12. S. Tsao, D. Hwang, N. Gajawelli, B. Wilkins, M. Law, H.C. Chui, " Mapping of ApoE- 4
Related White Matter Damage using Diffusion MRI, Southern California Alzheimer’s
Disease Centers 2013 Research Symposium, Irvine, CA, USA, 2013
Abstracts
1. N. Gajawelli, B Voytek, D. Wang , B. Zlokovic , A.W. Toga , M. Law , J. Morris , T.
Benzinger, J. Pa, “Power law exponent analysis of the resting state BOLD signal as a
potential measure of excitatory-inhibitory balance in Alzheimer’s disease”, ISMRM, 2017,
Hawaii
2. N. Gajawelli, S. Taso, D. Hwang, B. Wilkins, S. Kriger, M. Singh, ”Segmentation of brain
volumes with cerebral infarcts using FreeSurfer,” ISMRM, 2012, Melbourne
3. N. Gajawelli, M. Singh, B. Wilkins, "Hippocampal subfield ICA multifiber tractography
using 3T clinical diffusion data," ISMRM, 2012, Melbourne

64

References
[1] Neurological, Psychiatric, and Developmental Disorders., National Academies Press, Washington,
D.C. (2001).
[2] Boyle, C. A., Boulet, S., Schieve, L. A., Cohen, R. A., Blumberg, S. J., Yeargin-Allsopp, M., Visser,
S. and Kogan, M. D., “Trends in the prevalence of developmental disabilities in US children, 1997-
2008.,” Pediatrics 127(6), 1034–1042 (2011).
[3] Evans, A. C., “The NIH MRI study of normal brain development.,” Neuroimage 30(1), 184–202
(2006).
[4] Shi, F., Yap, P.-T., Wu, G., Jia, H., Gilmore, J. H., Lin, W. and Shen, D., “Infant Brain Atlases from
Neonates to 1- and 2-Year-Olds,” PLoS One 6(4), H. Okazawa, Ed., e18746 (2011).
[5] McArdle, C. B., Richardson, C. J., Nicholas, D. A., Mirfakhraee, M., Hayden, C. K. and Amparo, E.
G., “Developmental features of the neonatal brain: MR imaging. Part I. Gray-white matter
differentiation and myelination.,” Radiology 162(1 Pt 1), 223–229 (1987).
[6] Holland, B. A., Haas, D. K., Norman, D., Brant-Zawadzki, M. and Newton, T. H., “MRI of normal
brain maturation.,” AJNR. Am. J. Neuroradiol. 7(2), 201–208 (1986).
[7] Paus, T., Collins, D. ., Evans, A. ., Leonard, G., Pike, B. and Zijdenbos, A., “Maturation of white
matter in the human brain: a review of magnetic resonance studies,” Brain Res. Bull. 54(3), 255–266
(2001).
[8] White, L., Andrews, T. J., Hulette, C., Richards, A., Groelle, M., Paydarfar, J. and Purves, D.,
“Structure of the human sensorimotor system. I: Morphology and cytoarchitecture of the central
sulcus,” Cereb. Cortex 7(1), 18–30 (1997).
[9] Gilles, F. H. and Nelson, M. D., [The Developing Human Brain : Growth and Adversities], Mac
Keith Press (2012).
[10] Hopkins, W. D., Coulon, O. and Mangin, J.-F., “Observer-independent characterization of sulcal
landmarks and depth asymmetry in the central sulcus of the chimpanzee brain,” Neuroscience 171(2),
544–551 (2010).
[11] Li, S., Wang, S., Li, X., Li, Q. and Li, X., “Abnormal surface morphology of the central sulcus in
children with attention-deficit/hyperactivity disorder,” Front. Neuroanat. 9, 114 (2015).
[12] Jackowski, A. P. and Schultz, R. T., “Foreshortened Dorsal Extension of the Central Sulcus in
Williams Syndrome,” Cortex 41(3), 282–290 (2005).
[13] Richtsmeier, J. T. and Flaherty, K., “Hand in glove: brain and skull in development and
dysmorphogenesis.”
[14] Shim, K.-W., Park, E.-K., Kim, J.-S., Kim, Y.-O. and Kim, D.-S., “Neurodevelopmental Problems
in Non-Syndromic Craniosynostosis.,” J. Korean Neurosurg. Soc. 59(3), 242–246 (2016).
[15] Da Costa, A. C., Anderson, V. A., Savarirayan, R., Wrennall, J. A., Chong, D. K., Holmes, A. D.,
65

Greensmith, A. L. and Meara, J. G., “Neurodevelopmental functioning of infants with untreated
single-suture craniosynostosis during early infancy,” Child’s Nerv. Syst. 28(6), 869–877 (2012).
[16] Rutherford, M. A., [MRI of the neonatal brain], W.B. Saunders (2002).
[17] Casanova, M. F., El-Baz, A., Elnakib, A., Switala, A. E., Williams, E. L., Williams, D. L., Minshew,
N. J. and Conturo, T. E., “Quantitative analysis of the shape of the corpus callosum in patients with
autism and comparison individuals,” Autism 15(2), 223–238 (2011).
[18] Lao, Y., Nguyen, B., Tsao, S., Gajawelli, N., Law, M., Chui, H., Weiner, M., Wang, Y. and Leporé,
N., “A T1 and DTI fused 3D corpus callosum analysis in MCI subjects with high and low
cardiovascular risk profile,” NeuroImage Clin. 14, 298–307 (2017).
[19] Lao, Y., Gajawelli, N., Haas, L., Wilkins, B., Hwang, D., Tsao, S., Wang, Y., Law, M. and Leporé,
N., “3D pre-versus post-season comparisons of surface and relative pose of the corpus callosum in
contact sport athletes,” Prog. Biomed. Opt. Imaging - Proc. SPIE 9034 (2014).
[20] Luders, E., Thompson, P. M. and Toga, A. W., “The development of the corpus callosum in the
healthy human brain.,” J. Neurosci. 30(33), 10985–10990 (2010).
[21] Giedd, J. N., Rumsey, J. M., Castellanos, F. X., Rajapakse, J. C., Kaysen, D., Vaituzis, A. C., Vauss,
Y. C., Hamburger, S. D. and Rapoport, J. L., “A quantitative MRI study of the corpus callosum in
children and adolescents.,” Brain Res. Dev. Brain Res. 91(2), 274–280 (1996).
[22] Thompson, P. M., Giedd, J. N., Woods, R. P., MacDonald, D., Evans, A. C. and Toga, A. W.,
“Growth patterns in the developing brain detected by using continuum mechanical tensor maps,”
Nature 404(6774), 190–193 (2000).
[23] Li, Z., Park, B.-K., Liu, W., Zhang, J., Reed, M. P., Rupp, J. D., Hoff, C. N. and Hu, J., “A Statistical
Skull Geometry Model for Children 0-3 Years Old,” PLoS One 10(5), J. Cray Jr., Ed., e0127322
(2015).
[24] Libby, J., Marghoub, A., Johnson, D., Khonsari, R. H., Fagan, M. J. and Moazen, M., “Modelling
human skull growth: a validated computational model.,” J. R. Soc. Interface 14(130), 20170202
(2017).
[25] Sperber, G. H., Sperber, S. M., Guttmann, G. D. and Sperber, G. H., [Craniofacial embryogenetics
and development], People’s Medical Pub. House USA (2010).
[26] Lieberman, D., [The evolution of the human head], Belknap Press of Harvard University Press (2011).
[27] King, T. C., [Elsevier’s integrated pathology], Mosby, St. Louis : (2007).
[28] Kandel, E. R., Schwartz, J. H. (James H. and Jessell, T. M., [Principles of neural science], McGraw-
Hill, Health Professions Division (2000).
[29] Holland, D., Chang, L., Ernst, T. M., Curran, M., Buchthal, S. D., Alicata, D., Skranes, J., Johansen,
H., Hernandez, A., Yamakawa, R., Kuperman, J. M. and Dale, A. M., “Structural growth trajectories
and rates of change in the first 3 months of infant brain development.,” JAMA Neurol. 71(10), 1266–
1274 (2014).
66

[30] Marsh, R., Gerber, A. J. and Peterson, B. S., “Neuroimaging Studies of Normal Brain Development
and Their Relevance for Understanding Childhood Neuropsychiatric Disorders,” J. Am. Acad. Child
Adolesc. Psychiatry 47(11), 1233–1251 (2008).
[31] Lenroot, R. K. and Giedd, J. N., “Brain development in children and adolescents: Insights from
anatomical magnetic resonance imaging,” Neurosci. Biobehav. Rev. 30(6), 718–729 (2006).
[32] Paus, T., Zijdenbos, A., Worsley, K., Collins, D. L., Blumenthal, J., Giedd, J. N., Rapoport, J. L. and
Evans, A. C., “Structural maturation of neural pathways in children and adolescents: in vivo study.,”
Science 283(5409), 1908–1911 (1999).
[33] Giedd, J. N. and Rapoport, J. L., “Structural MRI of Pediatric Brain Development: What Have We
Learned and Where Are We Going?,” Neuron 67(5), 728–734 (2010).
[34] Gogtay, N., Giedd, J. N., Lusk, L., Hayashi, K. M., Greenstein, D., Vaituzis, A. C., Nugent, T. F.,
Herman, D. H., Clasen, L. S., Toga, A. W., Rapoport, J. L. and Thompson, P. M., “Dynamic mapping
of human cortical development during childhood through early adulthood.,” Proc. Natl. Acad. Sci.
U. S. A. 101(21), 8174–8179 (2004).
[35] Li, G., Nie, J., Wang, L., Shi, F., Gilmore, J. H., Lin, W. and Shen, D., “Measuring the dynamic
longitudinal cortex development in infants by reconstruction of temporally consistent cortical
surfaces,” Neuroimage 90, 266–279 (2014).
[36] Li, G., Nie, J., Wang, L., Shi, F., Lin, W., Gilmore, J. H. and Shen, D., “Mapping Region-Specific
Longitudinal Cortical Surface Expansion from Birth to 2 Years of Age,” Cereb. Cortex 23(11), 2724–
2733 (2013).
[37] Giedd, J. N., Blumenthal, J., Jeffries, N. O., Castellanos, F. X., Liu, H., Zijdenbos, A., Paus, T., Evans,
A. C. and Rapoport, J. L., “Brain development during childhood and adolescence: a longitudinal MRI
study.,” Nat. Neurosci. 2(10), 861–863 (1999).
[38] Dean, D. C., O’Muircheartaigh, J., Dirks, H., Waskiewicz, N., Lehman, K., Walker, L., Han, M.,
Deoni, S. C. L. and Deoni, S. C. L., “Modeling healthy male white matter and myelin development:
3 through 60months of age.,” Neuroimage 84, 742–752 (2014).
[39] Knickmeyer, R. C., Gouttard, S., Kang, C., Evans, D., Wilber, K., Smith, J. K., Hamer, R. M., Lin,
W., Gerig, G. and Gilmore, J. H., “A structural MRI study of human brain development from birth
to 2 years.,” J. Neurosci. 28(47), 12176–12182 (2008).
[40] Coch, D., Dawson, G. and Fischer, K. W., [Human Behavior, Learning, and the Developing Brain :
Typical Development], Guilford Publications, New York, UNITED STATES (2014).
[41] Klyachko, V. A. and Stevens, C. F., “Connectivity optimization and the positioning of cortical areas”
(2003).
[42] Richman, D. P., Stewart, R. M., Hutchinson, J. W. and Caviness, V. S., “Mechanical model of brain
convolutional development.,” Science 189(4196), 18–21 (1975).
[43] Fernández, V., Llinares-Benadero, C. and Borrell, V., “Cerebral cortex expansion and folding: what
have we learned?,” EMBO J. 35(10), 1021–1044 (2016).
67

[44] Van Essen, D. C., “A tension-based theory of morphogenesis and compact wiring in the central
nervous system.,” Nature 385(6614), 313–318 (1997).
[45] Cykowski, M. D., Coulon, O., Kochunov, P. V., Amunts, K., Lancaster, J. L., Laird, A. R., Glahn,
D. C. and Fox, P. T., “The Central Sulcus: an Observer-Independent Characterization of Sulcal
Landmarks and Depth Asymmetry,” Cereb. Cortex 18(9), 1999–2009 (2008).
[46] Boling, W. W. and Olivier, A., “Localization of hand sensory function to the pli de passage moyen
of Broca” (2004).
[47] Boling, W., Olivier, A., Bittar, R. G. and Reutens, D., “Localization of hand motor activation in
Broca’s pli de passage moyen,” J. Neurosurg. 91(6), 903–910 (1999).
[48] Boling, W., Parsons, M., Kraszpulski, M., Cantrell, C. and Puce, A., “Whole-hand sensorimotor area:
cortical stimulation localization and correlation with functional magnetic resonance imaging,” J.
Neurosurg. 108(3), 491–500 (2008).
[49] PENFIELD, W. and BOLDREY, E., “SOMATIC MOTOR AND SENSORY REPRESENTATION
IN THE CEREBRAL CORTEX OF MAN AS STUDIED BY ELECTRICAL STIMULATION,”
Brain 60(4), 389–443 (1937).
[50] Penfield, W. and Rasmussen, T., [The cerebral cortex of man; a clinical study of localization of
function.], Macmillan, Oxford, England (1950).
[51] Stiles, J. and Jernigan, T. L., “The basics of brain development.,” Neuropsychol. Rev. 20(4), 327–
348 (2010).
[52] Gray, H., [Anatomy of the Human Body, 20thed.], Lea & Febiger, New York; USA; (1918).
[53] Cunningham, M. L. and Heike, C. L., “Evaluation of the infant with an abnormal skull shape,” Curr.
Opin. Pediatr. 19(6), 645–651 (2007).
[54] Ridgway, E. B. and Weiner, H. L., “Skull deformities.,” Pediatr. Clin. North Am. 51(2), 359–387
(2004).
[55] Nikitina, N. V. and Bronner-Fraser, M., “Gene regulatory networks that control the specification of
neural-crest cells in the lamprey,” Biochim. Biophys. Acta - Gene Regul. Mech. 1789(4), 274–278
(2009).
[56] Lathia, J. D., Mattson, M. P. and Cheng, A., “Notch: from neural development to neurological
disorders,” J. Neurochem. 107(6), 1471–1481 (2008).
[57] Long, F., “Building strong bones: molecular regulation of the osteoblast lineage,” Nat. Rev. Mol.
Cell Biol. 13(1), 27–38 (2011).
[58] Henderson, J. H., Chang, L. Y., Song, H. M., Longaker, M. T. and Carter, D. R., “Age-dependent
properties and quasi-static strain in the rat sagittal suture,” J. Biomech. 38(11), 2294–2301 (2005).
[59] Mao, J. J. and Nah, H.-D., “Growth and development: hereditary and mechanical modulations.,” Am.
J. Orthod. Dentofacial Orthop. 125(6), 676–689 (2004).
68

[60] Pujol, J., Vendrell, P., Junqué, C., Martí-Vilalta, J. L. and Capdevila, A., “When does human brain
development end? Evidence of corpus callosum growth up to adulthood.,” Ann. Neurol. 34(1), 71–
75 (1993).
[61] Tanaka-Arakawa, M. M., Matsui, M., Tanaka, C., Uematsu, A., Uda, S., Miura, K., Sakai, T. and
Noguchi, K., “Developmental changes in the corpus callosum from infancy to early adulthood: a
structural magnetic resonance imaging study.,” PLoS One 10(3), e0118760 (2015).
[62] Shaw, P., Greenstein, D., Lerch, J., Clasen, L., Lenroot, R., Gogtay, N., Evans, A., Rapoport, J. and
Giedd, J., “Intellectual ability and cortical development in children and adolescents.”
[63] Giedd, J. N., “The teen brain: insights from neuroimaging.,” J. Adolesc. Health 42(4), 335–343
(2008).
[64] Sowell, E. R., Peterson, B. S., Thompson, P. M., Welcome, S. E., Henkenius, A. L. and Toga, A. W.,
“Mapping cortical change across the human life span,” Nat. Neurosci. 6(3), 309–315 (2003).
[65] Raznahan, A., Shaw, P., Lalonde, F., Stockman, M., Wallace, G. L., Greenstein, D., Clasen, L.,
Gogtay, N. and Giedd, J. N., “How does your cortex grow?,” J. Neurosci. 31(19), 7174–7177 (2011).
[66] Remer, J., Croteau-Chonka, E., Dean, D. C., D’Arpino, S., Dirks, H., Whiley, D. and Deoni, S. C.
L., “Quantifying cortical development in typically developing toddlers and young children, 1–6 years
of age,” Neuroimage 153, 246–261 (2017).
[67] Andersen, S. K., Jakobsen, C. E., Pedersen, C. H., Rasmussen, A. M., Plocharski, M. and Østergaard,
L. R., “Classification of Alzheimer’s Disease from MRI Using Sulcal Morphology.”
[68] Leroy, F., Cai, Q., Bogart, S. L., Dubois, J., Coulon, O., Monzalvo, K., Fischer, C., Glasel, H., Van
der Haegen, L., Bénézit, A., Lin, C.-P., Kennedy, D. N., Ihara, A. S., Hertz-Pannier, L., Moutard,
M.-L., Poupon, C., Brysbaert, M., Roberts, N., Hopkins, W. D., et al., “New human-specific brain
landmark: the depth asymmetry of superior temporal sulcus.,” Proc. Natl. Acad. Sci. U. S. A. 112(4),
1208–1213 (2015).
[69] Hill, J., Dierker, D., Neil, J., Inder, T., Knutsen, A., Harwell, J., Coalson, T. and Van Essen, D., “A
Surface-Based Analysis of Hemispheric Asymmetries and Folding of Cerebral Cortex in Term-Born
Human Infants,” J. Neurosci. 30(6) (2010).
[70] Churchwell, J. C. and Yurgelun-Todd, D. A., “Developmental Cognitive Neuroscience Age-related
changes in insula cortical thickness and impulsivity: Significance for emotional development and
decision-making,” Accid. Anal. Prev. 6, 80–86 (2013).
[71] Galaburda, A. M., Schmitt, J. E., Atlas, S. W., Eliez, S., Bellugi, U. and Reiss, A. L., “Dorsal
Forebrain Anomaly in Williams Syndrome,” Arch. Neurol. 58(11), 1865 (2001).
[72] Li, S., Han, Y., Wang, D., Yang, H., Fan, Y., Lv, Y., Tang, H., Gong, Q., Zang, Y. and He, Y.,
“Mapping Surface Variability of the Central Sulcus in Musicians,” Cereb. Cortex 20(1), 25–33
(2010).
[73] Amunts, K., Jäncke, L., Mohlberg, H., Steinmetz, H. and Zilles, K., “Interhemispheric asymmetry of
the human motor cortex related to handedness and gender,” Neuropsychologia 38(3), 304–312 (2000).
69

[74] Amunts, K., Schlaug, G., Schleicher, A., Steinmetz, H., Dabringhaus, A., Roland, P. E. and Zilles,
K., “Asymmetry in the Human Motor Cortex and Handedness,” Neuroimage 4(3), 216–222 (1996).
[75] McKay, D. R., Kochunov, P., Cykowski, M. D., Kent, J. W., Laird, A. R., Lancaster, J. L., Blangero,
J., Glahn, D. C. and Fox, P. T., “Sulcal depth-position profile is a genetically mediated neuroscientific
trait: description and characterization in the central sulcus.,” J. Neurosci. 33(39), 15618–15625
(2013).
[76] Deoni, S. C. L., Rutt, B. K. and Peters, T. M., “Rapid combined T1 and T2 mapping using gradient
recalled acquisition in the steady state.,” Magn. Reson. Med. 49(3), 515–526 (2003).
[77] Deoni, S. C. L., Rutt, B. K. and Peters, T. M., “Synthetic T1-weighted brain image generation with
incorporated coil intensity correction using DESPOT1,” Magn. Reson. Imaging 24(9), 1241–1248
(2006).
[78] Shattuck, D. W. and Leahy, R. M., “BrainSuite: an automated cortical surface identification tool.,”
Med. Image Anal. 6(2), 129–142 (2002).
[79] Jenkinson, M. and Smith, S., “A global optimisation method for robust affine registration of brain
images.,” Med. Image Anal. 5(2), 143–156 (2001).
[80] Jenkinson, M., Bannister, P., Brady, M. and Smith, S., “Improved optimization for the robust and
accurate linear registration and motion correction of brain images.,” Neuroimage 17(2), 825–841
(2002).
[81] Gaonkar, B., Macyszyn, L., Bilello, M., Sadaghiani, M. S., Akbari, H., Attiah, M. A., Ali, Z. S., Da,
X., Zhan, Y., Rourke, D. O., Grady, S. M. and Davatzikos, C., “Automated tumor volumetry using
computer-aided image segmentation.,” Acad. Radiol. 22(5), 653–661 (2015).
[82] Fischer, C., Operto, G., Laguitton, S., Perrot, M., Denghien, I., Riviere, D. and Mangin, J.-F.,
“Morphologist 2012: the new morphological pipeline of BrainVISA,” Proc. HBM (2012).
[83] Perrot, M., Rivière, D. and Mangin, J.-F., “Cortical sulci recognition and spatial normalization,” Med.
Image Anal. 15(4), 529–550 (2011).
[84] Gaonkar, B. and Shu, L., “Adaptive geodesic transform for segmentation of vertebrae on CT images,”
… Imaging 9035, 903516 (2014).
[85] Coulon, O., Lefevre, J., Kloppel, S., Siebner, H. and Mangin, J.-F., “Quasi-isometric length
parameterization of cortical sulci: Application to handedness and the central sulcus morphology,”
2015 IEEE 12th Int. Symp. Biomed. Imaging, 1268–1271, IEEE (2015).
[86] Coulon, O., Pizzagalli, F., Operto, G., Auzias, G., Delon-Martin, C. and Dojat, M., “Two new stable
anatomical landmarks on the Central Sulcus: Definition, automatic detection, and their relationship
with primary motor functions of the hand,” 2011 Annu. Int. Conf. IEEE Eng. Med. Biol. Soc. 2011,
7795–7798, IEEE (2011).
[87] Li, G., Wang, L., Shi, F., Lyall, A. E., Lin, W., Gilmore, J. H. and Shen, D., “Mapping longitudinal
development of local cortical gyrification in infants from birth to 2 years of age.,” J. Neurosci. 34(12),
4228–4238 (2014).
70

[88] Boling, W. and Olivier, A., “Anatomy of Important Functioning Cortex,” [Functional Mapping of
the Cerebral Cortex], Springer International Publishing, Cham, 23–40 (2016).
[89] Yousry, T., Schmid, U. D., Alkadhi, H., Schmidt, D., Peraud, A., Buettner, A. and Winkler, P.,
“Localization of the motor hand area to a knob on the precentral gyrus. A new landmark,” Brain
120(1), 141–157 (1997).
[90] Sun, Z. Y., Klöppel, S., Rivière, D., Perrot, M., Frackowiak, R., Siebner, H. and Mangin, J.-F., “The
effect of handedness on the shape of the central sulcus.,” Neuroimage 60(1), 332–339 (2012).
[91] White, L. E., Lucas, G., Richards, A. and Purves, D., “Cerebral asymmetry and handedness,” Nature
368(6468), 197–198 (1994).
[92] Kim, S. G., Ashe, J., Hendrich, K., Ellermann, J. M., Merkle, H., Uğurbil, K. and Georgopoulos, A.
P., “Functional magnetic resonance imaging of motor cortex: hemispheric asymmetry and
handedness.,” Science 261(5121), 615–617 (1993).
[93] Hopkins, W. D., Coulon, O., Meguerditchian, A., Autrey, M., Davidek, K., Mahovetz, L., Pope, S.,
Mareno, M. C. and Schapiro, S. J., “Genetic Factors and Orofacial Motor Learning Selectively
Influence Variability in Central Sulcus Morphology in Chimpanzees (Pan troglodytes).,” J. Neurosci.
37(22), 5475–5483 (2017).
[94] Fischl, B., “FreeSurfer.,” Neuroimage 62(2), 774–781 (2012).
[95] Tokimura, H., Asakura, T., Tokimura, Y., Oliviero, A. and Rothwell, J. C., “Speech-induced changes
in corticospinal excitability,” Ann. Neurol. 40(4), 628–634 (1996).
[96] Meister, I. G., Boroojerdi, B., Foltys, H., Sparing, R., Huber, W. and Töpper, R., “Motor cortex hand
area and speech: implications for the development of language,” Neuropsychologia 41(4), 401–406
(2003).
[97] Seyal, M., Mull, B., Bhullar, N., Ahmad, T. and Gage, B., “Anticipation and execution of a simple
reading task enhance corticospinal excitability,” Clin. Neurophysiol. 110(3), 424–429 (1999).
[98] Iverson, J. M. and Goldin-Meadow, S., “Why people gesture when they speak,” Nature 396(6708),
228–228 (1998).
[99] Collett, B. R., Starr, J. R., Kartin, D., Heike, C. L., Berg, J., Cunningham, M. L. and Speltz, M. L.,
“Development in Toddlers With and Without Deformational Plagiocephaly,” Arch. Pediatr. Adolesc.
Med. 165(7), 653 (2011).
[100] Robinson, S. and Proctor, M., “Diagnosis and management of deformational plagiocephaly,” J.
Neurosurg. Pediatr. 3(4), 284–295 (2009).
[101] Lillie, E. M., Urban, J. E., Lynch, S. K., Weaver, A. A. and Stitzel, J. D., “Evaluation of Skull Cortical
Thickness Changes With Age and Sex From Computed Tomography Scans,” J. Bone Miner. Res.
31(2), 299–307 (2016).
[102] Bastir, M., Rosas, A. and O’higgins, P., “Craniofacial levels and the morphological maturation of the
human skull.,” J. Anat. 209(5), 637–654 (2006).
71

[103] Scarr, G., “A model of the cranial vault as a tensegrity structure, and its significance to normal and
abnormal cranial development,” Int. J. Osteopath. Med. 11(3), 80–89 (2008).
[104] Warschausky, S., Angobaldo, J., Kewman, D., Buchman, S., Muraszko, K. M. and Azengart, A.,
“Early development of infants with untreated metopic craniosynostosis.,” Plast. Reconstr. Surg.
115(6), 1518–1523 (2005).
[105] Chieffo, D., Tamburrini, G., Massimi, L., Di Giovanni, S., Giansanti, C., Caldarelli, M. and Di Rocco,
C., “Long-term neuropsychological development in single-suture craniosynostosis treated early,” J.
Neurosurg. Pediatr. 5(3), 232–237 (2010).
[106] Faul, M., Xu, L., Wald, M. M., Coronado, V. and Dellinger, A. M., “Traumatic brain injury in the
United States: national estimates of prevalence and incidence, 2002-2006,” Inj. Prev. 16(Supplement
1), A268–A268 (2010).
[107] Coats, B. and Margulies, S. S., “Material properties of human infant skull and suture at high rates.,”
J. Neurotrauma 23(8), 1222–1232 (2006).
[108] de Gonzalez, A. B., Salotti, J. A., McHugh, K., Little, M. P., Harbron, R. W., Lee, C., Ntowe, E.,
Braganza, M. Z., Parker, L., Rajaraman, P., Stiller, C., Stewart, D. R., Craft, A. W. and Pearce, M.
S., “Relationship between paediatric CT scans and subsequent risk of leukaemia and brain tumours:
assessment of the impact of underlying conditions,” Br. J. Cancer 114(4), 388–394 (2016).
[109] Pearce, M. S., Salotti, J. A., Little, M. P., McHugh, K., Lee, C., Kim, K. P., Howe, N. L., Ronckers,
C. M., Rajaraman, P., Sir Craft, A. W., Parker, L. and Berrington de González, A., “Radiation
exposure from CT scans in childhood and subsequent risk of leukaemia and brain tumours: a
retrospective cohort study.,” Lancet (London, England) 380(9840), 499–505 (2012).
[110] Eley, K. A., Watt-Smith, S. R., Sheerin, F. and Golding, S. J., “‘Black Bone’ MRI: a potential
alternative to CT with three-dimensional reconstruction of the craniofacial skeleton in the diagnosis
of craniosynostosis,” Eur. Radiol. 24(10), 2417–2426 (2014).
[111] Fjørtoft, M. I., Sevely, A., Boetto, S., Kessler, S., Sarramon, M. F. and Rolland, M., “Prenatal
diagnosis of craniosynostosis: value of MR imaging,” Neuroradiology 49(6), 515–521 (2007).
[112] Chauveau, N., Franceries, X., Doyon, B., Rigaud, B., Morucci, J. P. and Celsis, P., “Effects of skull
thickness, anisotropy, and inhomogeneity on forward EEG/ERP computations using a spherical
three-dimensional resistor mesh model.,” Hum. Brain Mapp. 21(2), 86–97 (2004).
[113] Cuffin, B. N., “Effects of local variations in skull and scalp thickness on EEG’s and MEG’s,” IEEE
Trans. Biomed. Eng. 40(1), 42–48 (1993).
[114] Tustison, N. J., Avants, B. B., Cook, P. A., Egan, A., Yushkevich, P. A. and Gee, J. C., “N4ITK:
Improved N3 Bias Correction,” IEEE Trans. Med. Imaging 29(6), 1310–1320 (2010).
[115] Smith, S. M., “Fast robust automated brain extraction.,” Hum. Brain Mapp. 17(3), 143–155 (2002).
[116] Jenkinson, M., Pechaud, M. and Smith, S., “BET2 -MR-Based Estimation of Brain, Skull and Scalp
Surfaces,” Elev. Annu. Meet. Organ. Hum. Brain Mapping, 2005.
[117] Xu, L., Collignon, O., Wang, G., Kang, Y., Leporé, F., Shi, J., Lao, Y., Joshi, A., Leporé, N. and
72

Wang, Y., “Combining Thickness Information with Surface Tensor-based Morphometry for the 3D
Statistical Analysis of the Corpus Callosum” (2013).
[118] Wang, Y., Song, Y., Chou, Y.-Y., Toga, A. W. and Thompson, P. M., “Hippocampal and Ventricular
Differences in 804 ADNI subjects mapped with Multivariate Tensor-Based Morphometry.”
[119] Boas, D. A., “Tetrahedral mesh generation from volumetric binary and grayscale images,” 2009
IEEE Int. Symp. Biomed. Imaging From Nano to Macro, 1142–1145, IEEE (2009).
[120] Shi, J., Collignon, O., Xu, L., Wang, G., Kang, Y., Leporé, F., Lao, Y., Joshi, A. A., Leporé, N. and
Wang, Y., “Impact of Early and Late Visual Deprivation on the Structure of the Corpus Callosum:
A Study Combining Thickness Profile with Surface Tensor-Based Morphometry.,” Neuroinformatics
13(3), 321–336 (2015).
[121] Adamson, C. L., Wood, A. G., Chen, J., Barton, S., Reutens, D. C., Pantelis, C., Velakoulis, D. and
Walterfang, M., “Thickness profile generation for the corpus callosum using Laplace’s equation.,”
Hum. Brain Mapp. 32(12), 2131–2140 (2011).
[122] Jones, S. E., Buchbinder, B. R. and Aharon, I., “Three-dimensional mapping of cortical thickness
using Laplace’s equation.,” Hum. Brain Mapp. 11(1), 12–32 (2000).
[123] Chung, M. K., Dalton, K. M. and Davidson, R. J., “Tensor-based cortical surface morphometry via
weighted spherical harmonic representation.,” IEEE Trans. Med. Imaging 27(8), 1143–1151 (2008).
[124] Chung, M. K., [Computational Neuroanatomy: The Methods], World Scientific (2012).
[125] Meyer-Marcotty, P., Böhm, H., Linz, C., Kochel, J., Stellzig-Eisenhauer, A. and Schweitzer, T.,
“Three-dimensional analysis of cranial growth from 6 to 12 months of age.,” Eur. J. Orthod. 36(5),
489–496 (2014).
[126] de Lacoste, M. C., Kirkpatrick, J. B. and Ross, E. D., “Topography of the human corpus callosum.,”
J. Neuropathol. Exp. Neurol. 44(6), 578–591 (1985).
[127] Gilmore, J. H., Lin, W., Corouge, I., Vetsa, Y. S. K., Smith, J. K., Kang, C., Gu, H., Hamer, R. M.,
Lieberman, J. A. and Gerig, G., “Early postnatal development of corpus callosum and corticospinal
white matter assessed with quantitative tractography.,” AJNR. Am. J. Neuroradiol. 28(9), 1789–1795
(2007).
[128] Matsuzawa, J., “Age-related Volumetric Changes of Brain Gray and White Matter in Healthy Infants
and Children,” Cereb. Cortex 11(4), 335–342 (2001).
[129] Chura, L. R., Lombardo, M. V, Ashwin, E., Auyeung, B., Chakrabarti, B., Bullmore, E. T. and Baron-
Cohen, S., “Organizational effects of fetal testosterone on human corpus callosum size and
asymmetry.,” Psychoneuroendocrinology 35(1), 122–132 (2010).
[130] Giedd, J. N., Blumenthal, J., Jeffries, N. O., Rajapakse, J. C., Vaituzis, A. C., Liu, H., Berry, Y. C.,
Tobin, M., Nelson, J. and Castellanos, F. X., “Development of the human corpus callosum during
childhood and adolescence: a longitudinal MRI study.,” Prog. Neuropsychopharmacol. Biol.
Psychiatry 23(4), 571–588 (1999).
[131] Salat, D., Ward, A., Kaye, J. A. and Janowsky, J. S., “Sex differences in the corpus callosum with
73

aging.,” Neurobiol. Aging 18(2), 191–197.
[132] Hofer, S. and Frahm, J., “Topography of the human corpus callosum revisited—Comprehensive fiber
tractography using diffusion tensor magnetic resonance imaging,” Neuroimage 32(3), 989–994
(2006).
[133] Deoni, S. C. L., Dean, D. C., O’Muircheartaigh, J., Dirks, H., Jerskey, B. A. and Jerskey, B. A.,
“Investigating white matter development in infancy and early childhood using myelin water faction
and relaxation time mapping.,” Neuroimage 63(3), 1038–1053 (2012).
[134] Yushkevich, P. A., Piven, J., Cody Hazlett, H., Gimpel Smith, R., Ho, S., Gee, J. C. and Gerig, G.,
“User-Guided {3D} Active Contour Segmentation of Anatomical Structures: Significantly Improved
Efficiency and Reliability,” Neuroimage 31(3), 1116–1128 (2006).
[135] Lederman, C., Joshi, A., Dinov, I., Vese, L., Toga, A. and Van Horn, J. D., “The generation of
tetrahedral mesh models for neuroanatomical MRI.,” Neuroimage 55(1), 153–164 (2011).
[136] Styner, M., Lieberman, J. A., McClure, R. K., Weinberger, D. R., Jones, D. W. and Gerig, G.,
“Morphometric analysis of lateral ventricles in schizophrenia and healthy controls regarding genetic
and disease-specific factors.,” Proc. Natl. Acad. Sci. U. S. A. 102(13), 4872–4877 (2005).
[137] Nitzken, M. J., Casanova, M. F., Gimel, G., Inanc, T., Zurada, J. M. and El-Baz, A., “Shape Analysis
of Human Brain: A Brief Survey,” IEEE Trans. Inf. Technol. Biomed. (2013).
[138] Wang, Y., Panigrahy, A., Shi, J., Ceschin, R., Nelson, M. D., Thompson, P. M. and Lepor, N.,
“Subcortical Structures Morphometry in Premature Neonates: A Parametric Surface-based
Approach,” Proc. 17th Annu. Meet. Organ. Hum. Brain Mapp. (2011).
[139] Lepore, N., Brun, C., Chou, Y.-Y., Chiang, M.-C., Dutton, R. A., Hayashi, K. M., Luders, E., Lopez,
O. L., Aizenstein, H. J., Toga, A. W., Becker, J. T. and Thompson, P. M., “Generalized Tensor-Based
Morphometry of HIV/AIDS Using Multivariate Statistics on Deformation Tensors,” IEEE Trans.
Med. Imaging 27(1) (2008).
[140] Wang Phd, Y., Panigrahy Md, A., Shi, J., Ceschin, R., Nelson, M. D., Gutman, B., Thompson Phd,
P. M. and Leporé, N., “Surface Multivariate Tensor-based Morphometry on Premature Neonates: A
Pilot Study.”
[141] Wang, Y., Song, Y., Rajagopalan, P., An, T., Liu, K., Chou, Y.-Y., Gutman, B., Toga, A. W. and
Thompson, P. M., “Surface-based TBM boosts power to detect disease effects on the brain: an N=804
ADNI study.,” Neuroimage 56(4), 1993–2010 (2011).
[142] Leporé, N., Brun, C., Chou, Y.-Y., Lee, A., Barysheva, M., De Zubicaray, G. I., Meredith, M.,
Macmahon, K., Wright, M. and Toga, A., “Multi-Atlas Tensor-Based Morphometry and its
Application to a Genetic Study of 92 Twins,” 48–55 (2008).
[143] Pizer, S. M., Fritsch, D. S., Yushkevich, P. A., Johnson, V. E. and Chaney, E. L., “Segmentation,
registration, and measurement of shape variation via image object shape.,” IEEE Trans. Med.
Imaging 18(10), 851–865 (1999).
[144] Hotelling, H., “The Generalization of Student’s Ratio,” Ann. Math. Stat. 2(3), 360–378 (1931).
74

[145] Wang, Y., Yuan, L., Shi, J., Greve, A., Ye, J., Toga, A. W., Reiss, A. L. and Thompson, P. M.,
“Applying tensor-based morphometry to parametric surfaces can improve MRI-based disease
diagnosis.,” Neuroimage 74, 209–230 (2013).
[146] Gilliam, M., Stockman, M., Malek, M., Sharp, W., Greenstein, D., Lalonde, F., Clasen, L., Giedd, J.,
Rapoport, J. and Shaw, P., “Developmental trajectories of the corpus callosum in attention-
deficit/hyperactivity disorder.,” Biol. Psychiatry 69(9), 839–846 (2011).

Abstract (if available)

Abstract The characterization of healthy brain and cranial anatomical growth during early childhood is a vital step in our understanding of normal and pathological neurodevelopment. The brain and cranium develops rapidly in the first 5 years of life, and describing this process in vivo through medical imaging would allow comparison tools to aid in disease diagnosis. This thesis focuses more specifically on the anatomical development of three key brain and head structures that have so far been understudied in early childhood: the central sulcus (CS), the neurocranium, and the corpus callosum (CC). Below a brief abstract is presented for these three structures respectively. ❧ Sulcal growth begins at the fetal stage and continues evolving through the first years of life. The CS, which begins developing at 20 weeks gestational age, has been shown to vary in depth over the range of the sulcus. The CS is located adjacent to the precentral gyrus which plays an important role in motor function, yet, despite this significance, normal development of the CS has not been studied. This thesis explores CS depth changes in early childhood using MRI. ❧ Along with the brain, the neurocranium also grows rapidly in the first 2 years of life to accommodate brain growth. Neurocranium growth patterns have rarely been studied except for global values such as circumference and head volume. Neurocranium thickness, in particular, is an important feature to investigate especially to predict the predisposition of the neurocranium to disease or fracture. In this thesis, we present a pipeline to compute neurocranium thickness using the Laplace-Beltrami operator on volumetric meshes generated from processing MRI data in early childhood. ❧ One of the most studied structures in adults and aging, the CC is also susceptible to change in early childhood. Cognitive and behavioral deficits in neurodevelopmental disorders have been correlated with structural changes in the CC. However, studies on healthy development of the CC using structural MRI data, especially between the ages of 1 and 5 are scarce. To this end, a combined measure of surface multivariate tensor based morphometry (mTBM) and medial axial distance (MAD) was used in this thesis to detect sets of local geometric shape differences between the different subject groups. This combined metric has demonstrated strong statistical power over voxel based TBM or individual measures of MAD and mTBM respectively in a number of pediatric and adult neuroimaging studies.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Diffusion MRI of the human brain: signal modeling and quantitative analysis

PDF

Shape, pose, and connectivity in subcortical networks across the human lifespan

PDF

Pediatric magnetic resonance image processing: applications to posterior fossa cancer and normal development

PDF

Unveiling the white matter microstructure in 22q11.2 deletion syndrome with diffusion magnetic resonance imaging

PDF

A virtual reality exergaming system to enhance brain health in older adults at risk for Alzheimer’s disease

PDF

Theory and simulation of diffusion magnetic resonance imaging on brain's white matter

PDF

Pattern detection in medical imaging: pathology specific imaging contrast, features, and statistical models

PDF

Applications of graph theory to brain connectivity analysis

PDF

Susceptibility-weighted MRI for the evaluation of brain oxygenation and brain iron in sickle cell disease

PDF

Regional cerebral hemodynamic responses to hypoxia in humans using MRI and NIRS

PDF

Signal processing methods for brain connectivity

PDF

Characterization of lenticulostriate arteries using high-resolution black blood MRI as an early imaging biomarker for vascular cognitive impairment and dementia

PDF

Neuroimaging markers of risk & resilience to brain aging and dementia

PDF

Improving sensitivity and spatial coverage of myocardial arterial spin labeling

PDF

Mapping water exchange rate across the blood-brain barrier

PDF

3D vessel mapping techniques for retina and brain as an early imaging biomarker for small vessel diseases

PDF

Functional connectivity analysis and network identification in the human brain

PDF

Morphological and microstructural models of typical development in the hippocampus and white matter

PDF

Development of a toolbox for global functional brain imaging of wake and sleep states in zebrafish

PDF

Estimation of cognitive brain activity in sickle cell disease using functional near-infrared spectroscopy and dynamic systems modeling

Asset Metadata

Creator Gajawelli, Niharika Rao (author)

Core Title Characterization of the brain in early childhood

School Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Biomedical Engineering

Publication Date 08/23/2019

Defense Date 01/23/2018

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

Tag central sulcus,corpus callosum,Early childhood,human brain,magnetic resonance imaging,neurocranium,OAI-PMH Harvest,structural MRI analysis

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Khoo, Michael (committee chair), Lepore, Natasha (committee chair), Joshi, Anand (committee member), Law, Meng (committee member), Pa, Judy (committee member)

Creator Email niharika.gajawelli@gmail.om

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c40-478841

Unique identifier UC11266596

Identifier etd-GajawelliN-6059.pdf (filename),usctheses-c40-478841 (legacy record id)

Legacy Identifier etd-GajawelliN-6059.pdf

Dmrecord 478841

Document Type Dissertation

Rights Gajawelli, Niharika Rao

Type texts

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

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA