Developmental Aspects of Pediatric fMRI: Considerations for Image
Acquisition, Analysis, and Interpretation
William Davis Gaillard,*,† Cecile B. Grandin,†,‡ and Benjamin Xu†
*Department of Neurology, The Children’s National Medical Center, The George Washington University School of Medicine, Washington
DC; †The Epilepsy Research Branch, NINDS, NIH Bethesda, Maryland; and ‡Department of Radiology, St Luc University Hospital,
Universite ´Catholique de Louvain, Belgium
Received J anuary 21, 2000
F unctional MR I provides a powerful means to iden-
tify and trace the evolution, development, and consol-
idation of cognitive neural networks through normal
childhood. Neural network perturbations due to dis-
ease and other adverse factors during development
can also be explored. Studies performed to date sug-
gest that normal children older than 5 years show
activation maps comparable to adults for similar cog-
nitive paradigms. Minor differences in adult and pedi-
atric activation maps may reflect age dependent strat-
egies or maturation of cognitive networks. However,
there are important physiologic and anatomic differ-
ences in children, varying with age, that may affect
the acquisition, analysis, and interpretation of pediat-
ric fMR I data. Differences between children and adult
fMR I comparison studies may reflect technical aspects
of data acquisition as much as developmental and
brain maturation factors.
© 2001 Academic Press
Functional MRI (fMRI) provides a powerful means to
map neural networks that underlie cognitive functions
in adults and children. Unlike
sion tomography (PET), fMRI does not involve radia-
tion, and thus can be used to study normal children.
MRI studies can be repeated, and children trained to
remain still in the MRI environment. fMRI is well
suited toidentify and tracetheevolution, development,
and consolidation of cognitiveneural networks through
normal childhood. Furthermore, the effect of disease
states, such as epilepsy, and disorders, such as dys-
lexia and learning disabilities, on developing neural
networks can be explored.
A number of recent studies have used fMRI to iden-
tify the neural networks associated with cognitive, mo-
tor, and sensory function in children; they include
15O-water positron emis-
working memory, spatial memory, verbal fluency,
reading visual recognition, imagery, motor and sensory
tasks (Casey, 1995, 1997; Hertz-Pannier, 1997; Staple-
ton, 1997; Thomas, 1999; Pugliese, 1999; Booth, 1999;
Vaidya, 1998; Lee, 1999; Baird, 1999; Gaillard, 1999,
2000a; Nelson, 2000). Although some of these investi-
gations describe intriguing differences that may reflect
developmental strategies for task performance or mat-
uration of cognitive networks, they suggest that acti-
vation maps are fundamentally the same in normal
children older than 8 years and adults, when similar
paradigms are used.
Brain mapping is based on the observations of Roy
and Sherrington (1890), that cerebral activity is accom-
panied by an increase in cerebral blood flow (CBF). Fox
and colleagues (1986) used
sory task and found that local CBF and oxygen delivery
exceeded oxygen extraction and demand in activated
cortex. Optical imaging studies show that after stimu-
lus onset, initial increased oxygen extraction, de-
creased oxyhemoglobin, and concomitant increase in
deoxyhemoglobin is followed within 2–4 s, by “luxury
hyperperfusion” with an over abundance of oxygenated
hemoglobin. It is this physiologic epiphenomenon, with
a delayed change in oxy/deoxy hemoglobin ratio, that
provides the basis for blood oxygen level-dependent
(BOLD) fMRI (Malonek, 1997). Regional needs and
regulation of cerebral blood flow occur at the microvas-
cular level and may be restricted toregions as small as
250 ?m (Lou, 1987). The mediator of this response is
not known but may be a third factor rather than a
direct link between perfusion and metabolism (Lou,
1987). It is important to recall that current fMRI tech-
niques do not measure synaptic metabolic activity di-
rectly and that the BOLD response is detected as a
relative change in MRsignal and corresponding
change in blood flow: it is not an absolute measure and
depends heavily on the cognitive difference between
15O-water PET and a sen-
NeuroImage 13, 239–249 (2001)
doi:10.1006/nimg.2000.0681, available online at http://www.idealibrary.com on
Copyright © 2001 by Academic Press
All rights of reproduction in any form reserved.
control and task conditions. Furthermore, the statisti-
cal thresholds used to determine a significant signal
change are derived from adult studies. Identifying true
activation also depends on identifying factors that ob-
scure detection of signal, such as motion.
Many issues remain tobe clarified for proper interpre-
tation of fMRI in children. A wide range of anatomic,
physiological, and cognitive factors change during child
development, and may confound functional neuroimag-
ing data acquisition, analysis, and interpretation. They
may affect theability todetect a signal, influencehow the
signal is processed and analyzed, and determine its loca-
tion and magnitude. The particular challenge with child
studies is to take into account the constant growth and
change in physical as well as psychological capacity.
Studies comparing adults and children may also suffer
fromtechnical challenges that areunder recognized, such
as the common practice of using equipment originally
designed for adults that may not be optimal when used
with children. As developmental studies using fMRI in
children are in their infancy, it is time to identify and
consider issues that merit investigation and consider-
ation for future studies.
BOLD RESPONSE IN NEONATES, INFANTS,
fMRI studies of higher cortical function—working
memory, verbal fluency, reading—in awake children
between 7 and 15 years, relying on the BOLD fMRI
technique, provide empirical evidence that the princi-
ples underlying adult studies resemble those in chil-
dren (Casey, 1995, 1999; Hertz-Pannier, 1997; Staple-
ton, 1997; Booth, 1999; Lee, 1999; Pugliese, 1999;
Baird, 1999; Thomas, 1999; Gaillard, 1999, 2000a). The
hemodynamic response, seen in 2- to 3-year-olds dur-
ing passive sensory tasks (photic stimulation, tactile
stimulation, passive listening tosound) and in children
as young as 5 years whileperforming cognitivetasks, is
also similar (Stapleton, 1997; Logan, 1999a; Souwei-
dane, 1999) (Fig. 1). There is little cause to believe the
physiology of the BOLD response, or its detection in
children, is remarkably different than in adults.
Whether there are minor differences in threshold of
response, reactivity, or robustness of response, how-
ever, is less certain. Maturation of the response to
adult levels probably occurs at a young age. None of
these studies, however, have specifically examined the
time course or pattern of the BOLD response as has
D’Esposito, 1999); and it is not known whether the
developmental differences in myelination and synapse
formation discussed below affect the pattern and tem-
poral delay of the BOLD effect.
There is some evidence to suggest, however, that
neonates and infants may exhibit different vascular
responses than older children and adults (Yamada,
1997; Born, 1998; Martin, 1999). Born et al. (1998)
studied 17 neonates, infants, and toddlers, ranging
from 3 days to48 months, with BOLD fMRI. All studies
took placein thesleep (6) or sedated (11) state; thetask
monitored a response to a photic flicker compared to
rest. Contrary to expectations, they found a mean de-
crease in signal in 10 subjects during the photic flash
condition and a mean decrease in signal of 2.2% for the
group (Born, 1998). Yamada et al. (1997), also using
the same paradigm in 15 sedated children between 1
and 54 weeks, found an increase in signal in children
younger than 5 weeks and a decrease in signal in
children older than 8 weeks. The authors interpreted
their results as evidenceeither for decreased blood flow
or increased oxygen extraction. Martin et al. (1999),
whosuccessfully studied 58 of 75 children (47 sedated)
between 1 day and 12 years of age, found mixed results
of either increased or decreased signal in occipital cor-
tex in all populations. In contrast to BOLD fMRI, op-
tical imaging in neonates using visual response to
checkerboard pattern shows a hemodynamic response
similar to those in older humans and is similar to that
described by Malonek et al. (1997) in adult cats (Meek,
1998). Meek and colleagues (1998) obtained data in 20
of 58 children (10 control, 10 subject) who ranged be-
tween 2 days and 3 months. They found a mean 9%
signal increase, with contributions of 67% by oxygen-
ated hemoglobin. Thus, they describe an increase in
oxygen extraction but an overabundant supply of oxy-
genated hemoglobin similar to adult-based studies.
The varied and inconsistent responses found with
BOLD fMRI suggest that methodological factors may
be important in data acquisition and analysis in neo-
nates and infants. Sedation, which was used in nearly
all the younger children studied, significantly de-
presses CMRGlc on the order of 25–30% and may alter
blood flow responses (Theodore, 1989).
PEDIATRIC ANATOMICAL AND
Many physical and physiological differences between
children and adults continue tochange during a child’s
growth and maturation and may influence how data
can be collected, analyzed, and interpreted. Among the
most relevant factors are body size and morphologic
differences in children’s brain structure, including the
proportion of gray to white matter. Some factors, such
as the cardiorespiratory cycle, will increase motion ar-
tifact, which is always problematic in young children
who may be less likely to remain still than many
adults. Other factors, such as synaptic density and
differences in resting blood flow and metabolism, may
influence signal detection and subsequent analysis.
The most obvious difference is that children are
smaller than adults. Proportionately, head size differs
less than does body size, but for a 5 to 7-year-old child
GAILLARD, GRANDIN, AND XU
the head circumference may be 6–10% smaller than an
adult’s. Sinuses, which are a cause of MRI signal dis-
tortion, are not fully formed until adolescence. More
important, children’s skulls are thinner and less mus-
cular, and their necks are shorter. As a result, children
do not fit as readily into MR head coils primarily de-
signed for adults. The brain to be imaged may not rest
in the center of the head coil; signal may be distorted
by field inhomogeneity reducing the signal to noise
ratio upon which activation maps are based (Gaillard,
2000a, see below). Conversely, surface coils will lie
closer to the brain being imaged in children than in
adults because their skulls are thinner and are more
likely to enhance detection of increase signal induced
by the BOLD response (Gaillard, 2000a). Ideally, pedi-
atric studies should be conducted on equipment
adapted for pediatric specifications: head coils can be
made that are suitable for children.
A further consequence of shorter stature is that
transmission of movement from jaw movement (Eden,
1999) and thecardiorespiratory cyclemay begreater in
children. The cardiopulmonary cycle is more dynamic
in children, which may increase artifact engendered
from respiration and vascular pulsatility. Children
alsohavehigher heart rates and respiratory rates than
adults, which do not approach adult levels until ado-
lescence. The pulsatility is greatest in the basal gan-
glia but is also transmitted to the cortical parenchyma
(Poncelet, 1992). The combination of higher pulsatility
and possibly a higher parenchyma compliance in chil-
dren compared to adults may increase parenchymal
motion and introduce more physiological noise in pedi-
atric data. Finally, blood pressure is lower in children,
with some mild differences between boys and girls, and
rises steadily with age (Table 1). Unlike adults, chil-
dren compensate for increased cardiac output demand
with increased heart rate, rather than with greater
cardiac contractility. Cerebral perfusion pressure and
cerebral blood flow are tightly regulated and main-
tained in a narrow window to guard against low or
F IG. 1.
viewing large dot patterns (t ? 4.0). Activation is seen in left temporal language areas (middle temporal lobe). (B) The time course of the
signal change from the left temporal lobe seen in the activation map. The study was performed on a conventional 1.5 T scanner using an
echoplanar imaging BOLD technique.
(A) Activation map of a right-handed native English speaking 5.5-year-old girl obtained while reading a story compared to
DEVELOPMENTAL ASPECTS OF PEDIATRIC fMRI
elevated blood pressure or increased intracranial pres-
sure. For the purposes of studying normal children
these factors will not be as important as monitoring
and correcting for vascular artifact and motion induced
by the cardiorespiratory cycle.
The BOLD effect is directly affected by the oxygen
carrying capacity of blood and is suseptible to changes
in hemoglobin type and concentration. Fetal hemoglo-
bin, with its greater oxygen binding, gradually declines
during the first months of life—20% of hemoglobin by
four months, and less than 2% at 12 months (Garby,
1962)—and is mostly an issuefor investigations involv-
ing neonates and infants. Hemoglobin concentration
and hematocrit have a nadir in the first year of life, at
2 months, and then steadily increase through adoles-
cence (Table 1). In postmenarche females, alterations
in hemoglobin due to menstrual loss may affect the
magnitude of the BOLD response. Cerebral blood flow
is also increased by hypercarbia, and decreased with
hypocarbia. Hyperventilation will decrease the hemo-
dynamic response, and breath holding will increase
CBF (Logan, 1999b). These factors could conceivably
affect studies if the respiratory rate changes across
conditions and are more likely to affect younger chil-
dren who may be more prone to hyperventilate when
anxious in the scanner or hold their breath for some
tasks. These factors are more likely to affect the mag-
nitude and secondarily the extent of the hemodynamic
The principal factors central to brain development
that may influence either the hemodynamic response
or the analysis of pediatric data, necessary to deter-
mine the presence and location of BOLD response,
relate to brain morphology: brain size, neuronal con-
nectivity, and synaptic density. The bulk of brain
growth occurs before the fifth year of life. About 5–8%
of cerebral growth occurs between the fifth year and
adolescence (Caviness, 1996; Reiss, 1996; Giedd, 1996),
the ages during which most pediatric fMRI are per-
formed aside from neonatal studies. Brain maturation
occurs at different rates in different brain regions. The
primary sensory and motor areas are the first to com-
plete development, whereas the association areas, es-
pecially parietal and frontal regions, are the last to
mature. These changes are reflected in myelination,
which provides the bulk of regional brain size. Myeli-
nation in association areas continues throughout child-
hood, and is not complete in frontal association cortex
until young adulthood (Yakovlev, 1967). Structural
MRI can be used to detect these developmental differ-
ences long known from pathologically based studies
(Yakovlev, 1967; Giedd, 1999a; Sowell, 1999). A sec-
ondary effect of immature myelination is that synaptic
connections in association areas are not firmly estab-
lished in these areas. The relationship between struc-
tural immaturity and the physiology of the BOLD re-
sponse is unknown, but is likely to diminish its
magnitude and may limit its detection. Finally, the
cortical regions to be analyzed with standard process-
ing techniques may not have a uniform anatomy
within a given subject, let alone between different age
groups, reflecting differences in regional maturation.
The smaller brain size and age-dependent differ-
ences in proportional brain region size will affect warp-
ing children’s brain images into standard atlases com-
monly employed for group analysis. There are no
standard pediatric atlases for any age. The most com-
monly employed atlases are based on young adults (the
Montreal Neurologic Institute brain (Kollokian, 1997))
or an elderly lady (Talairach and Tourneaux 1988).
The distortion engendered by warping children’s
brains intoeither adult atlas will influence group anal-
ysis and comparisons across ages are likely to be in
error by several millimeters and perhaps by as much as
a centimeter. The Pediatric Brain Project, a joint
NINDS, NIMH, NICHD multicenter prospective longi-
tudinal initiative, to generate a normative MRI struc-
tural image data base will also provide data for the
creation of age appropriate brain atlases suitable for
group analysis involving children of different ages.
Another factor concerns the gray mantle, which is
thicker and more dense in children than adults. There
is less free space containing cerebral spinal fluid about
cerebral arteries and veins (Giedd, 1996; Caviness,
1996). This is the opposite difficulty that investigators
of aging encounter when comparing cognitive studies
in the elderly population toyoung adult controls. Older
subjects have greater atrophy and more generous sul-
cal space. In children, then, there is a smaller volume
averaging effect in gray matter voxels. The solution
used by investigators of the elderly may be adapted to
pediatric analysis which corrects signal magnitude in
gray matter for volume averaging effect (Ross, 1997).
The ratio of gray to white matter is also greater in
children than in adults (Giedd, 1999b): the number of
gray as opposed towhite matter pixels varies with age.
Such differences may not only affect warping into a
T ABL E 1
Age-Dependent Changes for Heart Rate, Respiratory Rate,
and Systolic Blood Pressure
47 (M)/41 (F)
Note. Adapted from the Harriet LaneHandbook, Manual for Pedi-
atric House Officers, 14th ed. BP, blood pressure; Hct, Hematocrit;
M, male; F, female.
GAILLARD, GRANDIN, AND XU
standard atlas, but also affect calculations to correct
for multiple corrections based on preordained assump-
tions about gray and white matter volume relation-
ships. One solution may be to segment brain volumes
for statistical analysis to account for developmental
changes in gray to white matter distribution (Giedd,
1999b); ideally segmentation algorithm thresholds
should be derived from pediatric high resolution struc-
There are additional difficulties for investigators of
very young children. The gray/white matter distinction
is considerably more difficult to discern in children
younger than two as myelination is immature, hence
segmentation programs may be less reliable. In con-
junction with synapse formation in gray matter and
myelination, the capillary bed undergoes developmen-
tal changes. In the third trimester vessels are radially
oriented. These precursors of the capillary bed de-
crease as the number of horizontal branches increase
(Norman, 1986). The spatial geometry of the capillary
network and draining veins influences the magnitude
of the BOLD fMRI signal which may evolve in parallel
to the maturation of the capillary bed.
The synaptic density within gray matter increases
during the first years of life and peaks between 4 and 8
years. As a consequence of the competitive establish-
ment of synaptic connections, synapses are pruned and
do not reach adult density until adolescence (Hutten-
locker, 1979; Bourgeois, 1994; Goldman-Rackic, 1987;
PET) studies in normal development show increased
glucose utilization in all brain areas that peaks be-
tween four and eight years (Fig. 2). This increase is a
18Fluorodeoxy-glucose PET (FDG-
reflection of synaptic density where most metabolic
activity occurs (Chugani, 1987; Van Bogaert, 1998;
Bentourkia, 1998). CBF, determined by Xenon single
photonemission computed tomography
shows parallel increases toFDG-PET with a maximum
between four and eight years (Fig. 3) (Chiron, 1992,
1997; Barthel, 1997). As with myelination there are
regional differences as towhen maturation is complete:
primary sensory cortex first, then motor cortex, and
finally association cortex reach adult levels of glucose
consumption and resting blood flow (Chugani, 1987;
Chiron, 1992). The central nervous system maintains a
constant level of perfusion tomatch metabolic demand.
There is a tight coupling between perfusion and me-
tabolism. This is reflected in the similar curves of CBF
and cerebral metabolic rate for glucose (CMRGlc) rel-
ative to age and is maintained in childhood and adult-
hood (Chugani, 1987; Chiron, 1992). Perfusion is ac-
tively regulated to maintain central CBF to meet
metabolic demand constant over a range of blood pres-
sures at the arteriolar level (Lou, 1987). The coupling
of perfusion to metabolism is constant except under
some disease states such as in acute stroke (Baron,
1982), chronic epilepsy (Gaillard, 1995; Breier, 1997),
and systemically ill neonates (Volpe, 1995).
These developmental changes in CBF and CMRGlc
have important implications for fMRI signal and anal-
ysis in children. The higher resting levels of CMRGlc
and CBF found in children compared to adults may
lead to absolute and proportionate differences in blood
flow accompanying activation and may influence the
appropriate choice of threshold (Hertz-Pannier, 1997;
Grandin, 1999; Gaillard, 2000a). With greater synaptic
F IG. 2.
abnormalities using FDG-PET. The peak glucose consumption, for all cortical brain areas, occurred in children between 4 and 8 years.
Adapted from Chugani et al. (1987).
Age distribution curve of regional glucose consumption from a cross sectional study performed in children with mild neurological
DEVELOPMENTAL ASPECTS OF PEDIATRIC fMRI
density, metabolism, and blood flow, a more robust
increase in CBF might be expected—that is, the acti-
vation ratio may be greater in children per unit cortex
activated (Gaillard, 2000a). On the other hand, a sta-
tistically “true” signal in terms of absolute change of
blood flow for an adult may not reach statistical signif-
icance for a child because of the overall relative greater
resting blood flow per voxel gray matter in children.
Furthermore, synaptic connections may be less well
formed and less efficient, and more diffusely spread
about cortical regions as neural networks and cortical
areas may not becompletely consolidated resulting in a
more widespread and lower signal that may escape
detection. The former scenario would result in a more
readily detected activation signal, the latter would de-
crease the activation signal relative to baseline and
make identification of true activation less certain.
These competing issues have not yet been closely ex-
amined but may influence studies designed to test the
concept of consolidation of cognitive processes into in-
creasingly discrete regions and networks during criti-
cal stages of cognitive development (Gaillard, 2000a;
Mu ¨ller, 2000). Determining absolute measures of blood
flow during rest and task conditions, which would help
resolve these issues, is not available with current
BOLD fMRI methods but may be possible with arterial
spin tagging techniques (Ye, 1998; Lia, 2000).
PEDIATRIC NEUROPSYCHOLOGICAL ISSUES
Neuropsychological issues can affect functional im-
aging based comparisons between children of different
ages and adults. An important concern is how to ac-
count for, while at the same time investigating, learn-
ing and skill acquisition: this is the paradox of imaging
brain development. Studies comparing brain activation
patterns between age groups engender the risk that
they may only identify differences in performance
rather than development. In individual children, the
relation of ageand skill levels may vary enormously. In
pediatric studies it is important to incorporate some
measure of task difficulty and performance, particu-
larly when comparing different agegroups or normal to
patient populations. Performance speed, response
time, and accuracy may all be influential variables.
Task performance and response times can be moni-
tored in children in the scanner, but may be obtained
at the cost of an increase in motion artifact or creating
an additional cognitive level to be considered in data
analysis. In adults, stimulus presentation rate and
duration affect activation patterns for some auditory
and visually based stimuli (Fox, 1986; Binder, 1994;
Price, 1992, 1994, 1996; Fiez, 1996, 1998). Theseeffects
are likely compounded in children. Facility with task
may effect thelocation, extent, and degreeof activation
(Raichle, 1994; Deiber, 1997; Thomas, 1999; Gaillard,
2000a,b). For instance, preliminary studies in a work-
ing memory task suggested significant differences be-
tween adult and pediatric study groups (Casey, 1997).
In a subsequent study, activation patterns were found
to be similar after taking into account the differences
in task performance (Thomas, 1999). For some tasks,
however, activation may reflect effort rather than task
performance: activation in dominant inferior frontal
gyrus during verbal fluency, word generation toa letter
or category, does not reflect performance over a wide
range of ability (Grandin, 1998, 1999; Gaillard, 2000a).
How to design experiments to account for these dif-
ferences is a challenge without an ideal solution.
BOLD fMRI depends upon across-task changes in re-
gional CBF. Altering experimental parameters may
affect cognitive processesinvolved inthe task
F IG. 3.
occurred between four to eight years of age. Adapted from Chiron et al. (1992).
Age distribution curve of cerebral blood flow (CBF) in a similar population using Xenon SPECT. The peak in brain region CBF
GAILLARD, GRANDIN, AND XU
(Bookheimer, 1995); and cognitive strategies may be
different among populations (Shaywitz, 1998; Pugh,
2000). They may alsovary depending on task construc-
tion (Bookheimer, 1995; Gaillard, 2000b) or according
to skill. There is no guarantee that processing strate-
gies are the same between all groups examined. For
example, slow readers may rely more heavily on pho-
nologic processing than more skilled readers (Bookhei-
mer, 2000). Changes in tasks may alter strategies and
hence activation patterns: adults reading the same
stimuli out loud rather than silently show greater ac-
tivation of phonological networks (Bookheimer, 1995).
One option when studying children of different ages
and skill is to use a common paradigm readily per-
formed by all study subjects. Imaging data can be
analyzed and corrected for task performance obtained
either in or out of the scanner. The risk here is that
little activation may be found in older or more skilled
subjects. Another method is totailor paradigms appro-
priate to an individual subject’s skill level based on
behavioral performance or to design self-paced stimu-
lation presentation paradigms. Imaging data may be
analyzed in relation to results from behavioral testing
or monitoring (Thomas, 1999; Gaillard, 2000a). An-
other means of addressing variable performance be-
tween study groups is to use a parametric design in-
corporating different levels of task difficulty for the
study groups to identify regions that do or do not re-
spond linearly to task difficulty (Price, 1992; Bookhei-
mer, 2000). Parametric designs alsoallow for groups to
be compared based on performance level rather than
absolute performance. Mixed models, of which growth
curve models and repeated measures models are spe-
cial cases, may also be well suited to analyze develop-
mental imaging and psychometric measure changes
over time in prospective longitudinal studies of cogni-
tive processes (Gizzle, 1969; Zerbe, 1994). These types
of models have been used to describe linear and non-
linear growth curves and may be adapted for individ-
ual and group data analysis and adjusted by covariates
in neuropsychological domains.
The choice of a control condition may also be critical
as they are active states in their own fashion (Binder,
1999); there may be significant differences between
adults and children in control conditions that affect
analysis of the task condition. For example, control
conditions for reading in adults commonly employ a
string of characters or symbols. Such a design for chil-
dren may evoke activation in language cortex as dis-
criminating these characters from words may involve
language systems (Gaillard, 1999). In another exam-
ple, an adult or older child may be more likely than a 5
year old to stare at a cross hair and successfully make
their mind “blank.” As a result, negative findings in
cortical regions where one anticipates a response must
be conservatively interpreted in individual analysis.
Caution is alsonecessary when interpreting absence
of typical activation found when group analyses are
conducted in patients with disease states that are in-
herently heterogeneous, such as epilepsy, attention
deficit, and reading disorders. Individual variability of
activation maps found among patient populations may
be subtracted and lost in group analysis when com-
pared to homogeneous populations where activation
patterns are augmented by a group analysis strategy.
For example, Bellgowan et al. (1998) described activa-
tion of middle left parahippocampal gyrus and hip-
pocampus during verbal encoding of a semantic deci-
sion task using a group analysis in right but not left
temporal lobe adult seizure focus patients. However,
there was no clinical difference in performance be-
tween right and left sided temporal lobe epilepsy pa-
tient groups, suggesting that the left temporal lobe
patients maintained the capacity to perform the task
and the location of which was not identified in the
analysis process. Not only children with disease states
or developmental conditions, but also normal children
may use different cognitive strategies or neural net-
works to perform a given task, which supports the
position that individual as well as group analysis may
need to be performed to identify such heterogeneity.
A further difficulty ensues when studying a patient
population where performance may clearly be different
from the control population—a difficulty not restricted
topediatric studies. Booth et al. (1999) showed variable
activation in homotopic language cortex in children
with left hemispheric stroke. In their decision based
auditory comprehension tasks, the investigators ob-
tained extensive behavioral data and suggested that
poor performance of some children indicated that ei-
ther they were guessing, or had given up because the
task was too difficult for them. It is problematic, but
not impossible, toassess learning disabilities when the
task at hand is poorly performed (Shaywitz, 1998;
It is also critical to assure that individuals and pop-
ulations targeted for functional imaging studies can
perform the experimental task. In the application of
fMRI to determine laterality and location of language
cortex in preparation for surgery, absent activation, or
highly atypical activation patterns, need to be either
repeated or confirmed with other means, such as the
intra-carotid amytal procedure or electrocortical stim-
ulation (Gaillard, 2000b,c).
AN EXAMPLE OF ADULT AND CHILD
Part of the difficulty in comparing pediatric and
adult data may be portrayed in three studies of verbal
fluency involving children and adults at 1.5 T using
echoplanar imaging. They highlight somedifficulties in
DEVELOPMENTAL ASPECTS OF PEDIATRIC fMRI
determining and comparing activation maps in adults
We used a verbal fluency paradigm, in which sub-
jects generated words from given letters, in 11 children
with localization-related epilepsy. The cross correla-
tion coefficient threshold needed toidentify frontal lan-
guage areas was considerably higher than commonly
used in adults (0.7–0.8 vs 0.5–0.6) even though the
hemodynamic response patterns were similar (Hertz-
Pannier, 1997). In a follow-up study we compared ac-
tivation patterns between 10 adults (mean age 28.7;
range 19.3–48.0) and 10 children (mean age 10.7;
range 8.1–13.1) using a common threshold (r ? 0.7) for
analysis. Although activation patterns were similar,
children consistently activated morecortex than adults
(390 vs 249 voxels, P ? 0.02) both in expected language
areas and in contralateral cortex. In addition, children
tended to have a higher activation ratio, or response
magnitude (Gaillard, 2000a). These results are in line
with the less established and lateralized networks hy-
pothesized toexist in children with a greater activation
ratio reflecting the increased synaptic density in neo-
cortex. A third verbal fluency MRI study, using t sta-
tistics, questions these conclusions (Grandin, 1999).
This study compared 32 adults (mean 29.3; range 22–
45) and 20 children (mean 10.4; range, 7–14). The
results were reversed: adults activated more pixels in
frontal regions than children (threshold t ? 5 (the
theoretical Bonferroni corrected at P ? 0.05) 25.0 vs
10.8; at t ? 4.0, 59.0 vs 31.6), and higher thresholds
were used to identify areas in the adults. However,
asymmetry indices, showing degree of language domi-
nance, were the same between adults and children (for
t ? 5: 0.72 vs 0.77; for t ? 4: 0.52 vs 0.52) (Grandin,
It is not clear whether these findings represent a
true developmental effect or a technical effect of opti-
mizing signal to noise. In the first two studies there
were no perceptible differences in motion artifact, but
they were performed without quantitating motion; the
third study quantitated motion and showed somewhat
greater motion in children than in adults and mildly
greater variance in voxel intensity in the children (this
may be related to macroscopic movements of the sub-
ject or tomicroscopic movements caused by physiologic
noise such as cardiopulmonary
above). Although statistical methods used in these
studies were different they should not alter the funda-
mental observations. Within groups there was a minor
trend to activate more pixels with higher verbal flu-
ency scores, but the adults performed better than the
children. Given the contrary results other factors must
besought. Theprincipal differencebetween thesestud-
ies may be explained by image acquisition techniques,
specifically in the coils used to gather the data. In the
first two studies surface coils were used that lie closer
to the brain imaged in children than adults. In con-
trast, the third study was performed with a standard
head coil designed for adults: here adults were posi-
tioned closer to the center of the coil than were the
children, which may have led to poorer detection of
signal in children or greater noise in children. We have
found similar results with reading paradigms where
children aged 8 to 12 years demonstrated nearly iden-
tical activation patterns to adults, but at lower t
thresholds than those used with adults (Gaillard,
CHOICE OF STATISTICAL THRESHOLD FOR
It is possible that assumptions about the proper
threshold for children may be overly conservative.
Many of the factors reviewed here may influence the
detection of signal. Children have a greater gray to
white matter ratio which may increase the threshold
relative to adults. Furthermore, the signal to noise
ratio may be poor in children, and exacerbated by vas-
cular artifact or coil design, so that real activity does
not reach the level of statistical significance. By lower-
ing the threshold, areas of “true” activation may be
discovered. Technical as well as developmental factors
may bean issuein determining theappropriatethresh-
old at which to analyze pediatric data and may affect
interpretation of either longitudinal or cross-sectional
age-based comparison studies.
Investigators who apply fMRI to identify language
function in individual epilepsy patients, adults and
children, report a need to examine individual data at
different thresholds (Binder, 1996; Hertz-Pannier,
1997; Stapleton, 1997). They find reliable activation
maps at lower thresholds that match group activation
maps established in normal populations at higher
thresholds and that have been subsequently confirmed
by either the intracarotid amytal procedure or electro-
cortical stimulation. The commonly employed thresh-
olds, set high toavoid false-positiveactivation, may fail
to identify significant areas or true activation extent.
When different aged populations are compared for a
cognitivetask, investigators may need toexaminetheir
data at different thresholds to identify the brain areas
that are either commonly or distinctly involved. Con-
sequently, broad comparisons between age groups re-
garding the general location of brain areas activated
may be more justified than interpretation based upon
extent and magnitude of activation.
fMRI provides a safe and reliable means for nonin-
vasive identification of neural networks that underlie
many cognitive processes during development. How-
ever, pediatric fMRI activation maps should be inter-
preted with care. Absence of expected activation in
GAILLARD, GRANDIN, AND XU
study populations may support the notion that the
networks to perform the task are not developmentally
established. Alternatively, it may reflect an inability to
perform the task; reflect failure to reach activation
threshold for any of the reasons discussed above; or,
result from task and control conditions that are not
suitable and are not sufficiently distinct (Gaillard,
The physiology underlying cerebral activation in
children and adults appears tobe the same but has not
been rigorously investigated. There are general simi-
larities to adult BOLD response and activation pat-
terns observed in children as young as five years for
cognitive processes, and children older than two years
for sensory tasks. The significance of neonatal and
infantile activation patterns are debated as fMRI gives
different results than optical imaging, but factors such
as sedation may, in part, explain these discrepancies.
Developmental issues need to be considered in data
acquisition and analysis. Synaptic pruning and brain
myelination with concomitant changes in glucose con-
sumption and blood flow may affect the detection, mag-
nitude, and extent of BOLD response. Morphological
differences between children and adult brains may in-
fluencethemagnitudeof signal and thelocation of task
activation, especially when studies are warped into
adult atlases for group analysis. Differences between
children and adults in fMRI comparison studies may
reflect technical aspects of data acquisition as much as
developmental and brain maturation factors. Finally,
there may be significant performance differences found
among individuals that may affect interpretation of
comparisons between age groups or between normal
populations and study groups—such as epilepsy or
reading disability. Sorting out developmental issues
from differences in performance will require care and
innovative study designs. Broad inferences may be
made, but caution is warranted when interpreting
comparison data involving children of different ages
This paper was presented at Tools for Pediatric Neuroimaging:
Inter-Institute Group for Pediatric Functional Neuroimaging, spon-
sored by NINDS, NIMH, NICHD September 1999, Leesburg VA. The
research was supported by NINDS K08 NS01663 and the Epilepsy
Research Branch, NINDS, NIH. We thank Lyn Balsamo and Su-
zanne Reigle for her help in preparing this manuscript.
Baird, A. A., Gruber, S. A., Fein, D. A., Maas, L. C., Steingard, R. J .,
Renshaw, P. F., Cohen, B. M., and Yurgelun-Todd, D. A. 1999.
Functional magnetic resonance imaging of facial affect recognition
in children and adolescents. J . Am. Acad. Child Adolesc. Psychia-
try 38: 195–199.
Baron, J . C., Lebrun-Grandie, P., Collard, P., Crouzel, C., Mestelan,
G., and Bousser, M. G. 1982. Noninvasive measurement of blood
flow, oxygen consumption, and glucose utilization in the same
brain regions in man by positron emission tomography: Concise
communication. J . Nucl. Med. 23: 391–399.
Barthel, H., Wiener, M., Dannenberg, C., Sattler, B., Martin, W. G.,
and Knapp, W. H. 1997. Age-dependence of local cerebral blood
flow—Investigations of 4 to 15-year-old children and adults with
Tc-99m-ECD. Nuclearmedicine 36: 110–116.
Bellgowan, P. S. F., Binder, J . R., Swanson, S. J ., Hammeke, T. A.,
Springer, J . A., Frost, J . A., Mueller, W. M., and Morris, G. L. 1998.
Side of seizure focus predicts left medial temporal lobe activation
during verbal encoding. Neurology 51: 479–484.
Bentourkia, M., Michel, C., Ferriere, G., Bol, A., Coppens, A., Sibo-
mana, M., Bausart, R., Labar, D., and DeVolder, A. G. 1998.
Evolution of brain glucose metabolism with age in epileptic in-
fants, children and adolescents. Brain Dev. 20: 524–529.
Binder, J . R., Rao, S. M., Hammeke, T. A., Frost, J . A., Bandettini,
P. A., and Hyde, J . S. 1994. Effects of stimulus rate on signal
response during functional magnetic resonance imaging of audi-
tory cortex. Brain Res. Cogn. Brain Res. 2: 31–38.
Binder, J . R., Swanson, S. J ., Hammeke, T. A., and Morris, G. L., et
al. 1996. Determination of language dominance using functional
MRI: A comparison with the WADA test. Neurology 46: 978–984.
Binder, J . R., Frost, J . A., Hammeke, T. A., Bellgowan, P. S., Rao,
S. M., and Cox, R. W. 1999. Conceptual processing during the
conscious resting state. A functional MRI study. J . Cogn. Neurosci.
Bookheimer, S. Y., Zeffiro, T. A., Blaxton, T., Gaillard, W. D., and
Theodore, W. 1995. Regional blood flow during object naming and
word reading. Hum. Brain Mapp. 3: 93–106.
Bookheimer, S. Y. 2000. Methodological issues in pediatric neuroim-
aging. Dev. Med. Mental Retard Rev. 6: 161–165.
Booth, J . R., Feldman, H. M., Macwhinney, B., Thulborn, K. R.,
Sacco, K., and Voyvodic, J . 1999. Functional activation patterns in
adults, children, and pediatric patients with brain lesions. Prog.
Neuropsychopharm. Biol. Psych. 23: 669–682.
Born, P., Leth, H., Miranda, M. J ., Rostrup, E., Stensgaard, A.,
Peitersen, B., Larsson, H. B. W., and Lou, H. C. 1998. Visual
activation in infants and young children studied by functional
magnetic resonance imaging. Pediatric Res. 44: 578–583.
Bourgeois, J . P., Goldman-Rakic, P. S., and Rakic, P. 1994. Synap-
togenesis in the prefrontal cortex rhesus monkeys. Cerebral Cortex
Breier, J . I., Mullani, N. A., Thomas, A. B., Wheless, J . W., Plenger,
P. M., Gould, K. L., Papanicolaou, A., and Willmore, L. J . 1997.
Effects of duration of epilepsy on theuncoupling of metabolism and
blood flow in the complex partial seizures. Neurology 48: 1047–
Casey, B. J ., Cohen, J . D., J ezzard, P., Turner, R., Noll, D. C.,
Trainor, R. J ., Giedd, J ., Kaysen, D., Hertz-Pannier, L., and Rap-
oport, J . L. 1995. Activation of prefrontal cortex in children during
a nonspatial working memory task with functional MRI. Neuro-
image 2: 221–229.
Casey, B. J ., Trainor, R. J ., Orendi, J . L., Schubert, A. B., Nystrom,
L. E., Giedd, J . N., Castellanos, F. X., Haxby, J . V., Noll, D. C., and
Cohen, J . D., et al. 1997. A developmental functional MRI study of
prefrontal activation during performance of a Go-No-Go task. J .
Cogn. Neurosci. 9: 835–847.
Casey, B. J . 1999. Images in Neuroscience. Brain development. XII.
Maturation in brain activation. Am. J . Psychiatry 156: 504.
Caviness, V. S., J r., Kennedy, D. N., Richelme, C., Rademacher, J .,
and Filipek, P. A. 1996. The human brain age 7–11 years: A
volumetric analysis based on magnetic resonance images. Cereb.
Cortex 6: 726–736.
DEVELOPMENTAL ASPECTS OF PEDIATRIC fMRI
Changeux, J . P., and Danchin, A. 1976. Selective stabilization of
developing synapses as a mechanism for the specification of neu-
ronal networks. Nature 264: 705–712.
Chiron, C., Raynaud, C., Maziere, B., Zilbovicius, M., Laflamme, L.,
Masure, M. C., Dulac, O., Bourguigon, M., and Syrota, A. 1992.
Changes in regional cerebral blood flow during brain maturation
in children and adolescents. J . Nucl. Med. 33: 696–703.
Chiron, C., J ambaque, I., Mabbout, R., Lounes, R., Dulac, O., and
Syrota, A. 1997. Theright brain is hemispheredominant in human
infants. Brain 120: 1057–1066.
Chugani, H. T., Phelps, M. E., and Mazziotta, J . C. 1987. Positron
emission tomography study of human brain functional develop-
ment. Ann. Neurol. 22: 487–497.
D’Esposito, M., Zarahn, E., Aguirre, G. K., and Rypma, B. 1999. The
effect of normal aging on the coupling of neural activity tothe bold
hemodynamic response. Neuroimage 10: 6–14.
Deiber, M. P., Wise, S. P., Honda, M., Catalan, M. J ., Grafman, J .,
and Hallett, M. 1997. Frontal and parietal networks for condi-
tional motor learning: A positron emission tomography study.
J . Neurophysiol. 78: 977–991.
Eden, G. F., J oseph, J . E., Brown, H. E., Brown, C. P., and Zeffiro,
T. A. 1999. Utilizing hemodynamic delay and dispersion to detect
fMRI signal change without auditory interference: The behavior
interleaved gradients technique. Magn. Reson. Med. 41: 13–20.
Fiez, J . A., Raichle, M. E., Balota, D. A., Tallal, P., and Peterson,
S. E. 1996. PET activation of posterior temporal region during
auditory word presentation and verb generation. Cereb. Cortex 6:
Fiez, J . A., and Petersen, S. E. 1998. Neuroimaging studies of word
reading. Proc. Natl. Acad. Sci. USA 95: 914–921.
Fox, P. T., and Raichle, M. E. 1986. Focal physiological uncoupling of
cerebral blood flow and oxidative metabolism during somatosen-
sory stimulation of human subjects. Proc. Nat. Acad. Sci. USA 83:
Gaillard, W. D., Fazilat, S., White, S., Malow, B., Sato, S., Reeves, P.,
Herscovitch, P., and Theodore, W. H. 1995. Interictal metabolism
and blood flow are uncoupled in temporal lobe cortex of patients
with complex partial epilepsy. Neurology 45: 1841–1847.
Gaillard, W. D., Pugliese, M., Grandin, C. B., Braniecki, S. H., Konda-
paneni, P., Hunter, K., Xu, B., Petrella, J . R., Balsamo, L., and Basso,
G. 1999. Cortical representation of reading in normal children: A
fMRI language study. Ann. Neurol. 46: 528.
Gaillard, W. D., Hertz-Pannier, L., Mott, S. H., Barnett, A. S., LeBi-
han, D., and Theodore, W. H. 2000a. Functional anatomy of cog-
nitive development: fMRI of verbal fluency in children and adults.
Neurology 54: 180–185.
Gaillard, W. D., Bookheimer, S. Y., and Cohen, M. 2000b. Advances
in Neurology. In Neocortical Epilepsy (P. D. Williamson, A. M.
Siegel, V. M. Thadani, and M. S. Gazzangia, Eds.) Vol. 84, pp.
391–404. Lippincott Williams & Wilkins.
Gaillard, W. D. 2000c. Cortical function and epilepsy. Curr. Opin.
Neurol. 13: 193–200.
Garby, L., Sjolin, S., and Vuille, J . C. 1962. Studies on erythro-
kinetics in infancy. Acta Pediatr. 51: 245–254.
Giedd, J . N., Snell, J . W., Lange, N., Rajapakse, J . C., Casey, B. J .,
Kozuch, P. L., Vaituzis, A. C., Vauss, Y. C., Hamburger, S. D.,
Kaysen, D., and Rapoport, J . L. 1996. Quantitative magnetic res-
onance imaging of human brain development: Ages 4–18. Cereb.
Cortex 6: 551–560.
Giedd, J . N., Blumenthal, J ., J effries, N. O., Rajapakse, J . C., Vai-
tuzis, A. C., Liu, H., Berry, Y. C., Tobin, M., Nelson, J ., and
Castellanos, F. X. 1999a. Development of the human corpus callo-
sum during childhood and adolescence: A longitudinal MRI study.
Prog. Neuropsychopharmacol. Biol. Psychiatry 23: 571–588.
Giedd, J . N., Blumenthal, J ., J effries, N. O., Castellanos, F. X., Liu,
H., Zijdenbos, A., Paus, T., Evans, A. C., and Rapoport, J . L. 1999b.
Brain development during childhood and adolescence: A longitu-
dinal MRI study. Nat. Neurosci. 2: 861–863.
Gizzle, J . E., and Allen, D. M. 1969. Analysis of growth and dose
response curves. Biometrics 25: 357–381.
Goldman-Rakic, P. S. 1987. Development of cortical circuitry and
cognitive function. Child Dev. 58: 601–602.
Grandin, C. B., Gaillard, W. D., Whitnah, J . R., Petrella, J . R.,
Braniecki, S., Hunter, K., and Theodore, W. H. 1998. Comparison
of phonological and semantic verbal fluency tasks: An fMRI study.
Neuroimage 7: S133.
Grandin, C. B., Gaillard, W. D., Petrella, J . R., Witnah, J . R.,
Braniecki, S. H., Hunter, K. E., and Theodore, W. H. 1999. Com-
parison of activated brain areas during verbal fluency tasks in
adults and children: An fMRI study. Proc. Am. Soc. Neurorad. 38:
Barrone, M., (Ed.) 1996. Manual for pediatric house officers. In
Harriet Lane Handbook, 14th Ed. Mosby, New York.
Hertz-Pannier, L., Gaillard, W. D., Mott, S., Cuenod, C. A., Bookhei-
mer, S. Y., Weinstein, S., Conry, J ., Papero, P. H., Schiff, S. J .,
LeBihan, D., and Theodore, W. H. 1997. Noninvasive assessment
of languagedominancein children and adolescencewith functional
MRI: A preliminary study. Neurology 48: 1003–1012.
Huttenlocher, P. R. 1979. Synaptic density in human frontal cortex—
Developmental changes and effects of aging. Brain Res. 163: 195–
Kollokian, R. K., Kwan, S., and Evans, A. C. 1997. BrainWeb: Online
interface to a 3D MRI simulated database. Neuroimage 5: S425.
Lee, B. C., Kuppusamy, K., Grueneich, R., El-Ghazzawy, O., Gordon,
R. E., Lin, W., and Haacke, E. M. 1999. Hemispheric language
dominance in children demonstrated by functional magnetic reso-
nance imaging. J . Child Neurol. 14: 78–82.
Lia, T., Guang Chen, Z., Ostergaard, L., Hindmarsh, T., and Mose-
ley, M. E. 2000. Quantification of cerebral blood flow by bolus
tracking and artery spin tagging methods. Magn. Reson. Imag. 18:
Logan, W. J . 1999a. Functional magnetic resonance imaging in chil-
dren. Semin. Pediatr. Neurol. 6: 78–86.
Logan, W. J ., Aleong, R., Crawley, A. P., and Mikulis, D. J . 1999b.
Functional magnetic resonance imaging evaluation of cerebrovas-
cular reactivity in Moyamoya. Ann. Neurol. 46: 551.
Lou, H. C., Edvinsson, L., and MacKenzie, E. T. 1987. The concept of
coupling blood flow to brain function: Revision required? Ann.
Neurol. 22: 289–297.
Malonek, D., Dirnagl, U., Lindauer, U., Yamada, K., Kanno, I., and
Grinvald, A. 1997. Vascular imprints of neuronal activity: Rela-
tionships between thedynamics of cortical blood flow, oxygenation,
and volume changes following sensory stimulation. Proc. Natl.
Acad. Sci. USA 94: 14826–14831.
Martin, E., J oeri, P., Loenneker, T., Ekatodramis, D., Vitacco, D.,
Hennig, J ., and Marcar, V. L. 1999. Visual processing in infants
and children studied using functional MRI. Pediatr. Res. 46: 135–
Meek, J . H., Firbank, M., Elwell, C. E., Atkinson, J ., Braddick, O.,
and Wyatt, J . S. 1998. Regional hemodynamic responses to visual
stimulation in awake infants. Pediatr. Res. 43: 840–843.
Mu ¨ller, R-A., and Courchesne, E. 2000. The duplicity of plasticity: A
conceptual approach tothe study of early lesions and developmen-
tal disorders. In The Foundation and Future of Functional Neuro-
imaging in Child Psychiatry (M. Ernst and J . Rumsey, Eds.), pp.
335–365. Cambridge Univ. Press, Cambridge, UK.
Nelson, C. A., Monk, C. S., Lin, J ., Carver, L. J ., Thomas, K. M., and
Truwit, C. L. 2000. Functional neuroanatomy spatial working
memory in children. Dev. Psychol. 36: 109–116.
GAILLARD, GRANDIN, AND XU
Norman, M. G., and O’Kusky, J . R. 1986. The growth and develop-
ment of microvasculature in human cerebral cortex. J . Neuro-
pathol. Exp. Neurol. 45: 222–232.
Poncelet, B. P., Wedeen, V. J ., Weisskoff, R. M., and Cohen, M. S.
1992. Brain parenchyma motion: Measurement with cine echo-
planar MR imaging. Radiology 185: 645–651.
Price, C. R., Wise, R., Ramsay, S., Friston, K., Howard, D., Patterson,
K., and Frackowiak, R. 1992. Regional response differences within
the human auditory cortex when listening to words. Neurol. Lett.
Price, C. J ., Wise, R. J . S., Watson, J . D. G., Patterson, K., Howard,
D., and Frackowiak, R. S. J . 1994. Brain activity during reading:
The effects of exposure and duration of task. Brain 117: 1255–
Price, C. J ., Wise, R. J . S., and Frackowiak, R. S. J . 1996. Demon-
strating the implicit processing of visually presented words and
pseudowords. Cereb. Cortex 6: 62–70.
Pugh, K. R., Mencl, W. E., J enner, A. R., Katz, L., Frost, S. J ., Lee,
R. J ., Shaywitz, S. E., and Shaywitz, B. A. 2000. Functional neu-
roimaging studies of reading and reading disability (developmen-
tal dyslexia). Dev. Med. Mental Retard. Rev. 6: 207–213.
Pugliese, M., Gaillard, W. D., Basso, G., Braniecki, S. H., Xu, B.,
Balsamo, L. M., Cavazzuti, G. B., Grafman, J ., and Theodore,
W. H. 1999. Functional brain mapping of visual mental imagery
generation in children. Neuroimage 9: S352.
Raichle, M. E., Fiez, J . A., Videen, T. O., MacLeod, A. M. K., Pardo,
J . V., Fox, P. T., and Peterson, S. E. 1994. Practice-related changes
in human brain functional anatomy during nonmotor learning.
Cereb. Cortex 4: 8–26.
Reiss, A. L., Abrams, M. T., Singer, H. S., Ross, J . L., and Denckla,
M. B. 1996. Brain development, gender and IQ in children. A
volumetric imaging study. Brain 119: 1763–1774.
Ross, M. H., Yurelun-Todd, D. A., Renshaw, P. F., Maas, L. C.,
Mendelson, J . H., Mello, N. K., Cohen, B. M., and Levin, J . M.
1997. Age-related reduction in functional MRI response to photic
stimulation. Neurology 48: 174–176.
Roy, C. S., and Sherrington, C. S. 1890. On the regulation of blood
flow to the brain. J . Physiol. 11: 85–108.
Shaywitz, S. E., Shaywitz, B. A., Pugh, K. R., Fulbright, R. K.,
Constable, R. T., Mencl, W. E., Shankweiler, D. P., Liberman,
A. M., Skudlarski, P., Fletcher, J . M., Katz, L., Marchione, K. E.,
Lacadie, C., Gatenby, C., and Gore, J . C. 1998. Functional disrup-
tion in the organization of the brain for reading in dyslexia. Proc.
Natl. Acad. Sci. USA 95: 2636–2641.
Souweidane, M. M., Kim, K. H. S., McDowall, R., Ruge, M. I., Lis, E.,
Krol, G., and Hirsch, J . 1999. Brain mapping in sedated infants
and young children with passive-functional magnetic resonance
imaging. Ped. Neurosurg. 30: 86–92.
Sowell, E. R., Thompson, P. M., Holmes, C. J ., Batth, R., J ernigan,
T. L., and Toga, A. W. 1999. Localizing age-related changes in
brain structure between childhood and adolescence using statisti-
cal parametric mapping. Neuroimage 9: 587–597.
Stapleton, S. R., Kiriakipoulos, E., and Mikulis, D., et al. 1997.
Combined utility of functional MRI, cortical mapping, and frame-
less stereotaxy in the resection of lesions in eloquent areas of brain
in children. Pediatr. Neurosurg. 26: 68–82.
Talairach, J ., and Tournoux, P. 1988. Co-planar Stereotaxic Atlas of
the Human Brain. Thieme Medical, New York.
Theodore, W. H. 1989. The role of fluorodeoxyglucose-positron emis-
sion tomography in the evaluation of seizure disorders. Semin.
Neurol. 9: 301–306.
Thomas, K. M., King, S. W., Franzen, P. L., Welsh, T. F., Berkowitz,
A. L., Noll, D. C., Birmaher, V., and Casey, B. J . 1999. A develop-
mental functional MRI study of spatial working memory. Neuro-
image 10: 327–338.
Vaidya, C. J ., Austin, G., Kirkorian, G., Ridlehuber, H. W., Desmond,
J . E., Glover, G. H., and Gabrieli, J . D. 1998. Selective effects of
methylphenidate in attention deficit hyperactivity disorder: A
functional magnetic resonance study. Proc. Natl. Acad. Sci. USA
Van Bogaert, P., David, P., Gillain, C. A., Wikler, D., Damhaut, P.,
Scalais, E., Nuttin, C., Wetzburger, C., Szliwowski, H. B., Metens,
T., and Goldman, S. 1998. Perisylvian dysgenesis. Clinical, EEG,
MRI and glucose metabolism features in 10 patients. Brain 121:
Volpe, J . J . 1995. Neurology of theNewborn, 3rd ed. W. B. Saunders.
Yakovlev, P. I., and Lecours, A. R. 1967. The myelenetic cycles of
regional maturation of the brain. In Regional Development of the
Brain in Early Life (A. Minkowski, Ed.), pp. 3–70. Davis, Phila-
Yamada, H., Sadato, N., Konishi, Y., Kimura, K., Tanaka, M.,
Yonekura, Y., and Ishii, Y. 1997. A rapid brain metabolic changein
infants detected by fMRI. NeuroReport 8: 3775–3778.
Ye, F. Q., Smith, A. M., Mattay, V. S., Ruttimann, U. E., Frank, J . A.,
Weinberger, D. R., and McLaughlin, A. C. 1998. Quantitation of
regional cerebral blood flow increases in prefrontal cortex during a
working memory task: A steady-state arterial spin-tagging study.
Neuroimage 8: 44–49.
Zerbe, G. O., Wu, M. C., and Zucker. 1994. Studying the relationship
between change and initial value in longitudinal studies. Stat.
Med. 13: 759–768.
DEVELOPMENTAL ASPECTS OF PEDIATRIC fMRI