Quantitative Magnetic Resonance Imaging
of Human Brain Development: Ages 4-18
Jay N. Giedd,1 John W. Snell,2 Nicholas Lange,3 Jagath C.
Rajapakse,1 B. J. Casey,1 Patricia L. Kozuch,1 A. Catherine
Vaituzis,1 Yolanda C. Vauss,1 Susan D. Hamburger,1 Debra
Kaysen1 and Judith L. Rapoport1
'National Institute of Mental Health, Child Psychiatry Branch,
Bethesda, MD 20892-1600, 2University of Virginia, Department
of Neurosurgery, VA and }National Institute of Neurological
Disorders and Stroke, Bethesda, MD, USA
Brain magnetic resonance images (MRI) of 104 healthy children and
adolescents, aged 4-18, showed significant effects of age and
gender on brain morphometry! Males had larger cerebral (9%) and
cerebellar (8%) volumes (P < 0.0001 and P = 0.008. respectively),
which remained significant even after correction for height and
weight After adjusting for cerebral size, the putamen and gjobus
pallidus remained larger in males, while relative caudate size was
larger in females. Neither cerebral nor cerebellar volume changed
significantly across this age range. Lateral ventricular volume
increased significantly in males (trend for females), with males
showing an increase in slope after age 11. In males only, caudate and
putamen decreased with age (P = 0.007 and 0.05, respectively). The
left lateral ventricles and putamen were significantly greater than the
right (/> = 0.01 and 0.0001, respectively). In contrast, the cerebral
hemispheres and caudate showed a highly consistent right-
greater-than-left asymmetry [P < 00001 for both). All volumes
demonstrated a high degree of variability. These findings highlight
gender-specific maturational changes of the developing brain and the
need for large gender-matched samples in pediatric neuro-
Surprisingly little is known about human anatomical brain
development between the ages 4 and 18. Mortality is low, with
accidents the leading cause of death, and autopsies are rarely
performed. This point is exemplified by the Yakovlev brain
collection in Washington, DC, in which only 12 of the 483
normal brains from the second embryonic week to the tenth
decade of life are from subjects aged 4-18 years (Haleem, 1990).
Although by age 2 the brain has reached 75% of its adult
weight (Carmichael, 1990) and the processes of synaptic
pruning and cell death are most active during these early years,
changes in brain structure and physiology continue throughout
life (Huttenlocher, 1979; Huttenlocher etal, 1982; Easter etal.,
1985; Kretschmann et al, 1986; Chugani et al, 1987). For
example, the associative neocortex continues to develop well
into the third decade (Yakovlev and Lecours, 1967), as does the
corpus callosum, which connects all major subdivisions of the
cerebrum (Pujol et al, 1993).
Magnetic resonance imaging (MRI), with its lack of ionizing
radiation and excellent anatomical resolution, provides an
unprecedented opportunity to obtain in vivo neuroanatomical
information of children and adolescents. To date, however, few
studies have been carried out for this group. One study of 39
subjects aged 8-35 (Jernigan et al, 1991) found an apparent
linear age-related decrease in cortical (frontal and parietal
regions) and subcortical structures (gray matter nuclei) and an
increase in ventricular volume across this age range. A second
study of 88 clinically referred subjects aged 3 months-30 years
found a steady increase in cortical white matter until the age of
20, with cortical gray matter volume peaking at age 4 and then
decreasing. Cortical and ventricular cerebrospinal fluid (CSF)
volumes remained constant (Pffererbaum et al, 1994). An
increase in intracranial volume of -300 ml was seen between 3
months and 10 years, with most of this increase occurring by the
age of 5. Finally, an MRI study of dyslexia that included 14
non-impaired children, aged 7-9, noted larger brain sizes in male
subjects and age-related increases in brain structure sizes
These studies provide important information about key
aspects of developmental neuroanatomy. Clearly, there are
substantial brain maturational changes in these years that may
reflect or predict normal behavioral development. However,
most childhood neuropsychiatric disorders are diagnosed and
followed between the ages of 4 and 18—an age range
underrepresented in these previous studies. Large data sets of
well-defined normal subjects are still needed to obtain accurate
quantification of the highly variable developmental changes of
children and adolescents.
This need is particularly relevant for ongoing MRI studies
addressing hypothesized subtle deviations in brain development
in children with severe neuropsychiatric disorders. Several gross
anatomic structures have already been implicated in a variety of
childhood-onset disorders including basal ganglia anomalies in
attention-deficit/hyperactivity disorder (ADHD) (Hynd et al,
1993; Castellanos etal, 1994), Sydenham's chorea (Giedd etal,
1995b) and Tourette's syndrome (Peterson etal, 1993; Singer et
al, 1993); midsagittal corpus callosum area differences in
dyslexia (Hynd etal, 1990, 1995; Duarzetal, 1991)andADHD
(Hynd etal, 1990, 1991; Giedd etal, 1994; Semrud-Clikeman
etal., 1994); and cerebral volume, planum temporale and
asymmetry differences in dyslexia and other learning disorders
(Galaburda etal, 1985; Rumsey etal, 1986; Duara etal, 1991;
Larsen et al, 1991; Galaburda, 1993; Kushch et al, 1993).
Interpretation of these studies has been limited by the small
sample sizes and lack of normative data.
To address this lack, and to assess normal brain maturational
changes, a large group of medically and psychiatrically healthy
children and adolescents were recruited from the local
community for participation in a quantitative MRI study. The
common practice of using as controls children referred clinically
for MRI and whose scans were subsequently read as normal
was avoided because children referred for clinical scans
are overrepresented in diagnostic groups, such as ADHD.
Conversely, some clinically normal children may have scans read
by radiologists as 'abnormal' and excluding these subjects would
confound statistical comparisons with diagnostic groups.
This initial report is the first of a series examining the
relationship between age, gender and brain morphometry in a
sample of >100 healthy children and adolescents. Based on
earlier studies, we anticipated that a number of late maturational
changes would be seen, such as decreases in subcortical nuclei
volumes and increases in ventricular volumes (Jernigan et al.
Cerebral Cortex July/Aug 1996:6:551-560; 1047-3211/96/J4.00
by guest on July 13, 2011
1991; Pfefferbaum et aL, 1994). We hoped that with our larger
sample size and more accurate methodology, effects of gender
and laterality could also be addressed.
Materials and Methods
From 624 responses to our local newspaper advertisements and postings,
234 were excluded by telephone screening due to personal or familial
histories of learning disorders, ADHD or ongoing medical or psychiatric
disorders. The remaining 390 were sent packets containing the Child
Behavior Checklist (Achenbach and Edelbrock, 1983), an NIH medical
history form and Conners' 48-item Parent Questionnaire (werry et aL,
1975; Goyette et aL, 1978). Conners' 39-item Teacher Questionnaires
were sent directly to the children's teachers. Based on this information,
187 children were excluded due to histories of learning disorders,
behavioral problems at home or school, or medical problems such as head
injury, migraines or use of medication. The remaining 203 were brought
into the clinic for a physical and neurological examination; the 12
handedness items from the Physical and Neurological Examination for
Subtle Signs (PANESS) inventory (Denckla, 1985); a clinical psychiatric
interview of the parents and child using the Child and Parent Diagnostic
Interview for Children (Welner etaL, 1987); a clinical interview of parent
and child by a board-certified child psychiatrist 0.N.G.) including family
history assessment; Vocabulary, Block Design, and Digit Span subtests of
the Wechsler Intelligence Scale for Children-Revised (WISC-R)
(Wechsler, 1974) for subjects 6-16 years of age or the Wechsler Adult
Intelligence Scale-Revised (WAIS-R) (Wechsler, 1981) for subjects aged
16 or older, spelling subtest of the Wide Range Achievement
Test—Revised (Juscak and Wilkinson, 1984); and reading achievement
cluster (consisting of letter-word identification, word attack, and passage
comprehension) of the Woodcock-Johnson Psycho-educational Battery
(Woodcock and Johnson, 1977). Individuals with physical, neurological
or lifetime histories of psychiatric abnormalities or learning disabilities,
or who had first-degree relatives or >20% of second-degree relatives with
major psychiatric disorders were excluded. Older siblings were removed
from the data set to maintain independence between subjects. One
hundred and twelve subjects met all of the above criteria and returned for
the scanning procedure. Four children (ages 5,7,8 and 11) who had been
accepted for the study were unable to complete the scan due to
claustrophobia or anxiety, and four scans had excessive motion artifact,
which prevented accurate measurement.
Fifty-five male and 49 female subjects (mean age = 11.6 years, SD = 35,
range 4.7-17.8 years) were included in this analysis. There were
significant male greater than female group differences for height (* =2.37,
P - 0.02) and Vocabulary subtest score of the WISC-R (t = 1.99, P • = 0.05),
and a trend for weight (t = 1.90, P = 0.06). There were no significant group
differences on age, handedness, Tanner stage, total academic score on the
Woodcock-Johnson test, or Digit Span and Block Design subtests of the
WISC-R. Subject characteristics are shown in Table 1. As can be seen, the
subjects were above average on Vocabulary, Block Design and Digit Span
subtests. Our strict inclusion criteria make this outcome likely, although it
does limit the generalizability of these findings.
The protocol was approved by the Institutional Review Board of the
National Institute of Mental Health. Written consent from the parents and
assent from the children were obtained.
All subjects were scanned on the same GE 1.5 tesla Signa scanner. Three
Ti-weighted three-dimensional image sets, with slice thickness of 1.5 mm
in the axial and sagittal planes and 2.0 mm in the coronal plane were
obtained using three-dimensional spoiled gradient recalled echo in the
steady state (3D SPGR). Imaging parameters were as follows: time to
echo, 5 ms; repetition time, 24 ms; flip angle, 45°; acquisition matrix, 192
x 256; number of excitations, 1; field of view, 24 cm. Vitamin E capsules,
wrapped in gauze and placed in the meatus of each ear, were used to help
standardize head placement. A third capsule was taped to the lateral
aspect of the left inferior orbital ridge. The vitamin E capsules are readily
identifiable on the scans and were used to define a reference plane for our
images. The patient's head was aligned in a head holder so that a narrow
Characteristics of healthy MRI subjects, ages 4-18
"Subtests of the Wechsler Intelligence Scale for Children—Revised.
"P < 0.05 difference between genders.
guide light passed through each of the vitamin E capsules. Foam padding
was placed on both sides of the patient's head to minimize head
movement. A sagittal localizing plane was acquired and from this a
multi-echo axial series to ensure that one of the axial slices contained all
three of the capsules. If no slice clearly contained all three capsules, the
patient was realigned until this criterion was met. Alignment in the
remaining plane was standardized by having the subject's nose at the
'12:00' position. Subjects were scanned in the evening to promote their
falling asleep in the scanner. Younger children were allowed to bring
blankets or stuffed animals into the scanner and have their parents read to
them. No sedation was used.
All scans were evaluated by a clinical neuroradiologist. Two subjects
were found to have clinically insignificant increased T2 signal intensities:
one in the area of the left semiovale and the other in the right parietal
lobe. They were retained in the data set. No other gross abnormalities
Cerebrum and Cerebellar Quantification
A novel image analysis technique developed by one of the authors 0.W.S.)
was employed to separate the image of the brain from the surrounding
intracranial cavity and to quantify cerebrum and cerebellar volumes. This
technique used an active surface template of a brain to supplement the
sometimes ambiguous MR imaging characteristics. The elastically
deformable model of the brain surface was conformed to the features of
the input image through successive iterations of an energy minimization
function (Fig. 1, top and bottom). The resulting image was then hand
edited by experienced raters examining each axial slice and removing
artifacts related to patches of eyeball or dura. Intraclass correlations for
the volumes of the edited brains were 0.99 for interrater reliability and
0.92 by comparison to volumes derived from more conventional
slice-by-slice hand tracing through all axial slices on which brain matter is
visible. The technique can be qualitatively evaluated by comparison with
a post-mortem specimen (Fig. 2). Further details are provided elsewhere
Spatial orientation was standardized by rotating the brains in three
dimensions so that operator-selected midline anterior and posterior
commissure points were in the same axial plane and that this plane was
perpendicular to an operator-selected midsagittal plane. Once each
brain's spatial orientation was standardized, the left and right cerebral
hemispheres were further subdivided into five regions based on internal
landmarks. The boundaries for the regional brain volumes are shown in
Region I ('prefrontal' sector) consists of all brain matter in front of the
anteriormost point of the corpus callosum. Region II ('premotor/
temporal' sector) is bounded by the coronal plane intersecting the
anteriormost point of the corpus callosum and the coronal plane
intersecting the anterior commissure (AC). Region III ("precentral/
552 MRI of Human Brain Development: Ages 4-18 * Giedd ct al.
by guest on July 13, 2011
Figure 1. (Top) The brain active surface template deforms in response to the image data set such that the mode! surfaces are brought into correspondence with the brain. Curvature
and topology constraints are used to overcome low-contrast boundary ambiguities. [Bottom) The final surface configuration of the active surface template is used to segment and
quantify the left and right cerebrum and cerebellum.
Cerebral Cortex Juty/Aug 1996, V6N4 553
by guest on July 13, 2011
temporal' sector) is between the anterior and posterior commissures
(PC). The two posterior regions—region IV ('parietal/temporal' sector)
and region V ('occipital' sector)—are arbitrarily divided by a plane 1.5
times the AC-PC length posterior to the PC. The descriptive names are in
quotations to emphasize that they are used only for the sake of
communication and are not based on sulcal/gyral patterns or
cytoarchitectonic information, and thus should be interpreted only as
containing 'mostly' prefrontal tissue, or 'mostly' premotor and temporal
tissue, and so on.
Lateral Ventricle Quantification
Lateral ventricular volumes were measured in the coronal plane on all
slices in which they were visible using an operator-supervised
thresholding technique, which segmented cerebrospinal fluid from brain
tissue (Rasband, 1993). Because this process required little subjectivity,
interrater reliability was extremely high (ICC > 0.99).
Figure 2. The active surface template method was validated by comparison with
post-mortem specimens. The brain was segmented from three-dimensional MRI of
whole cadaver heads [above right) and compared visually and volumetrically with
manual dissection results [above left).
Figure 3. Boundaries for cerebral subdivisions are defined by coronal planes
intersecting internal landmarks. Planes intersecting the anteriormost point of the genu of
the corpus callosum, the anterior commissure (AC) and the posterior commissure (PC)
were used to demarcate regions I. II and III (prefrontal, premotor/temporal and
precentral/temporal). Region IV (parietal/temporal) and region V (occipital) are arbitrarily
divided by a plane 1.5 times the AC-PC length posterior to the PC.
Subcortical Gray Matter Quantification
The caudate and putamen were manually outlined from coronal slices on
a Macintosh Ilfx workstation using NIH Image software (Rasband, 1993).
Since the sum of areas from the odd-numbered slices for the first 20
subjects correlated highly with the sum of the areas from the
even-numbered slices (ICC = 0.98), subsequent outlining was done on
every other slice and then multiplied by a slice thickness of 4 mm to
derive volume. Interrater reliability (ICC = 0.88 and 0.84 for the caudate
and putamen, respectively) was assessed initially and periodically during
the analyses to monitor potential 'drifts' in operator measurements.
Manual outlining of basal ganglia structures by experienced raters was
judged to be superior to a variety of automated techniques examined by
ANOVA and ANCOVA (adjusting for total cerebral volume) for brain structures by gender and side
in healthy children and adolescents, aged 4-18 (n = 104)
554 MRI of Human Brain Development: Ages 4-18 • Giedd el al.
by guest on July 13, 2011
The globus pallidus, bounded medially by the internal capsule and
laterally by the putamen, was a)so measured on coronal sections, but
included every slice, beginning 2 mm anterior to the anterior commissure
and proceeding posteriorly for a total of 14 mm. Limiting sampling to this
domain, which encompassed almost the entire globus pallidus in the
majority of subjects, was necessary to achieve adequate interrater
reliability (ICC = 0.82).
Because volumetric quantification of the thalamus was beyond our
current methodology, the thalamic area was outlined using a supervised
thresholding technique (Rasband, 1993) from a single midsagittal slice
reconstructed from the axial series. Reslicing from the axial series allows
more precise designation of the midsagittal plane than choosing a 'best'
midsagittal slice from the sagittal series. The intraclass correlation
coefficient of interrater reliability for the thalamic area was 0.85.
The SAS general linear model procedure was used to examine the
relationship between age, gender and brain morphology (SAS Institute,
1990). This included linear regression models for total group and
gender-specific effects of age on brain structure volumes. Since total
cerebral volume differed significantly between genders, gender
differences were analyzed using ANOVA and then ANCOVA to adjust for
total cerebral volume.
In addition, linearity and constant variance assumptions were relaxed
by use of a local regression procedure that retained the subtle
non-linearities in the data (a 'super-smoother'; see Hastie and Tibshirani,
1990) to yield smooth, curvilinear and gender-specific adaptive fits to the
scatterplots of structure volumes by age.
Combining male and female data in a single classical statistical model
usually makes linear and equal variance assumptions that were not always
supported by our data and could, in some cases, have yielded artifactual
results. We employed the local regression procedure as a descriptive
graphical tool and, for statistical inference, fitted classical linear and
piecewise linear regression models separately by gender. We have,
however, included results from combined analyses (Table 3) to enable
comparison of our results with previous reports.
Resists are summarized in Tables 2 and 3 and Figures 3 and 4. A
striking feature, evident from the scatterplots, is the high degree
of variability in brain structure size even for our well-screened,
healthy population. Consistent with a previous report,
subcortical nuclei volumes decrease and ventricular volume
increases (Jernigan et al., 1991). Not previously reported for this
age group, however, are the gender and laterality effects for both
volumes and maturational changes.
Table 2 shows ANOVAs for gender and side (left or right) and
ANCOVAs adjusting for total cerebral volume. Table 3 shows the
linear regression slopes with age, by gender and side, for the
Total (left plus right) volumes are presented as scatterplots,
with respect to gender and age, in Figures 4 and 5. Curvilinear
summaries for each gender are superimposed. Linear summaries
appear in the upper right portion of each plot. Notable
maturational changes are the increases in total ventricular
volume and decreases in caudate and putamen volumes, which
are significant only for males.
Robust gender effects were seen for several measures. Male
cerebral volumes were larger than female by 8.7% (F = 198, P <
0.0001). This effect remained after correction for height and
weight (F= 16.5, P< 0.0001), and was a fairly uniform difference
in that none of the cerebral subdivisions showed sexual
dimorphism when corrected for total cerebral volume. The
cerebellum was also larger (8.0%) in males (F= 5.4, P = 0.02). For
subcortical structures, the putamen and globus pallidus were
larger in males (i7 = 16.1, P = 0.001 and P = 8.0, P =0.006,
respectively), and remained so after adjusting for total cerebral
volume (F = 6.3, P = 0.01 and P = 4.1, P = 0.05). In contrast,
caudate was larger in females after adjustment for total cerebral
volume (F = 6.5, P = 0.01). The unadjusted volumes of the
caudate, lateral ventricles and thalamic area did not differ
Neither right, left, nor total cerebral or cerebellar volume
increased significantly with age for either gender. The regional
subdivisions of the cerebrum also did not show significant age
effects. Lateral ventricular volume increased with age (slope =
0.88 ml/year, P = 0.0007 and slope = 0.47 ml/year, P = 0.06 for
males and females, respectively). Interestingly, the increase for
males occurred almost entirely after the age of 11. A piecewise
linear model for males revealed a significant change in slope after
the age of 11 (P = 0.03) not shared by females at that or other
ages. Both caudate and putamen volume decreased in males
(slope = -0.01 ml/year, P = 0.007 and slope = -0.007 ml/year, P =
0.05, respectively), but not in females. Neither globus pallidus
Linear regression of brain structures with age by gender and side in healthy children and adolescents (n = 104)
Cerebral Cortex July/Aug 1996, V 6 N 4 555
by guest on July 13, 2011
. ; # ° ^ •
:y = -2.2x+1199.9
y = 5.9x + 1007.5
n = 104
•# o o
00 ° 0
6 8 10
12 14 16 18
M:y = 0.9x + 0.19
F : y = 0.47x + 4.05o
n = 104
12 14 16 18
y = 0.0008X + 0
y = 0.0004X + 0
o %// •
o # ***5w»--'#"
y = 0.25x +
y = 1.23x +
...&. «. o
- - • •
Ilm »n •
6 8 10
12 14 16 18
6 8 10 12 14 16 18
Hgure 4. Scatterplots by age and gender of total cerebral volume, ventricular volume, ventricle:brain ratio and cerebellar volume for children and adolescents (n = 104).
volume nor thalamic area changed significantly with age for
Several right/left asymmetries were highly significant (Table 2).
Right cerebral hemisphere and caudate volumes were larger than
left (f = 38.3, P < 0.0001 and F = 58.6, P < 0.0001, respectively),
whereas left lateral ventricles and putamen were larger than the
right (F = 6.5, P = 0.01 and F = 97.5, P < 0.0001, respectively).
There were no significant differences between genders for
these asymmetries. No asymmetry was seen for the cerebellar
hemispheres. Consistent with reports from the adult literature
(Bilder et aL, 1994), the anteriormost subdivision of the
cerebrum demonstrated a right-greater-than-lef t asymmetry (F =
4.8, P = 0.03).
A ventriclerbrain ratio, calculated from the lateral ventricle and
556 MM of Human Brain Dcvdopmenc Ages 4-18 • Gicdd et al.
by guest on July 13, 2011
M:y = -0.12x+11.64
F:y = 0.01x+10.17
0 * 0 o
—•°——^? 2 • ° t °
• »O # 0
° °o* °
* • °«8*°
o ? •
6 8 10
12 14 16 18
^ o *
y = 0.01x +
y = -0.01x +
* . °
' °* • lv-rr*o
•o • »o o
o • °
6 8 10
12 14 16 18
M:y = -0.08x+12.3
F:y = -0.05x +11.18
6 8 10 12 14 16 18
M:y = -3.1x+185.7
F:y = 1.2x+140o
° * • *
10° a o
0 0 o ° ^ \
: • . - • : • • • •
6 8 10 12 14 16 18
Rgure 5. Scatterplots by age and gender of subcortical gray matter volumes (right + left) for children and adolescents in = 104).
cerebral hemisphere measures, was nearly collinear with lateral
ventricular volume (Fig. 4) and did little to reduce variance of the
ventricular measures. It is presented for comparison with the
wide body of literature related to this measure.
A number of findings emerge from these data confirming and
extending previous reports for pediatric subjects. As expected,
there are age-related decreases in caudate and putamen and
increase in ventricular volume. As reported for adult samples
(Breier et aL, 1992; Flaum et al, 1995), there is a highly
significant right-greater-than-left asymmetry of the caudate. In
addition, gender-specific maturational effects were noted for
The larger sample size in this study permitted closer
examination of sexual dimorphism than possible in earlier
reports. For instance, a previous study of 23 males and 16
females, aged 8-35, showed a decrease with age for the caudate
and lenticular nucleus, but no effect for gender (Jernigan et aL,
1991). The present study demonstrated similar declines in
Cerebral CortexJuly/Aug 1996, V6 N 4 557
by guest on July 13, 2011
caudate and lenticular volume (derived by summing putamen
and globus pallidus for this comparison), but only for males.
More global gender effects, e.g., larger male brains with no
regional differences when adjusting for total brain volume, are
also consistent with previous reports (Jernigan et al, 1991;
Pfefferbaum efa/:, 1994).
Sexual dimorphism of brain structures may be related to the X
chromosome, hormonal effects, environmental effects or a
combination of these. A recent study of Turner's syndrome
(Murphy et al, 1993) suggested that the X chromosome is
involved in determining the adult size of the caudate, lentiform
nucleus, thalamus and gray matter of the cerebral cortex.
Hormonal effects seem to be instrumental in overall brain size
and cerebral asymmetries (Kelley, 1993).
The larger size of the male brain in this age group parallels
both autopsy (Blinkov and Glezer, 1968; Ho et al, 1980) and
imaging studies (Andreasen et al, 1993; Filipek et al, 1994;
Pfefferbaum et al, 1994; Schultz et al., 1994). Although there are
several smaller structures that are thought to be sexually
dimorphic (anterior commissure, corpus callosum, and certain
thalamic nuclei), at a macroscopic level the larger size of the
male cerebrum appears relatively uniform, as none of the
regional subdivisions showed sexual dimorphism after
correction for total cerebral volume. Of course, gross structural
size may not be sensitive to sexually dimorphic differences in
connectivity between different neurons, known differences in
receptor density, or more subtle differences in the size or
connectivity of various nuclei. Given the multiple parameters
determining brain size, a larger size should not be interpreted as
imparting functional advantage or disadvantage.
The lack of increase in total cerebral size across this age range
is consistent with available post-mortem data arid previous MRI
studies (Jernigan et al, 1991; Pfefferbaum et al, 1994),
indicating a leveling off in total brain size at -5 years (Kretsch-
mann et al, 1986), although other investigators reported an
-100 cm3 increase from ages 5 to 18 (Blinkov and Glezer, 1968).
Head circumference increases by -2.0 in. in boys and -1.9 in. in
girls from ages 5 to 18 (Nellhaus, 1968), which may be
accounted for by the increase in ventricular volume and/or skull
thickening that is known to occur across this age range (Shapiro
and Janzen, I960). The phenomenon of a progressively
decreasing brain:body weight ratio during development is well
described in the literature (Dekaban and Sadowsky, 1978),
although a complete lack of increase in brain size during this
developmental period is perhaps surprising.
While the age-related changes in the cerebellum, caudate and
putamen are of interest, the interpretation of these changes is
not clear. The size of brain structures is determined by the
number, size and packing density of constituent cells, namely
neurons and glial cells. Like the nervous systems of other higher
vertebrates, human brain development takes place by an
overproduction and then selective elimination of cells, with the
number of neurons reaching its maximum in utero (Rabinowicz,
1986). The balance between cell proliferation and cell death
during neurogenesis largely accounts for the total number of
neurons. For a short time after birth, certain types of neurons,
granule cells in the cerebellum, olfactory bulb, hippocampal
dentate gyrus and brainstem nuclei may proliferate, but these
account for a small fraction of the total number of neurons
(Jacobson, 1991). Individual neurons undergo many cyclical
changes in size throughout development (Thatcher, 1992), but in
general enlarge with age (Blinkov and Glezer, 1968). As
neighboring neurons are lost through apoptosis or cell death, the
remaining neurons sprout greater numbers of dendrites, axons
become thicker and the number of synaptic boutons increases.
Axonal and dendritic changes continue throughout life and are
presumably involved in the mechanisms by which we learn
Glial cells outnumber neurons, with reports of glial
cell:neuron ratios ranging from 1.7 to 10 (Brizzee et al, 1964).
Unlike neurons, glial cells undergo a constant cycle of
proliferation and cell death. The relationship between glial cell
volume and the size, number or activity of neurons is poorly
understood, although both metabolic activity and neuronal cell
death are thought to influence glial proliferation (Jacobson,
1991). Myelination by oligodendrocytes is the activity of glial
cells most influential in determining brain size during this age
range. Myelination continues actively at least through the first
decade (Yakovlev and Lecours, 1967) and longer in certain parts
of the brain, such as the superior medullary lamina along the
surface of the parahippocampal gyrus, where there is a doubling
in the extent of myelination relative to brain weight between the
first and second decades, and an additional 60% increase
between the fourth and sixth decades (Benes et al, 1994).
The balance between decreasing numbers of neurons and
increasing size of neurons and glial cells, largely attributable to
myelination, is primarily responsible for determining the overall
size of the brain and its components. Synaptic pruning alone,
despite its ongoing activity during this age period, is less likely to
be a major factor in overall structure size. Based on work
involving the primary visual cortex of the macaque monkey
(Bourgeois and Rakic, 1993), it is estimated that even a total loss
of boutons would account for only a 1-2% decrease in volume.
However, the effect synaptic pruning has on the remaining
thickness of the parent axon or dendritic branches has yet to be
determined. Another parameter in structure size is packing
density, which is influenced by hydration, extracellular volume
and degree of vascularity.
Neurons, glial cells and packing density are, in turn, affected
by many factors, including genetics, hormones, growth factors
and nutrients in the developing nervous system (Jacobson,
1991). In addition, diet and other external factors such as
infections, toxins, trauma, stress or degree of enriched
environment (Diamond et al, 1964) may also have a role in
determining structure size. The extent to and the mechanisms
by which neuropsychiatric disorders affect these parameters
must be part of future investigations.
The caudate and putamen, which decrease significantly in
size with age for males only, and the striking ventricular
enlargement found in males exemplify important gender-by-age
differences. These sexually dimorphic effects are of great
interest in normal development as they occur in regions
implicated in various neuropsychiatric disorders that also have
male preponderance (Hynd et al, 1993; Peterson et al, 1993;
Castellanos et al, 1994; Giedd et al, 1994). Thus, sexually
dimorphic brain maturational changes may interact with other
unknown pathological influences, providing the striking sex
differences seen in most pediatric behavioral disorders.
It is unclear from our data to what degree ventricular
enlargement is at the expense of surrounding tissue.
Nevertheless, in light of the frequent interpretation of increased
ventricle:brain ratio as a general measure of cerebral damage it is
noteworthy that such an increase is an integral part of normal
pediatric development. It is also of note that the use of this ratio
does not decrease the variability when compared to the
variability in lateral ventricular volume alone.
558 MR] of Human Brain Development: Ages 4-18- Giedd et al.
by guest on July 13, 2011
Asymmetries of both the caudate and putamen are small in
degree (3.2% for the caudate and 130% for the putamen), but
highly consistent, occurring in the dominant direction in 85% of
the subjects for both structures. The right-greater-than-left
caudate volume is in keeping with several studies of normal
adults using large samples (Breier et al, 1992, Flaum et al,
1995). With growing awareness of more complex cognitive and
motor functions of the basal ganglia (Graybiel et al., 1994),
further evidence of develop- mental complexity in man is of
considerable interest. We will be examining asymmetries in
relation to motor and cognitive development in future analyses.
The methodological limitations of cross-sectional study
designs for developmental studies should be noted. Longitudinal
studies, encouraged by findings of high rescan reliability for
quantification of brain structure sizes (Giedd et al, 1995a), are
currently underway. The enormous variability of brain structure
sizes noted in this population, and the heterochronous nature
of most developmental curves, necessitates large samples to
characterize neuroanatomic changes in human brain develop-
ment. Correction for height, weight or total brain size only
partially reduces the variability. The relative merits of using
actual sizes or sizes corrected for total brain volume are a source
of considerable debate (Arndt et al, 1991), although both
approaches have potential utility in elucidating form/function
relationships in the brain.
The quantification of natural, gender-specific variability and
covariability provided by our study is in itself a useful tool in
on-going efforts to identify and to characterize greater than
expected inter- and intrasubject deviations in brain structure
sizes. The robust maturational effects detectable across this
clinically relevant age range will be important comparison
measures in studies of neurodevelopmentally impaired children.
For example, in males with ADHD, age-related decreases in
caudate volume were not observed and increases in ventricular
enlargement were significantly diminished (Castellanos et al,
It is anticipated that the interaction of gender effects and
specific disease processes will be a major contribution of this
large normative study in our ongoing comparative studies of
pediatric neuropsychiatric disorders.
We acknowledge Dan Dickstein and Wendy Marsh for assistance in image
analysis and data management, and Wayne Rasband for customizing
image analysis software. This work was supported in part by a grant from
the Gulton Foundation.
Address correspondence to Dr Jay N. Giedd, National Institute of
Mental Health, Child Psychiatry Branch, Building 10, Room 6N24O, 10
Center Drive MSC 1600, Bethesda, MD 20892-1600, USA.
Achenbach TM, Edelbrock CS (1983) Manual for child behavior checklist
and revised behavior profile. Burling, VT: Department of Psychiatry,
University of Vermont.
Andreasen NC, Flaum M, Swayze VS, O'Leary DS, Alliger R, Cohen G,
Ehrhardt J, Yuh TC (1993) Intelligence and brain structure in normal
individuals. Am J Psychiat 150:130-134.
Arndt S, Cohen G, Alliger RJ, Swayze VW, Andreasen NC (1991) Problems
with ratio and proportion measures of imaged cerebral structures.
Bull Clin Neurosci 55:131-136.
Benes FM, Turtle M, Khan Y, Farol P (1994) Myelination of a key relay
zone in the hippocampal formation occurs in human brain during
childhood, adolescence and adulthood. Arch Gen Psychiat
Bilder RM, Wu H, Bogerts B, Degreef G, Ashtari M, Alvir JM, Snyder PJ,
Lieberman JA (1994) Absence of regional hemisphere volume
asymmetries in first episode schizophrenia. Am J Psychiat
Blinkov SM, Glezer II (1968) The human brain in figures and tables. A
quantitative handbook. New York: Plenum.
Bourgeois JP, Rakic P (1993) Changes of synaptic density in the primary
visual cortex of the macaque monkey from fetal to adult stage. J
Breier A, Buchanan RW, Elkashef A, Munson RC, Kirkpatrick B, Gellad F
(1992) Brain morphology and schizophrenia. Arch Gen Psychiat
Brizzee KR, Vogt J, Kharetehko X (1964) Postnatal changes in glia neuron
index with a comparison of methods of cell enumeration in the white
rat. Prog Brain Res 4:136-149.
Carmichael A (1990) Physical development and biological influences. In:
Handbook of studies in child psychiatry (Tonge B, Burrows GD,
WerryJS, eds). Amsterdam: Elsevier.
Castellanos FX, Giedd JN, Eckburg P, Marsh WL, King AC, Hamburger SD,
Rapoport JL (1994) Quantitative morphology of the caudate nucleus
in attention-deficit hyperactivity disorder. Am J Psychiat
Castellanos FX, Giedd JN, Marsh WL, Hamburger SD, Vaituzis AC, Snell
JW, Lange N, Dickstein DP, Vauss YC, Kaysen D, Ritchie GF, Rajapakse
JC, Rapoport JL (1996) Quantitative brain magnetic resonance
imaging in attention-deficit/hyperactivity disorder. Arch Gen
Psychiat, in press.
Chugani HT, Phelps ME, Mazziotta JC (1987) Positron emission
tomography study of human brain functional development. Ann
Dekaban AS, Sadowsky D (1978) Changes in brain weight during the span
of human life: relation of brain weights to body heights and body
weights. Ann Neurol 4:345-356.
Denckla MB (1985) Revised physical and neurological examination for
subtle signs. Psychopharmacol Bull 21:773-800.
Diamond MC, Krech D, Rosenzweig MR (1964) The effects of an enriched
environment on the histology of the rat cerebral cortex. J Comp
Duara R, Kushch A, Gross-Glenn K, Barker WW, Jallad B, Pascal S,
Lowenstein DA, Sheldon J, Rabin M, Levin B, Lubs H (1991)
Neuroanatomic differences between dyslexic and normal readers on
magnetic resonance imaging scans. Arch Neurol 48:410-416.
Easter SS, Purves D, Rakic P, Spitzer NC (1985) The changing view of
neural specificity. Science 230:507-511.
Filipek PA, Richelme C, Kennedy DN, Caviness VS (1994) The young adult
human brain: an MRI-based morphometric analysis. Cereb Cortex
Flaum M,' Swayze VW, O'Leary DS, Yuh WTC, Ehrhardt JC, Arndt SV,
Andreasen NC (1995) Effects of diagnosis, laterality, and gender on
brain morphology in schizophrenia. Am J Psychiat 152:704-714.
Galaburda AM (1993) Dyslexia and development. London: Harvard
Galaburda AM, Sherman GF, Rosen GD, Aboitiz F, Geschwind N (1985)
Developmental dyslexia: four consecutive patients with cortical
anomalies. Ann Neurol 18:222-233.
Giedd JN, Castellanos FX, Casey BJ, Kozuch PL, King AC, Hamburger SD,
Rapoport JL (1994) Quantitative morphology of the corpus callosum
in attention deficit hyperactivity disorder. Am J Psychiat 151:
Giedd JN, Kozuch P, Kaysen D, Vaituzis AC, Hamburger SD, Bartko JJ,
Rapoport JL (1995a) Reliability of cerebral measures in repeated
examinations with magnetic resonance imaging. Psychiat Res
Giedd JN, Rapoport JL, Kruesi MJP, Parker C, Schapiro MB, Allen AJ,
Leonard HL, Kaysen D, Dickstein D, Marsh W, Kozuch PL, Vaituzis AC,
Hamburger SD, Swedo SE (1995b) Sydenham's chorea: magnetic
resonance imaging of the basal ganglia. Neurology 25:11 -14.
Goyette CH, Conners CK, Ulrich RF (1978) Normative data on the Revised
Conners' Parent and Teacher Rating Scales. J Abnorm Child Psychol
Graybiel AM, Aosaki T, Flaherty AW, Kimura M (1994) The basal ganglia
and adaptive motor control. Science 265:1826-1831.
Haleem M (1990) Diagnostic categories of the Yakovlev collection of
normal and pathological anatomy and development of the brain.
Washington, DC: Armed Forces Institute of Pathology.
Hastie T, Tibshirani R (1990) Generalized additive models. London:
Chapman & Hall.
Cerebral CortexJuly/Aug 1996, V6N 4 559
by guest on July 13, 2011
Ho KC, Roessmann U, Straumijord JV, Monroe G (1980) Analysis of brain
weight. I. Adult brain weight in relation to sex, age, and race. Arch
Pathol Lab Med 104:635-639.
Huttenlocher PR (1979) Synaptic
cortex—developmental changes and effects of aging. Brain Res
Huttenlocher PR, de Courten C, Garey LJ, Van Der Loos H (1982)
Synaptogenesis in human visual cortex—evidence for synapse
elimination during normal development. Neurosci Lett 33:247-252.
Hynd GW, Semrad-Clikeman M, Lorys AR, Novey ES, Eliopulos D (1990)
Brain morphology in developmental dyslexia and attention deficit
disorder/hyperactivity. Arch Neurol 47:919-926.
Hynd GW, Semrud-Clikeman M, Lorys AR, Novey ES, Eliopulos D,
Lyytinen H (1991) Corpus callosum morphology in attention
deficit-hyperactivity disorder morphometric analysis of MRI. J Learn
Hynd GW, Hern KL, Novey ES, Eliopulos D, Marshall R, GonzalezJJ (1993)
Attention deficit hyperactivity disorder (ADHD) and asymmetry of
the caudate nucleus. J Child Neurol 8:339-347.
Hynd GW, Hall J, Novey ES, Eliopulos D, Black K, Gonzales JJ, Edmonds
JE, Riccio C, Cohen M (1995) Dyslexia and corpus callosum
morphology. Arch Neurol 52:32-38.
Jacobson M (1991) Developmental neurobiology. New York: Plenum.
Jastak S, Wilkinson GS (1984) Wide range achievement test, revised
edition. Wilmington, DE: Jastak Assessment Systems.
Jernigan TL, Trauner DA, Hesselink JR, Tallal PA (1991) Maturation of
human cerebrum observed in vivo during adolescence. Brain
Kelley DB (1993) Androgens and brain development: possible
contributions to developmental dyslexia. In: Dyslexia and
development. Neurobiological aspects of extra-ordinary brains
(Galaburda AM, ed), pp 21-41. Cambridge, MA: Harvard University
Kretschmann HJ, Kammradt G, Krauthausen I, Sauer B, Wingert F (1986)
Brain growth in man. Bibl Anat 28:1-26.
Kushch A, Gross-Glenn K, Jallad B, Lubs H, Rabin M, Feldman E, Duara R
(1993) Temporal lobe surface area measurements on MRI in normal
and dyslexic readers. Neuropsychologia 31:811-821.
Larsen JP, Hoein T, Lundberg I, Odegaard H (1990) MRI evaluation of the
size and symmetry of the planum temporale in adolescents with
developmental dyslexia. Brain Lang 39:289-301.
Leonard CM, Voeller KKS, Lombardino LJ, Morris MK, Hynd GW,
Alexander AW, Andersen HG, Garofalakis M, Honeyman JC, Mao J,
Agee OF, Staab EV (1993) Anomalous cerebral structure in dyslexia
revealed with magnetic resonance imaging. Arch Neurol 50:461-469.
Middleton FA, Strick PL (1994) Anatomical evidence for cerebellar and
basal ganglia involvement in higher cognitive function. Science
Murphy DG, DeCarli C, Daly E, Haxby JV, Allen G, White BJ, Mclntosh
AR, Powell CM, Horwitz B, Rapoport SI (1993) X-chromosome effects
on female brain: a magnetic resonance imaging study of Turner's
syndrome. Lancet 342:1197-1200.
Nellhaus G (1968) Head circumference: girls and boys 2-18 years.
density in human frontal
Peterson B, Riddle MA, Cohen DJ, Katz LD, Smith JC, Hardin MT,
Leckman JF (1993) Reduced basal ganglia volumes in Tourette's
syndrome using three-dimensional reconstruction techniques from
magnetic resonance images. Neurology 43:941-949.
Pfefferbaum A, Mathalon DH, Sullivan EV, Rawles JM, Zipursky RB, Iim
KO (1994) A quantitative magnetic resonance imaging study of
changes in brain morphology from infancy to late adulthood. Arch
Pujol J, Vendrell P, Junque C, Marti-Vilalta JL, Capdevila A (1993) When
does human brain development end? Evidence of corpus callosum
growth up to adulthood. Ann Neurol 34:71-75.
Rabinowicz T (1986) The differentiated maturation of the cerebral
cortex. In: Human growth (Falkner F, Tanner JM, eds), Vol 2, pp
385-410. New York: Plenum.
Rasband W (1993) Image, version 1.6. Bethesda, MD: National Institutes
of Health (Public Domain).
Rumsey JM, Dorwart R, Vermess M, Denckla MB, Kruesi MJP, Rapoport
JL (1986) Magnetic resonance imaging of brain anatomy in severe
developmental dyslexia. Arch Neurol 43:1045-1046.
SAS Institute (1990) SAS, version 6. Cary, NC: SAS Institute.
Schultz RT, Cho NK, Staib LH, Kier LE, Fletcher JM, Shaywitz SE,
Shankweiler DP, Katz L, Gore JC, Duncan JS, Shaywitz BA (1994)
Brain morphology in normal and dyslexic children: the influence of
sex and age. Ann Neurol 35:732-742.
Shapiro R, Janzen AH (I960) The normal skull. New York: Hoeber.
Singer HS, Reiss AL, Brown JE, Aylward EH, Shih B, Chee E, Harris EL,
Reader MJ, Chase GA, Bryan RN, Denckla MB (1993) Volumetric MRI
changes in basal ganglia of children with Tourette's syndrome.
Snell JW, Merickel M, Ortega J, Goble J, Brookeman J, Kassell N (1995)
Boundary estimation of complex objects using hierarchical active
surface templates. Pattern Recognit 28:1599-1609.
Thatcher RW (1992) Cyclic cortical reorganization during early
childhood. Brain Cognit 20:24-50.
Wechsler D (1974) Wechsler Intelligence Scale for Children-Revised.
New York: Psychological Corporation.
Wechsler D (1981) Wechsler Adult Intelligence Scale-Revised. New York:
Welner Z, Reich W, Herjanic B, Jung K, Amado H (1987) Reliability,
validity and child agreement studies of the diagnostic interview of
children and adolescents (DICA). J Am Acad Child Adoles Psychiat
Werry JS (1991) Brain and behavior. In: Child and adolescent psychiatry:
a comprehensive textbook (Lewis M, ed), pp 76-85. Baltimore, MD:
Werry JS, Sprague RL, Cohen MN (1975) Conners' Teacher Rating Scale
for use in drug studies with children—an empirical study. J Abnorm
Child Psychol 3:217-229.
Woodcock RW, Johnson BB (1977) Woodcock-Johnson psycho-
educational battery. Allen, TX: DLM Teaching Resources.
Yakovlev PI, Lecours AR (1967) The myelogenetic cycles of regional
maturation of the brain. In: Regional development of the brain in early
life (Minkowski A, ed), pp 3-70. Oxford: BlackweU.
560 MRlof Human Brain Development: Ages 4-18 • Giedd a aL
by guest on July 13, 2011