matter changes from longitudinal studies are lacking. We quantified longitudinal magnetic resonance imaging (MRI) scans of 92 non-
demented older adults (age 59–85 years at baseline) in the Baltimore Longitudinal Study of Aging to determine the rates and regional
distribution of gray and white matter tissue loss in older adults. Using images from baseline, 2 year, and 4 year follow-up, we found
tissue loss were 5.4 ? 0.3, 2.4 ? 0.4, and 3.1 ? 0.4 cm3per year for total brain, gray, and white volumes, respectively, and ventricles
and occipital, lobar regions showed greater decline. Gray matter loss was most pronounced for orbital and inferior frontal, cingulate,
Stereological studies of neuron counts in the human brain sug-
Hof, 1997; Pakkenberg and Gundersen, 1997). The conclusions
drawn from these postmortem studies, which are necessarily
cross-sectional, stand in contrast to data from in vivo imaging
studies, which reveal age differences in brain volumes and CSF
spaces (Gur et al., 1991; Coffey et al., 1992; Pfefferbaum et al.,
in interpreting both the postmortem and in vivo imaging studies
is the cross-sectional nature of the majority of investigations.
secular changes in nutrition, medical care, and other factors. In-
deed, secular drifts in body and brain weight have been docu-
mented (Miller and Corsellis, 1977), and rates of brain changes
can only be determined from longitudinal investigations.
An advantage of the use of in vivo imaging to study brain
By controlling for variability between individuals, within-
individual comparisons can identify subtle changes over time.
There have been few longitudinal neuroimaging studies, and
white matter using high-resolution magnetic resonance imaging
(MRI). One longitudinal MRI study reported puzzling findings
of longitudinal increases in some brain volume measures and
decreases in subarachnoid CSF over a 3–6 year interval (Mueller
et al., 1998). Low reliability (inter-rater agreement of 0.71) for
reported a 2.1% annual rate of cerebral volume loss over 4.4. years,
using stereological techniques to estimate a global measure of cere-
Since 1994, we have been conducting a longitudinal brain
imaging study of older adults in the Baltimore Longitudinal
Study of Aging (BLSA) to identify brain changes that may be
predictors of cognitive decline and Alzheimer’s disease. High-
resolution MRI (1.5 mm thickness) and a validated approach for
for quantitation of global and regional gray and white volumes.
We hypothesized that loss of both gray and white matter would
(Resnick et al., 2000), we hypothesized that ventricular volume
would continue to increase over the course of the study. We also
examined the effects of age and sex on the rates of change for
provide the first evidence of substantial longitudinal loss of both
in ventricular CSF (V-CSF). The rate of increase in V-CSF is
influenced by age but is similar in male and female older adults.
Subjects. The present sample includes 92 participants (50 men, 42
women) in the neuroimaging study of the BLSA (Resnick et al., 2000)
TheJournalofNeuroscience,April15,2003 • 23(8):3295–3301 • 3295
who completed a baseline MRI study and assessments at year 3 (2 year
interval) and year 5 (4 year interval). Three additional participants were
excluded from these analyses because they had developed serious CNS
pathology (but remained free of dementia). Neuroimaging participants
in this sample are a subset of BLSA volunteers, aged 59–85 at baseline,
exclusionary criteria at initial evaluation: CNS disease [epilepsy, stroke,
bipolar illness, previous diagnosis of dementia according to Diagnostic
and Statistical Manual (DSM)-III-R criteria (Spitzer and Williams,
1987)], severe cardiovascular disease (myocardial infarction, coronary
artery disease requiring angioplasty or bypass surgery), severe pulmo-
decline that did not meet criteria for dementia (n ? 4) and those with
past or current depression (n ? 8) were included, because these factors
may be risk factors for dementing illness. All participants remained free
of dementia at year 5 follow-up, using diagnostic procedures described
previously (Kawas et al., 2000). Demographic characteristics and mea-
sures of functional status are presented in Table 1 for the entire sample
and for a subgroup of 24 participants who remained very healthy (no
medical conditions or cognitive impairment) at year 5 evaluation. This
research protocol was approved by the local institutional review board,
junction with each neuroimaging visit.
Image acquisition. MR acquisition procedures are detailed in Resnick
et al. (2000). MR scanning was performed on a GE Signa 1.5 Tesla scan-
ner. The current results are based on a high-resolution volumetric
“spoiled grass” (SPGR) series (axial acquisition; repetition time ? 35;
echo time ? 5; flip angle ? 45; field of view ? 24; matrix ? 256 ? 256;
number of excitations ? 1; voxel dimensions of 0.94 ? 0.94 ? 1.5 mm
slice thickness). Two similarly configured scanners were used inter-
changeably over the course of the data collection, which spanned from
LX software upgrade. Quality control scans were performed daily.
Image analysis. Quantitative analysis of MR volumes is accomplished
using a semi-automated approach, with demonstrated validity and high
reliability (Goldszal et al., 1998). Images are first reformatted parallel to
the plane containing the anterior and posterior commissures. Extracra-
millary bodies are removed by a single, highly experienced, image-
processing technician using a semi-automated procedure, with high
inter-rater and test–retest reliability (Goldszal et al., 1998). The remain-
ing tissue is classified, using an adaptive Bayesian segmentation algo-
rithm (Yan and Karp, 1995), into gray matter, white matter, and CSF.
V-CSF is defined by drawing a crude region of interest to mask out any
nonventricular CSF. After this step, all image processing is fully auto-
mated and operator independent. The segmented images provide quan-
titative volumetric measures of total gray, white, and brain (gray ?
white) matter, as well as V-CSF volumes and ventricle-to-brain ratios
(VBRs). Stabilities over 1 year for these measures are all ?0.95. Sulcal
CSF is not quantified, because the interface between CSF and the cra-
nium is difficult to determine reliably on SPGR images. Moreover, our
Regional analysis of volumes examined in normalized space, the
RAVENS approach (Goldszal et al., 1998), is used for quantification of
regional volumes and investigation of local brain changes. The seg-
mented images are transformed to the Talairach stereotaxic coordinate
space (Talairach and Tournoux, 1988), using the elastic deformation
algorithm of Davatzikos (1996). This approach allows quantitation of
absolute volumes within the standard reference space and applies a
boundary constraint for the ventricles, which are often enlarged in the
elderly. Volumes of frontal, parietal, temporal, and occipital brain re-
Talairach coordinate space (Andreasen et al., 1996), yielding stabilities
over 1 year between 0.86 and 0.97 (Resnick et al., 2000). In addition, the
RAVENS approach yields average brain maps for analysis of local differ-
ences between groups of subjects or longitudinal change. Image intensi-
ties reflect the amount of expansion or contraction relative to a template
brain and correspond to the distribution of gray, white, or CSF volumes
in the average RAVENS maps. For example, if a subject’s ventricles are
larger relative to other subjects, this subject’s RAVENS V-CSF map will
be relatively brighter. The same holds for gray matter and white matter
structures. Accordingly, regional volumetric measurements are per-
ences in intensities over time reveal local longitudinal changes.
Statistical analysis. Statistical analysis was performed using SAS Ver-
sion 6.12 on a DEC ? computer running OpenVMS. Two and 4 year
stabilities were estimated by Pearson product–moment correlations.
Mixed-effects regression was used to investigate longitudinal changes
changes for individual brain regions. Separate analyses were conducted
for total brain, gray, and white matter volumes, frontal, parietal, tempo-
ral, and occipital volumes, ventricular volumes, and VBR. Height was
entered as a covariate for volume but not ratio measures. Longitudinal
change was included in each model by addition of time (baseline, year 3,
year 5) as a fixed-effects term. In addition, all two-way interactions,
excluding those with height, and three-way interactions of interest were
included in the initial model. A backward elimination procedure was
used, whereby all lower-order terms remained in the final model but
were calculated from the slope of volume versus age at assessment, pro-
viding estimates that can be compared across the various brain volume
measurements. Finally, the differential effects of age and sex across dif-
ferent tissue types (i.e., gray versus white), brain regions (i.e., frontal,
parietal, temporal, occipital), and hemispheres (right, left) were exam-
ined using multivariate ANOVA (MANOVA). In these analyses, sex and
age (?70 years versus 70–85 years at baseline) were grouping factors;
time (baseline, year 3, year 5) and tissue type or region were repeated-
yses, with the analysis of lobar volumes including both gray and white
matter. Two sets of additional analyses were performed. The first ex-
cluded the four subjects with mild cognitive impairment. In the second,
not meet criteria for mild cognitive impairment.
Statistical analysis of local gray and white matter volume changes was
performed using the RAVENS images and voxel-based paired t tests as
3296 • J.Neurosci.,April15,2003 • 23(8):3295–3301 Resnicketal.•BrainAginginOlderAdults
1995). Tissue outside the brain, as well as V-CSF, was masked from each
image before SPM analysis. Images were smoothed using a 9 mm3filter,
chosen on the basis of our previous validation experiments (Davatzikos
et al., 2001) and the spatial specificity of our normalization approach.
Because we were interested in identifying regions of absolute volumetric
changes, we did not adjust for global changes in brain volume. Longitu-
dinal change was calculated as baseline minus year 5 images, with signif-
icance set at p ? 0.001, uncorrected. Anatomic localization was deter-
mined from our average gray matter map for the 92 subjects, using
standard anatomical references (Mai et al., 1997; Duvernoy, 1999).
All measurements were highly stable over both 2 and 4 year in-
tervals, with test–retest correlations ranging from 0.94 to 0.99.
changing absolute volumes. These scatterplots show declines in
brain volumes and increases in V-CSF for the majority of indi-
viduals. Although the magnitude of decline in brain volume ap-
pears relatively consistent across subjects, there is a tendency for
greater V-CSF increases in individuals with larger baseline
Results of the mixed-effects regression analyses are summarized
in Tables 2 and 3 for the whole sample and very healthy sub-
sample, respectively. Cross-sectional effects of age and sex on
brain and ventricular volumes were consistent with our previous
report of baseline and 1 year follow-up assessments (Resnick et
al., 2000). As indicated by the significant effects of time, longitu-
dinal brain changes reached statistical significance for all regions
examined. These findings indicated longitudinal decreases in
brain volume measurements and increases in VBR. For ventric-
ular volume and VBR, there was a significant age by time inter-
action, revealing a faster rate of change in both measures with
longitudinal increases were evident when the interaction was
volumes did not differ significantly for men and women. Longi-
tudinal changes remained significant excluding the four individ-
uals with mild cognitive impairment and for the analyses based
on 24 individuals free of medical problems and even very mild
cognitive change. Interestingly, cross-sectional effects of age did
not reach significance in the very healthy group, highlighting the
greater sensitivity of intra-individual measurements of longitu-
Direct comparison of longitudinal tissue loss for gray versus
white matter wasperformed
MANOVA. Consistent with the mixed-effects regression results,
the overall loss of brain tissue over time was highly significant:
F(2,88)? 146.0; p ? 0.0001. However, there were no significant
differences between gray and white matter in the magnitude of
tissue loss. In contrast to these findings, analysis of the magni-
tudes of longitudinal tissue loss for frontal, parietal, temporal,
and occipital regions revealed significant differences among re-
gions (time ? region: F(6,84)? 4.73; p ? 0.001). To further ex-
amine this interaction, values were standardized within each re-
gion using a z-transformation based on the mean and SD at
baseline for each region separately. Longitudinal tissue loss was
greater for frontal and parietal lobes than temporal and occipital
lobes (Fig. 2). The magnitudes of longitudinal changes in the
lobar regions were not significantly influenced by age and sex.
MANOVA restricted to the very healthy subsample showed sig-
nificant longitudinal changes that did not differ as a function of
tissue type or lobar brain region.
The magnitudes of the annual rates of change for each of the
volumetric measurements are presented in Figure 3 for total
brain, gray, white, and ventricular volumes and in Table 4 for
lobar regions. Remarkably, this community-dwelling sample of
increase of 1.4 cm3in ventricular volume each year. (Note that
the discrepancy between volume loss and ventricular volume in-
crease is attributable to the fact that we have not measured the
4 years. Against a background of highly stable measurement, brain volumes decrease and
Resnicketal.•BrainAginginOlderAdultsJ.Neurosci.,April15,2003 • 23(8):3295–3301 • 3297
four participants with mild cognitive impairment yielded nearly
identical rates of change for the global measures. The magnitude
of brain tissue loss was somewhat reduced in the very healthy
elderly, but these individuals still showed significant volume
change. As described above, the apparent trend toward greater
decline in rates of change for white compared with gray matter
volume was not statistically significant when compared directly
using repeated-measures MANOVA.
Voxel-based paired t tests were performed on the RAVENS gray
and white matter images, separately, to examine localized loss of
tissue over 4 years (Figs. 4, 5). Substantial gray matter tissue loss
(Fig. 6) is observed in the regions of the interhemispheric and
Sylvian fissures, affecting cingulate and insular cortex, respec-
tively. Consistent with our cross-sectional observations (Resnick
temporal cortex, albeit to a lesser extent, are also vulnerable to
a lateralized pattern to the gray matter volume loss, with greater
change for the right compared with left hemisphere in many
regions. The right greater than left asymmetry is most pro-
nounced in inferior frontal and anterior temporal regions, but
shows a reversed pattern in the inferior parietal region. Local
analysis of white matter tissue loss reveals widespread changes
throughout the brain. Although there is some asymmetry in the
white matter tissue loss, this asymmetry appears limited to the
temporal lobe, with greater loss of left than right temporal white
declines in both gray and white matter brain volumes in older
adults ranging in age from 59 to 85 years at baseline. Moreover,
we provide the first quantitative data of the rates and regional
distribution of local gray and white matter tissue loss in older
adult men and women. Although other investigations have indi-
cated brain volume loss in specific brain regions, such as the
hippocampus (Kaye et al., 1997; Jack et al., 1998; Mueller et al.,
1998), or more global volume loss (Chan et al., 2001; Tang et al.,
2001), our findings reveal significant longitudinal declines
Brain Gray White Vent VBR Frontal Parietal Temporal Occipital
Brain GrayWhiteVent VBR FrontalParietal TemporalOccipital
**** ** ****
**** ****** ******* *****
To illustrate the significant region-by-time interaction, mean longitudinal change for each
nificant trend for gray matter tissue loss in the healthy subgroup is significant in the more
Brain Gray White BrainGrayWhite
3298 • J.Neurosci.,April15,2003 • 23(8):3295–3301Resnicketal.•BrainAginginOlderAdults
throughout the brain involving multiple selected brain regions
studies, primarily using older imaging techniques, have yielded
inconsistent results regarding differential age effects on gray ver-
sus white matter (Pfefferbaum et al., 1994; Raz et al., 1997; Gutt-
mann et al., 1998). In a voxel-based analysis of high-resolution
MR images from 465 individuals ranging in age from 17 to 79
years, Good and colleagues (2001) reported cross-sectional de-
clines in global gray but not white matter volumes during adult-
hood. However, this study included few individuals over age 65
years. Using high-resolution MR images and a validated ap-
proach to image processing, we previously reported cross-
sectional age effects for both gray and white matter volumes in
our present longitudinal results from serial MRI evaluations
for greater longitudinal tissue loss in white compared with gray
matter, this difference was not statistically significant.
Rates of tissue loss were similar in men and women and in
older and younger adults. In contrast, the rates of increase in
ventricular volume, reflecting central brain atrophy, were signif-
icantly greater in older compared with younger individuals. Al-
having larger ventricles than older women, trends toward sex
differences in rates of V-CSF enlargement did not reach signifi-
cance when adjusted for baseline ventricular volume. Although
in ventricular size (Kaye et al., 1992), the magnitude and age at
which sex differences appear remain unclear. We have begun
enrolling younger BLSA participants into our neuroimaging
study to investigate the age at which rates of brain atrophy accel-
erate in both men and women.
Although tissue loss was distributed across gray and white
local changes revealed regional patterns of vulnerability to age-
related tissue loss. Consistent with other studies and our own
and occipital lobes. Voxel-based analyses of local regions pro-
vided complementary information and indicated that cingulate,
mesial temporal regions showed longitudinal changes in gray
matter. Relative vulnerability of insular and cingulate regions to
age-related gray matter loss is consistent with the cross-sectional
voxel-based analysis of Good and colleagues (2001), but our ob-
structures is contrary to their report of relative preservation of
samples may account in part for these discrepancies. Region-
based analyses of hippocampal volumes have also yielded differ-
ing results across studies, although recent longitudinal investiga-
(Jack et al., 1998; Mueller et al., 1998).
The local pattern of gray matter tissue loss observed in our
study of older adults is intriguing in light of the staging by Braak
and Braak (1997) of the deposition of amyloid in a nonselected
series of autopsy brains. Initial pathology appears in the basal
neocortex, including perirhinal and orbital cortex, and next
spreads into adjacent neocortical association areas and the hip-
pocampal formation, followed finally by its appearance in pri-
segmentation and stereotaxic normalization. Brighter regions of the gray matter image are
Local changes in white matter volumes. Longitudinal declines in white matter
Resnicketal.•BrainAginginOlderAdults J.Neurosci.,April15,2003 • 23(8):3295–3301 • 3299
function of age, the neurotoxic consequences of amyloid may
contribute to our observations of a regional pattern of gray mat-
ter tissue loss. Longitudinal declines in local white matter vol-
umes were more widespread throughout the brain, most likely
al., 1988; Svennerholm et al., 1997). There was a notable hemi-
spheric asymmetry in white matter volume loss in the temporal
lobe, with greater left compared with right tissue loss.
One important limitation of our analysis of local gray and
white matter changes is that there is greater sensitivity for detec-
formation. Registration errors limit our ability to detect change
in voxel-based analyses, and these difficulties will be most pro-
the development of new algorithms for elastic deformation (Da-
vatzikos et al., 2001; Shen and Davatzikos, 2001) will result in
substantial improvement in registration accuracy and will en-
hance identification of change in regions of greater anatomic
variability. The next phase of our analysis will use these new
methods for detection of more highly localized age effects.
Nevertheless, the current approach identified substantial
magnitudes of age-related tissue loss. Annual rates of cerebral
tissue loss were 5.4 cm3(0.5% per year) and of V-CSF increase
of 21.6 and 5.6 cm3, respectively. This volume loss cannot be
explained by disproportionate changes in a limited number of
individuals. Inspection of Figure 1, top and bottom, indicates
that, against a background of highly stable measurement, longi-
tudinal decreases in brain tissue volume and increases in V-CSF
occur across almost all individuals in this age range. Although
some argue that any tissue loss reflects pathological changes as-
sociated with preclinical dementia rather than normal aging, the
uniformity of our findings across individuals argues against this
interpretation unless all are in a preclinical stage of dementia.
Thus, our data provide normative values against which potential
pathological increases in rates of change can be evaluated. How-
ever, it should be cautioned that some of these individuals will
ultimately develop dementia. We would hypothesize that those
frontal regions are more vulnerable to disease, given our obser-
with Alzheimer’s disease (Braak and Braak, 1997). Thus, height-
ened vulnerability to disease may be indicated by accelerated
change in specific regions against a background of age-related
Although our findings indicate that most older adults show
some tissue loss over time, there is substantial variability in the
reduced in individuals who remain medically and cognitively
healthy. Our sample is composed of relatively healthy individu-
als, and they have shown, on average, little cognitive change over
this initial 4 year interval. However, with continued longitudinal
follow-ups we expect to detect cognitive change and impairment
determine the clinical relevance of the observed brain changes
and other behavioral changes. There is already substantial evi-
dence that loss of mesial temporal lobe tissue, particularly in
hippocampus and entorhinal cortex, is associated with memory
brain regions, e.g., orbital-frontal and insular cortex, also show
significant longitudinal tissue loss and may predict other cogni-
tive and functional impairments in elderly individuals. Finally,
our findings from in vivo MRI studies can help direct more fo-
cused neuropathological studies of specific brain regions. Such
correlative neuropathological studies will be critical in elucidat-
ing the cellular basis of these in vivo imaging findings and will
help clarify the apparent paradox between volume loss observed
on MRI and reports of minimal neuron loss with aging.
Andreasen NC, Rajarethinam R, Cizadlo T, Arndt S, Swayze VWI, Flashman
LA, O’Leary DS, Ehrhardt JC, Yuh WTC (1996) Automatic atlas-based
volume estimation of human brain regions from MR images. J Comput
Assist Tomogr 20:98–106.
BraakH,BraakE (1997) FrequencyofstagesofAlzheimer-relatedlesionsin
different age categories. Neurobiol Aging 18:351–357.
ChanD,FoxNC,JenkinsR,ScahillRI,CrumWR,RossorMN (2001) Rates
tia. Neurology 57:1756–1763.
Figiel GS, Webb MC, Spritzer CE, Djang WT (1992) Quantitative cere-
bral anatomy of the aging human brain: a cross-sectional study using
magnetic resonance imaging. Neurology 42:527–536.
Convit A, De Leon MJ, Tarshish C, De Santi S, Tsui W, Rusinek H, George A
matter regions. Three-dimensional views of the t statistic are shown projected on the outer
cortical surface of a representative subject in our sample. The projection on the surface was
hemispheres highlight tissue loss in inferior frontal, insular, and posterior temporal regions
insula (right ? left; inferior and coronal views), orbital frontal cortex (inferior and sagittal
Three-dimensional views of significant longitudinal tissue loss in specific gray
3300 • J.Neurosci.,April15,2003 • 23(8):3295–3301Resnicketal.•BrainAginginOlderAdults
(1997) Specifichippocampalvolumereductionsinindividualsatriskfor Download full-text
Alzheimer’s disease. Neurobiol Aging 18:131–138.
Courchesne E, Chisum HJ, Townsend J, Cowles A, Covington J, Egaas B,
Harwood M, Hinds S, Press GA (2000) Normal brain development and
aging: quantitative analysis at in vivo MR imaging in healthy volunteers.
Davatzikos C (1996) Spatial normalization of 3D brain images using de-
formable models. J Comput Assist Tomogr 20:656–665.
Davatzikos C, Genc A, Xu D, Resnick SM (2001) Voxel-based morphome-
try using the RAVENS maps: methods and validation using simulated
longitudinal atrophy. NeuroImage 14:1361–1369.
Duvernoy H (1999) The human brain: surface, blood supply and three-
dimensional sectional anatomy, Ed 2. New York: Springer.
Folstein MF, Folstein SE, McHugh PR (1975) Mini-mental state: a practical
method for grading the cognitive state of patients for the clinician. J Psy-
chiat Res 12:189–198.
Friston KJ, Holmes AP, Worsley K, Poline JB, Frith CD, Frackowiak RS
approach. Hum Brain Mapp 2:189–210.
Fuld PA (1978) Psychological testing in the differential diagnosis of the de-
mentias. In: Alzheimer’s disease: senile dementia and related disorders
(Katzman R, Terry RD, Bick KL, eds), pp 185–193. New York: Raven.
Goldszal AF, Davatzikos C, Pham DL, Yan MXH, Bryan RN, Resnick SM
metric analysis of brain images. J Comput Assist Tomogr 22:827–837.
human brains. NeuroImage 14:21–36.
Grant R, Condon B, Lawrence A, Hadley DM, Patterson J, Bone I, Teasdale
GM (1987) HumancranialCSFvolumesmeasuredbyMRI:sexandage
influences. Magn Reson Imaging 5:465–468.
Gur RC, Mozley PD, Resnick SM, Gottlieb G, Kohn M, Zimmerman RA,
Herman G, Atlas S, Grossman R, Berretta DA, Erwin R, Gur RE (1991)
Gender differences in age effect on brain atrophy measured by magnetic
resonance imaging. Proc Natl Acad Sci USA 88:2845–2849.
MS (1998) White matter changes with normal aging. Neurology
KokmenE (1998) Rateofmedialtemporallobeatrophyintypicalaging
and Alzheimer’s disease. Neurology 51:993–999.
Kawas C, Gray S, Brookmeyer R, Fozard J, Zonderman A (2000) Age-
specific incidence rates of Alzheimer’s disease: the Baltimore Longitudi-
nal Study of Aging. Neurology 54:2072–2077.
Kaye JA, DeCarli C, Luxenberg JS, Rapoport SI (1992) The significance of
age-related enlargement of the cerebral ventricles in healthy men and
women measured by quantitative computed x-ray tomography. J Am
Geriatr Soc 40:225–231.
Kaye JA, Swihart T, Howieson D, Dame A, Moore MM, Karnos T, Camicioli
R, Ball M, Oken B, Sexton G (1997) Volume loss of the hippocampus
and temporal lobe in healthy elderly persons destined to develop demen-
tia. Neurology 48:1297–1304.
Mai JK, Assheuer J, Paxinos G (1997) Atlas of the human brain. San Diego:
Miller AKH, Corsellis JAN (1977) Evidence for a secular increase in human
brain weight during the past century. Ann Hum Biol 4:253–257.
Morrell CH, Pearson JD, Brant LJ (1997) Linear transformations of linear
mixed-effects models. Am Stat 51:338–343.
Morrison JH, Hof PR (1997) Life and death of neurons in the aging brain.
Quinn JF, Kaye JA (1998) Brain volume preserved in healthy elderly
through the eleventh decade. Neurology 51:1555–1562.
Murphy DGM, DeCarli C, McIntosh AR, Daly E, Mentis MJ, Pietrini P,
Szczepanik J, Schapiro MB, Grady CL, Horwitz B, Rapoport SI (1996)
quantitative magnetic resonance imaging and positron emission tomog-
raphy study on the effect of aging. Arch Gen Psychiatry 53:585–594.
Pakkenberg B, Gundersen HJ (1997) Neocortical neuron number in hu-
mans: effect of sex and age. J Comp Neurol 384:312–320.
TangalosEG,BoeveBF,KokmenE (2000) MemoryandMRI-basedhip-
pocampal volumes in aging and AD. Neurology 54:581–587.
(1994) A quantitative magnetic resonance imaging study of changes in
brain morphology from infancy to late adulthood. Arch Neurol
Raz N, Gunning FM, Head D, Dupuis JH, McQuain J, Briggs SD, Loken WJ,
Thornton AE, Acker JD (1997) Selective aging of the human cerebral
cortex observed in vivo: differential vulnerability of the prefrontal gray
matter. Cereb Cortex 7:268–282.
RN,ZondermanAB (2000) One-yearagechangesinMRIbrainvolumes
in older adults. Cereb Cortex 10:464–472.
Shen D, Davatzikos C (2001) HAMMER: hierarchical attribute matching
on Mathematical Methods in Biomedical Image Analysis, pp 29–36.
Kauai, Hawaii, December.
Spitzer R, Williams J (1987) Diagnostic and statistical manual of mental
disorders (Ed 3, revised) DSM-III-R. Washington, D.C.: American Psy-
SvennerholmL,BostromK,JungbjerB (1997) Changesinweightandcom-
positions of major membrane components of human brain during the
Talairach J, Tournoux P (1988) Co-planar stereotaxic atlas of the human
brain. New York: Thieme Medical.
Tang Y, Whitman GT, Lopez I, Baloh RW (2001) Brain volume changes on
longitudinal magnetic resonance imaging in normal older people. J Neu-
Wiggins RC, Gorman A, Rolsten C, Samorajski T, Ballinger Jr WE, Freund G
(1988) Effects of aging and alcohol on the biochemical composition of
histologically normal human brain. Metab Brain Dis 3:67–80.
Yan MXH, Karp JN (1995) An adaptive Bayesian approach to three-
dimensional MR brain segmentation. In: Proceedings of XIVth Interna-
tional Conference on Information Processing in Medical Imaging, pp
201–213. Ile de Berder, France, June.
Yue NC, Arnold AM, Longstreth WT, Elster AD, Jungreis CA, O’Leary DH,
Poirier VC, Bryan RN (1997) Sulcal, ventricular, and white matter
changes at MR imaging in the aging brain: data from the Cardiovascular
Health Study. Radiology 202:33–39.
Resnicketal.•BrainAginginOlderAdultsJ.Neurosci.,April15,2003 • 23(8):3295–3301 • 3301