Short Communications

Prefrontal Compensatory Mechanism May Enable

Normal Semantic Memory Performance in Mild Cognitive
Impairment (MCI)
Ariela Gigi, PhD, Reuven Babai, PhD, Avichen Penker, BA, Talma Hendler, PhD, MD, Amos D. Korczyn, MA, MD
From the Department of Behavioral Sciences, Ariel University Center, Ariel 44837, Israel (AG, AP); Department of Science Education, The Constantiner School of Education,
Tel Aviv University, Tel Aviv 69978, Israel (RB); The Functional Brain Imaging Unit, Whol Institute for Advanced Imaging, Sourasky Medical Center, Tel Aviv 64239, Israel (TH);
Department of Psychiatry, Sackler School of Medicine, Tel-Aviv University, Tel Aviv 69978, Israel (TH); Department of Psychology, Tel Aviv University, Tel Aviv 69978, Israel
(TH); Sieratzki Chair of Neurology, Tel Aviv University, Tel Aviv 69978, Israel (ADK).

ABSTRACT
Mild cognitive impairment (MCI) is considered as a potential transitional state between Keywords: Mild cognitive impairment
normal aging and dementia. Studies addressing semantic memory in patients with incipi- (MCI), dementia, Alzheimer’s disease,
ent dementia and MCI show inconsistent results. In the current report we focused on MCI prefrontal cortex, fMRI, compensation,
and examined memory performance (semantic, episodic, and working memory) in addi- neuropsychology, Naming test.
tion to structural and functional magnetic resonance imaging (fMRI) measurements. We Acceptance: Received December 26,
studied 6 MCI, 6 normal controls, and 4 Alzheimer disease (AD) patients. MCI participants 2007, and in revised form November 8,
demonstrated normal semantic memory performance while fMRI examination revealed 2008. Accepted for publication November
distinct patterns of activations between MCI and normal aged subjects. According to 12, 2008.
our previous study, time courses were taken from parietal, dorsolateral-prefrontal-cortex Correspondence: Address correspon-
(DLPFC), hippocampal formation, and fusiform gyri. A small number of activated voxels dence to Ariela Gigi, PhD, Department
in parietal regions were depicted in MCI participants and were correlated with structural of Behavioral Sciences, Ariel Univer-
changes in this region. In contrast, significantly higher activation (intensity and number sity Center, Ariel 44837, Israel. E-mail:
of voxels) was observed in DLPFC of MCI subjects. The overactivity seen in DLPFC of MCI Arielag@ariel.ac.il.
may represent a compensatory mechanism that enables them to perform normally. These J Neuroimaging 2010;20:163-168.
preliminary results suggest that functional imaging may be useful for early diagnosis of DOI: 10.1111/j.1552-6569.2009.00386.x
dementia and call to develop reliable tests and criteria that will enable using such methods
clinically.

Introduction
Mild cognitive impairment (MCI) is documented to progress
to Alzheimer disease (AD) at rates of 10-15% per year com-
pared with a much lower rate of 1-2% per year in the general
elderly population, suggesting that many MCI subjects have in-
cipient dementia, particularly AD. The majority of longitudinal
and cross-sectional behavioral studies report deficits in verbal
episodic learning and retrieval abilities in both AD and MCI.
Regions within the medial temporal lobe have been identified
as important for memory formation. The medial temporal lobe
is damaged in AD and was found to harbor the first changes
in MCI.1 Studies that examined memory aspects in MCI sub-
jects or healthy people at genetic risk for AD showed incon-
sistent results. Reduced activation during naming and verbal
fluency tasks as well as reduced medial temporal metabolism
were found in several studies for participants at risk for AD.2
Other studies show that during memory performance, partici-
pants at risk for AD have increased signal intensity and more
activated voxels in the hippocampal, parietal, and prefrontal
regions.3,4 Yet, some studies failed to find differences between
MCI and controls in regional cerebral blood flow ratio in the
frontal temporal or parietal cortices.5
Naming difficulties often occur in pathological aging, as re-
ported for patients with dementia of the Alzheimer type and for
subjects with MCI.6 Naming performance that involves percep-
tion, semantic activities, and language is often studied to assess
semantic memory. Studies suggest that naming is impaired in
AD.7 In addition, data on semantic memory in MCI show fre-
quent category fluency and naming difficulties.8,9
The current study focuses on semantic memory in MCI. We
used behavioral tests to examine memory performance (seman-
tic, episodic, and working memory). In addition, we performed
structural measurements and applied functional magnetic res-
onance imaging (fMRI) to study cerebral activation patterns
during naming performance. Based on the literature described
above, our regions of interest were the parietal, dorsolateral pre-
frontal cortex (DLPFC), hippocampal formation, and fusiform
complex. We focused on comparing the behavioral, structural,
and functional results of the MCI subjects to those of healthy
age-matched controls. In addition, we relate in the current

Copyright ◦ 2009 by the American Society of Neuroimaging

C 163
Table 1. Demographic and MMSE Information (Mean ± SEM)
Controls N = 6
Age (years) 62.0 ± 5.2
Education (years) 12.8 ± 2.0
MMSE 29.3 ± .4
∗
Significance was measured by one-way analysis of variance.
For MMSE performance Scheffe Post Hoc showed that the only significant difference was between the AD group and the other
two (P < .05).
study to the results of 4 AD patients who underwent the same
examination.
Methods
Subjects
Six MCI subjects (all females), 4 AD patients (1 female), and 6
healthy age-matched controls (4 females) participated in the
study. Table 1 summarizes their demographic Mini-Mental
State Examination (MMSE) data. An expert neurologist diag-
nosed the MCI and AD participants, according to accepted cri-
teria.10,11 Inclusion criteria for amnesic MCI were12 subjective
memory complains, objective memory dysfunction, absence of
dementia, MMSE score of 24 or above, and absence of any
neurological or psychiatric illnesses.13 T2 and fluid-attenuated
inversion recovery (FLAIR) MRI scans were performed to ex-
clude other localized brain pathologies. The Tel Aviv Sourasky
Medical Center Ethics Committee approved the experimental
procedure, and a written informed consent was obtained from
each participant.
General Procedure
Each participant underwent the behavioral, anatomical, and
fMRI assessments. The behavioral assessment was done in one
session, while fMRI and anatomical examinations were done in
a separate session, a few days apart. Each session lasted 1.5-2.5
hour.
Behavioral Neuropsychological Test Battery
Episodic memory evaluation included verbal learning and re-
call abilities measured by the Hebrew version of the “Rey Audi-
tory Verbal Learning Test”14 (AVLT) and the “Logical Memory
Test” (by stories learning) from the Wechsler Memory Scale.15
Semantic memory abilities related to general knowledge were
ascertained by using the “General Information Test” from the
Wechsler Adult Intelligence Scale, revised.16 Word fluency was
examined as described by Lezak.17 The behavioral “Object
Naming Task” was described before.18 In brief, during this test
40 picture cards (in black and white) of common objects are
presented to the participant. Half of them are depicted from a
usual viewpoint and the other half from an unusual one. The
participants are asked to name the object depicted in each pho-
tograph. Working memory span was tested by the “Digit Span
Test” (from WAIS-R) and the “Sentence Repetition Test.”18
164 Journal of Neuroimaging Vol 20 No 2 April 2010
MCI N = 6 AD N = 4 Significance∗
64.0 ± 8.4 61.7 ± 6.1 NS
12.5 ± .5 13.0 ± 1.2 NS
27.7 ± .8 20.0 ± .4 .001
MRI Data Acquisition
The MRI measurements were performed in a whole-body 1.5
Tesla scanner (Signa Horizon, Echo speed, LX MRI scanner,
GE Healthcare, Milwaukee, WI) located at the Wohl Insti-
tute for advanced imaging in the Tel Aviv Sourasky Medical
Center. The MRI session included structural and functional
protocols. Structural scans included T1, fast spin echo-T2
weighted images, and FLAIR images. The fMRI protocols
were based on a multislice gradient echo, echo-planar imag-
ing (EPI) using a standard head coil. The functional data were
obtained using the following parameters: T R = 3 s, T E =
55 ms, flip angle = 90◦ , imaging matrix = 80 × 80, field of
view = 24 cm. Seventeen slices, 5-mm thick with 1-mm gaps,
were selected, based on a mid-sagittal slice, oriented approxi-
mately in the axial position, covering all cortical areas. In ad-
dition, 3-dimensional anatomical volumes were collected using
a T1 SPGR sequence with high resolution, for each subject, for
volume analyses.
Structural Semi-Quantitative and Linear Measurements
Semi-quantitative measurements were used for assessing the
subarachnoid space and the lateral ventricles size. The assess-
ment was performed by grading the size on a scale of 0—3: 0 =
no enlargement, 1 = mild, 2 = moderate, 3 = severe atrophy.1
Linear measurements included assessment of the third ventricle
width (V3; by using an axial T1 slice at the level of the foramen
of Monro).
Visual Stimulation during Functional-MRI
The fMRI “Object Naming Task” follows the scheme described
above in the behavioral neuropsychological tests but included
different common pictured objects that were computerized pre-
sented by Realslid software (GE Healthcare). Behavioral per-
formance was not measured during the fMRI session.
Stimuli
The stimuli consisted of 40 black and white pictures of common
objects. Twenty objects were pictured from a usual viewpoint,
and 20 different objects were pictured from an unusual point of
view.
General Procedures
The test session included two naming conditions, “usual” and
“unusual,” presented in a block design fashion. Four different
blocks were constructed, and each condition-block was pre-
sented in alternating order during the session. Ten epochs were
Table 2. Behavioral Neuropsychological Memory Assessment Results (Mean ± SEM)

Test Controls MCI AD

Episodic AVLT maximum 12.2 ± .6 9.8 ± 1.2∗ 5.0 ± .0

Memory AVLT delayed recall 8.9 ± .9 5.7 ± 1.9 0
AVLT recognition (A & B lists) 18.7 ± 1.1 13.2 ± 3.1∗ 0
Percentile logical memory—immediate 64.8 ± 8.8 31.2 ± 9.4∗∗ 5.5 ± 4.5
Percentile logical memory—delayed 67.7 ± 6.8 27.2 ± 9.3∗∗∗ .5 ± .4
Semantic General information (WAIS) 12.1 ± .6 11.8 ± 1.0 4.3 ± 2.3
Memory Word fluency (animals category) 21.3 ± 2.2 18.7 ± 4.0 7.5 ± 3.5
Word fluency (phonological) 35.9 ± 4.4 28.7 ± 10.3 21.5 ± 6.5
Object naming—usual viewpoint 18.7 ± .4 19.6 ± .4 16.0 ± 2.5
Object naming—unusual viewpoint 13.1 ± 1.1 12.6 ± 1.4 9.7 ± 3.0
Working Digit span forward 8.0 ± .8 7.4 ± .7 3.7 ± 2.0
memory Digit span backward 6.2 ± .7 4.8 ± .5 3.3 ± 1.2
Sentences repetition 31.5 ± 1.2 27.0 ± .9∗ 18.7 ± 5.0

Significant differences between MCI and controls by independent sample t-test: ∗ P < .05; ∗∗ P < .03; ∗∗∗ P < .003. The AD patients
differed significantly from MCI and controls in all tests performance (P < .01).

presented in one block, each of them included an experimental
stimulus followed by a picture of a centered “fixation point”
on a gray background, presented for 550 and 1,550 ms, re-
spectively. Participants were instructed to fixate on the center
of the image and to covertly name the presented object as ac-
curately and as fast as possible. The blocks were separated by
6-15 second intervals, in which participants viewed a gray back-
ground with a centered fixation point (ie, blank condition).
The fMRI session included the addition of 2-3 blocks of
scrambled pictures (“scrambled” condition), for mapping low
visual areas. The “scrambled” stimuli were generated by split-
ting pictures into 1,024 rectangles, which were then randomly
gathered.19
Prior to the test runs, subjects practiced the task with different
pictures.
Functional-MRI Data Analysis
els in each ROI. The average signal intensity across all subjects
was also calculated. Across-subjects analysis was done using
the general linear model approach (GLM),23 in which all stim-
uli conditions are positive predictors, with a lag of 3-6 seconds
(to account for the hemodynamic response delay). To create
the maps, the time-courses of all subjects were transformed into
Talairach space.24 The regression weights for each condition
were estimated using the GLM.
Results
General
Our three tested groups were quite similar in age and education
with no significant difference between the groups (Table 1). The
MMSE differed significantly (F (2,15) = 59.3; P = .001), with
the AD group being significantly different from the other two
(Scheffe Post Hoc: P < .05).

fMRI data were processed using BrainVoyager 4.4 software
package (www.brainvoyager.com, Brain Innovation, Maas-
tricht, The Netherlands). For each subject, the 2-dimensional
functional data were aligned to 2-dimensional anatomical slices
of the same subject. Before statistical analysis, raw data were
examined for motion and signal artifacts. Head motion correc-
tion and high-pass temporal smoothing in the frequency do-
main were applied in order to remove drifts and to improve
the signal-to-noise ratio. Only voxels with a correlation coeffi-
cient above .4 with cognitive condition (P < .01, not corrected
for multiple comparisons) were included in the regions of in-
terest (ROIs). The obtained maps were superimposed on to
3-dimensional anatomical reference scans. The Talairach co-
ordinates were determined for each ROI. Several functional
imaging studies showed that the lateral prefrontal cortex and the
medial temporal cortex play a critical role in semantic memory
and in successful retrieval of target information.20 Accordingly,
time-courses were taken from Brodman areas (BA) 7 and 40, the
DLPFC that included the BA 9, 46, and 47; the hippocampal
formation; and the fusiform gyri.19,21,22 The subject’s average
signal intensity was estimated by averaging across all the vox-
Neuropsychological Data
Table 2 summarizes the results of the neuropsychological mem-
ory assessments. The table also depicts the statistically signifi-
cant differences that were found between the MCI and control
groups. As can be seen in the table, the episodic memory per-
formance was significantly lower in MCI compared with the
controls, but no significant differences were found for the se-
mantic memory tests performance. Lower performance in MCI
was also observed in the working memory tests, although only
one of them (sentences repetition) was statistically significant.
The AD group showed poor results for all tests, compared with
controls or MCI.
Structural Data
All of the MCI subjects showed an enlargement of the pari-
etal subarachnoid spaces and scored 1-2 in semi-quantitative
structural assessments. The linear measurements showed that
the V3 of the MCI group was wider (2-8 mm) than normal val-
ues (1.7-6.6 mm).25 In addition, enlargement of the temporal
horns (both sides) was found in 3 of the MCI subjects, and this

Gigi et al: Prefrontal Compensatory Mechanism and MCI 165

Table 3. Functional MRI Results during Naming Performance (Usual and Unusual) in Control and MCI Participants (Mean ±
SEM)

Number of Voxels Percent Signal Change

Controls MCI Ratio Sig.∗ Controls MCI Ratio Sig.∗

R-fusiform 5,289 ± 158 2,980 ± 79 .6 .0001 1.7 ± .2 1.2 ± .1 .7 .009

L-fusiform 6,441 ± 226 2,769 ± 93 .4 .0001 1.1 ± .1 .9 ± .1 .8 NS
R-hipp. 395 ± 14 447 ± 23 1.1 .04 2.1 ± .2 1.5 ± .1 .7 .01
L-hipp. 409 ± 13 415 ± 8 1.0 NS 3.0 ± .3 2.2 ± .5 .7 NS
R-parietal 4,628 ± 205 585 ± 29 .1 .0001 .5 ± .03 1.0 ± .1 2.0 .0001
L-parietal 2,519 ± 151 532 ± 12 .2 .0001 .5 ± .03 .9 ± .1 1.8 .0001
R-DLPFC 1,476 ± 65 1,315 ± 45 .9 NS .5 ± .04 2.1 ± .3 4.2 .0001
L-DLPFC 672 ± 29 754 ± 26 1.1 .03 .5 ± .06 3.4 ± .4 6.8 .0001

Number of voxels and percent signal change detected by fMRI during naming performance. The table describes the combined
results of the usual and unusual conditions.
Ratio refers to the proportion of MCI to controls (MCI/control) activation.
DLPFC = dorsolateral prefrontal cortex; hipp = hippocampal formation.
∗Significance relates to the comparison of control and MCI groups’ performance by independent sample t-test and was calculated
independently for the number of activated voxels or percent signal change.

was significantly correlated with low performance in the AVLT
episodic memory task (r > .97; P < .01).
Functional MRI Brain Activity
MCI, AD, and control participants performed the semantic
memory task (“usual and unusual naming test”) during the fMRI
session. During the analysis of the MRI data we measured in
each ROI the number of activated voxels as well as the mean
percent signal change for both conditions together (“usual” and
“unusual”). These results are presented in Table 3. MCI partici-
pants showed higher activation in several ROIs compared with
the controls, while in other locations lower or similar activations
were observed. Higher activation in MCI, in terms of percent
signal change, were found in the parietal and DLPFC in both
hemispheres (Fig 1). In the latter the degree of activation was
more than 4 times higher compared with the controls. Higher
activation in terms of number of activated voxels was observed
in the left DLPFC and the right hippocampal formation of the
MCI subjects. Lower activation, a fifth or less compared with
controls, was observed in terms of number of voxels in the
parietal lobe.
Overall, brain activation (percent signal change and num-
ber of activated voxels) in AD patients was poor in all ROIs
compared with the MCI or control groups (data not shown).
AD patients showed little percent signal change (less then .8%)
in all tested regions.
Discussion
Research on AD and MCI has mostly focused on episodic and
working memory, studied behaviorally and by imaging tech-
niques. In the current study we used neuropsychological assess-
ments, structural measurements, and fMRI to study semantic
memory in MCI.

Fig 1. Differences in brain activation patterns between MCI subjects (blue to green) and controls (yellow to red) during naming task
performance (see methods). The scale on the right represents levels of activation. Compensatory brain activation can be seen in the dorsolateral
prefrontal cortex of MCI participants in comparison with controls. Mapping was constructed from two controls and two MCI subjects, for
illustration.

166 Journal of Neuroimaging Vol 20 No 2 April 2010

 In the structural measures we found a wider V3 in the MCI
group compared with normal norms.25 This result is in agree-
ment with a previous study that showed correlation between
V3 width and severity of the disease in AD patients.26
The neuropsychological results showed that the MCI sub-
jects had low scores in episodic memory performance and also
a deficit in working memory, as was shown before.27 In the cur-
rent study a significant correlation was found between the per-
formance in the AVLT episodic memory test and the enlarge-
ment of the temporal horns. An earlier study has found that the
width of the temporal horns is a sensitive marker discriminating
between controls and mild or moderate AD cases.28 Thus, our
results support these findings connecting between performance
and structural changes related to the temporal lobe.
Although our MCI participants complained of daily diffi-
culties in words and in finding object names, their semantic
memory performance on the behavioral tests were only slightly
lower (non-significantly) than the performance observed for the
controls (Table 2). One of the semantic tests was a naming task
of pictured objects. Although in this naming task the behav-
ioral performance of MCI participants matched that of normal
controls, fMRI examination of the cerebral activity during such
task revealed that the performance of the MCI participants was
achieved by a distinct pattern of activation, different from the
one found in the controls.
MCI participants in comparison to controls showed en-
hanced activation of parietal and frontal regions during object
naming. Enhanced activation in parietal regions (BA 7 and 40)
of MCI participants is in agreement with recent literature that
shows that the parietal lobe is a vulnerable region in MCI and
non-demented subjects with memory loss.29 For example, it was
shown by positron emission tomography that MCI participants
who later converted to AD had reduced glucose metabolism
compared with non-converters, only in the parietal region.30
Moreover, metabolic impairment in this region was found to
be strongly related with the degree of cognitive impairment in
very mild AD.31 Structural damage in this region was revealed
by examining T1-MRI scans of our MCI participants. Using
semi-linear measurements we found enlargement of the pari-
etal subarachnoid space.32 In addition, the number of activated
voxels in the current fMRI study is in agreement with the above
data showing structural damage to the parietal region. Table 3
shows that during the naming task, the number of activated
voxels in the parietal region of MCI subjects was only 13% or
21% (in the right or left, respectively) of the number of acti-
vated voxels in the controls. The higher deficiency in the right
compared with the left parietal was also previously reported
using other methods of inspection.30,33 The data show that the
fusiform region of MCI subjects also has fewer activated voxels
and less intensity during the naming performance compared
with the controls. This is in accordance with recent findings in
MCI that showed a medial temporal volume loss that included
this specific region.33 Since the parietal region was revealed to
be important for semantic processes and both regions (parietal
and fusiform) were seen to be important for the accumulation
of visual semantic information and were shown to be highly
activated during naming task performance in controls,19 the
Gigi et al: Prefrontal Compensatory Mechanism and MCI
question arises as to how our MCI participants were able to
perform behaviorally equally well as the controls?
The DLPFC [BA 9, 46, and 47] was also one of our ROIs. BA
47 is usually associated with semantic search.34 One possibility
to overcome the structural loss might be to hyperactivate this
region, and indeed, this was detected. However, the increased
activation is restricted to the small number of active voxels in
the parietal region (13% or 21% of the control). Overcoming this
deficiency might also be obtained by recruiting additional cere-
bral regions that will compensate for the lack of activity in the
parietal regions. It can be seen that a significantly higher activa-
tion was observed bilaterally in the DLPFC of the MCI group
compared with the control group. In addition, the number of
active voxels in the DLPFC of the MCI participants seems to be
comparable or even larger than the number of activated voxels
in the controls. Overactivity in DLPFC was also seen previously
and was suggested to reflect a compensatory mechanism for sev-
eral behavior deficits.35 For example, elevated frontal activation
in mild AD patients during a rehearsal task was interpreted as
reflecting compensatory recruitment of neural resources.36 In
addition, a compensatory role for frontal areas related to verbal
episodic memory was described by Becker for AD patients.37
Accordingly, we suggest that the overactivity seen here in the
DLPFC of our MCI subjects may represent a compensatory
mechanism that, in this case, enables them to perform as well
as the controls during the naming task in spite of the structural
deficits found in their parietal region and fusiform gyri.
In summary, in this preliminary study the memory perfor-
mance of MCI participants was investigated compared with
healthy age-matched controls. It should be noted that we used
a small sample size and that the tested patients might have
been early-onset subjects; therefore, further study is needed in
order to generalize these findings. Yet, our findings are in line
with many studies that showed for MCI subjects low scores
in episodic- and working-memory performances. In line with
earlier studies, we did not find a deficit in semantic memory in
MCI in terms of behavioral performance (for review see Arnaiz
and Almkvist38 ). Functional imaging examination was found to
be a sensitive tool to asses that matter, suggesting the use of
functional imaging methods for early diagnosis of dementia,
which further calls for developing reliable tests and criteria that
will enable using such methods clinically.
The study was supported by the Israeli Ministry of Health (No. 5707–
1). We would like to thank Prof. Irith I. Reider-Groswasser for her
guidance and caring out the structural linear measurements.
References
1. Visser PJ, Verhey FRJ, Scheltens P, et al. Diagnostic accuracy of
the Pre-clinical AD Scale (PAS) in cognitively mildly impaired
subjects. J Neurol 2002;249:312-319.
2. Santi S, de-Leon MJ, Rusinek H, et al. Hippocampal formation
glucose metabolism and volume losses in MCI and AD. Neurobiol
Aging 2001;22:529-539.
3. Bookheimer SY, Strojwas MH, Cohen MS, et al. Patterns of brain
activation in people at risk for Alzheimer’s disease. N Engl J Med
2000;343:450-456.
167
 4. Smith CD, Andersen AH, Kryscio RJ, et al. Women at risk for AD
show increased parietal activation during a fluency task. Neurology
2002;58:1197-1202.
5. Encinas M, De Juan R, Marcos A, et al. Regional cerebral blood
flow assessed with 99mTc-ECD SPET as a marker of progression
of mild cognitive impairment to Alzheimer’s disease. Eur J Nucl
Med Mol Imaging 2003;30:1473-1480.
6. Dudas RB, Clague F, Thompson SA, et al. Episodic and se-
mantic memory in mild cognitive impairment. Neuropsychologia
2005;43:1266-1276.
7. Nagata K, Yoshinari K, Ozawa S, et al. The relationship between
naming and semantic knowledge for different categories in demen-
tia of Alzheimer’s type. Neuropsychologia 1997;35:1251-1260.
8. Vogel A, Gade A, Stokholm J, et al. Semantic memory impairment
in the earliest phases of Alzheimer’s disease. Dement Geriatr Cogn
Disord 2005;19:75-81.
9. Duong A, Whitehead V, Hanratty K, et al. The nature of lexico-
semantic processing deficits in mild cognitive impairment. Neu-
ropsychologia 2006;44:1928-1935.
10. McKhann G, Drachman D, Folstein M, et al. Clinical diagnosis of
Alzheimer’s disease: report of the NINCDS-ADRDA Work Group
under the auspices of Department of Health and Human Services
Task Force on Alzheimer’s disease. Neurology 1984;34:939-944.
11. Petersen RC, Smith GE, Waring SC, et al. Mild cognitive
impairment: clinical characterization and outcome. Arch Neurol
1999;56:303-308.
12. Smith GE, Petersen RC, Parisi JE, et al. Definition, course and
outcome of mild cognitive impairment. Aging Neuropsychol Cogn
1996;3:141-147.
13. Folstein MF, Folstein SE, McHugh PR. Mini-mental state: a prac-
tical method for grading the cognitive state of patients for the clin-
ician. J Psychiatr Res 1975;12:189-198.
14. Vakil E, Agmon-Ashkenazi D. Baseline performance and learning
rate of procedural and declarative memory tasks: younger versus
older adults. J Gerontol B Psychol Sci Soc Sci 1997;52:229-234.
15. Wechsler D. WMS-R Manual . New York: Psychological Corp,
1987.
16. Wechsler D. WAIS-R Manual . New York: Psychological Corp,
1981.
17. Lezak MR. Neuropsychological Assessment. USA: Oxford University
Press, 1995.
18. Gigi A, Vakil E, Kahana E, et al. Presymptomatic signs in healthy
CJD mutation carriers. Dement Geriatr Cogn Disord 2005;19:246-
255.
19. Gigi A, Babai R, Katzav E, et al. Prefrontal and parietal regions
are involved in naming of objects seen from unusual viewpoints.
Behav Neurosci 2007;121:836-844.
20. Backman L, Andersson JLR, Nyberg L, et al. Brain regions as-
sociated with episodic retrieval in normal aging and Alzheimer’s
disease. Neurology 1999;52:1861-1870.
21. Thompson-Schill SL, Aguirre GK, DwEsposito M, et al. Aneural
basis for category and modality specificity of semantic knowledge.
Neuropsycologia 1999;37:671-676.
168 Journal of Neuroimaging Vol 20 No 2 April 2010
22. Thompson-Schill SL. Neuroimaging studies of semantic memory:
inferring “how” from “where.” Neuropsychologia 2003;41:280-292.
23. Friston KJ, Holmes AP, Worsley KJ, et al. Statistical parametric
maps in functional imaging: a general linear approach. Hum Brain
Map 1995;2:189-210.
24. Talairach J, Tournoux P. Co-Planar Stereotaxic Atlas of the Human
Brain. NY: Theime Medical Publishing, 1988.
25. Gyldensted C. Measurements of the normal ventricular system
and hemispheric sulci of 100 adults with computed tomography.
Neuroradiology 1977;14:183-192.
26. Soininen H, Reinikainen KJ, Puranen M, et al. Wide third ventricle
correlates with low choline acetyltransferase activity of the neocor-
tex in Alzheimer patients. Alzheimer Dis Assoc Disord 1993;7:39-
47.
27. Nordahl CW, Ranganath C, Andrew PY, et al. Different mecha-
nisms of episodic memory failure in mild cognitive impairment.
Neuropsychologia 2005;43:1688-1697.
28. Frisoni GB, Beltramello A, Weiss C. Linear measures of atrophy
in mild Alzheimer disease. AJNR Am J Neuroradiol 1996;17:913-
923.
29. Tanaka M, Fukuyama H, Yamauchi H, et al. Regional cerebral
blood flow abnormalities in nondemented patients with memory
impairment. J Neuroimaging 2002;12:112-118.
30. Mosconi L, Perani D, Sorbi S, et al. MCI conversion to demen-
tia and the APOE genotype: a prediction study with FDG-PET.
Neurology 2004 28;63:2332-2340.
31. Desgranges B, Baron JC, Lalevée C, et al. The neural substrates
of episodic memory impairment in Alzheimer’s disease as re-
vealed by FDG–PET: relationship to degree of deterioration. Brain
2002;125:1116-1124.
32. Gigi A, Babai B, Hadar U, et al. Increased cerebral activation in
MCI may reflect compensatory mechanisms. European Journal of
Neurology 2003;10(Suppl 1)234-244.
33. Chételat G, Landeau B, Eustache F, et al. Using voxel-based
morphometry to map the structural changes associated with
rapid conversion in MCI: a longitudinal MRI study. Neuroimage
2005;27:934-946.
34. Bernard FA, Desgranges B, Eustache F, et al. Neural correlates
of age-related verbal episodic memory decline: a PET study
with combined subtraction/correlation analysis. Neurobiol Aging
2007;28:1568-1576.
35. Bondi MW, Houston WS, Eyler LT, et al. fMRI evidence of com-
pensatory mechanisms in older adults at genetic risk for Alzheimer
disease. Neurology 2005;64:501-508.
36. Woodard JL, Grafton ST, Votaw JR, et al. Compensatory recruit-
ment of neural resources during overt rehearsal of word lists in
Alzheimer’s disease. Neuropsychology 1998;12:491-504.
37. Becker JT, Mintun MA, Aleva K, et al. Compensatory reallo-
cation of brain resources supporting verbal episodic memory in
Alzheimer’s disease. Neurology 1996;46:692-700.
38. Arnaiz E, Almkvist O. Neuropsychological features of mild cog-
nitive impairment and preclinical Alzheimer’s disease. Acta Neurol
Scand 2003;107(Suppl 179):34-41.