is currently a fourth-year Radiology Resident at the Mallinckrodt
Institute of Radiology, Washington University School of Medicine,
St. Louis, MO. She earned a bachelor's degree from Williams
College and her MD and PhD degrees from the University of
Chicago-Pritzker School of Medicine, Chicago, IL. After
completing her residency, she will begin a Neuroradiology
Fellowship at the Mallinckrodt Institute of Radiology.
Alzheimer's disease (AD) is an irreversible dementia of the
elderly that has an insidious onset but eventually leads to severe
debilitation and death. It currently affects more than 4 million
people in the United States, and it is estimated that 14 million
Americans will be afflicted by the year 2050.
Given the aging population in the United States, the National
Institutes of Health have designated research in AD to be an urgent
As new therapies for early intervention in AD become available,
such as the vaccine trials currently under way,
the need for accurate, early diagnostic tools and the ability to
assess response to therapy will become imperative. Radiologic
imaging potentially holds the key to providing accurate, early
diagnosis of this devastating disease. An overview of current
routine, specialized, and research protocols for the imaging of AD
is provided in Table 1.
Imaging of dementia: Conventional computed tomography and
magnetic resonance imaging
Current recommendations from the American Academy of Neurology
for the evaluation of patients with suspected AD call for a
complete neurological assessment plus a computed tomographic (CT)
or magnetic resonance imaging (MRI) scan.
The current role of imaging is to exclude other causes of dementia,
such as multiple infarctions (the second most common form of
dementia), brain tumors, head trauma (acute, subacute, or chronic),
normal pressure hydrocephalus, prion diseases, or inflammatory
diseases (such as human immunodeficiency virus [HIV] encephalitis).
The finding of diffuse atrophy, with enlargement of the ventricles
and sulci, is suggestive of AD; however, this pattern of atrophy
can also be seen in normal aging. In contrast, basal ganglia or
thalamic strokes can lead to focal cortical atrophy. Atrophy
limited to the frontal and temporal cortex is often found in
frontotemporal dementia. Other diseases that may be identified at
MRI, based upon the topological distribution of atrophy, include
Pick's atrophy, corticobasal degeneration, Huntington's disease,
and multisystem degeneration.
Periventricular white matter disease is another finding commonly
seen in AD. It is manifested by decreased attenuation on CT and
more readily by increased signal intensity on T2-weighted MRI.
However, periventricular white matter disease may result from
chronic microinfarctions and can be seen in patients with and
Most MRI protocols for initial evaluation of dementia also
include diffusion-weighted imaging (DWI) sequences. There are no
findings specific for AD on conventional DWI. However, DWI is
helpful in the evaluation of stroke as a cause of the patient's
dementia and to establish its chronicity. DWI is also helpful in
exclusion of rare causes of dementia such as Cruetzfeldt-Jakob
disease (CJD), a prion disease. In CJD, patients may present with
increased signal on DWI in the basal ganglia and thalami (Figure
Conventional CT and MRI, even with the inclusion of DWI, currently
offer little information specific to diagnosing AD, nor do they
offer any prognostic information.
Specialized clinical imaging protocols
Nuclear medicine offers 2 examinations, single-photon-emission
CT (SPECT) and positron emission tomography (PET), which may be
useful in the diagnosis of AD. SPECT and PET can be used to
evaluate global and regional metabolic and blood flow
abnormalities. Specifically, patients with AD show symmetric,
bilateral hypometabolism in the temporal and parietal lobes.
In contrast, multi-infarct dementia shows asymmetric, focal
abnormalities, and frontal and temporal lobe dementias show
hypometabolism limited to those lobes.
PET allows higher resolution and greater sensitivity than SPECT,
whereas SPECT is more commonly available. A recently published
meta-analysis of PET in AD found an overall sensitivity of 86% and
a specificity of 86%, but suggested that clear parameters including
patient selection, threshold values, and correlations with
cognitive impairment must be established before PET can be
routinely used in the evaluation of AD.
However, in cases in which early diagnosis of AD is critical (such
as in trials for early administration of investigational
therapeutics) or in cases in which the differential diagnosis is
difficult, PET can be combined with careful clinical examination to
improve diagnostic certainty.
Similarly, perfusion MR can be used to show areas of decreased
cerebral blood flow and cerebral blood volume. Compared with SPECT,
perfusion MR allows superior spatial resolution for distinguishing
between multi-infarct dementia, which has multifocal, subcortical
decreases in cerebral blood flow, and AD, which has generalized
cortical decreases in cerebral blood flow.
Perfusion MR has been found to show localized findings of decreased
cerebral blood volume in the temporoparietal lobes with 90%
sensitivity and 87% specificity in patients who have AD.
More recent research using a transgenic mouse model of AD suggests
that regional decreases in cerebral blood volume in the
hippocampus, thalamus, and temporoparietal cortex may occur prior
to the development of any neurological dysfunction or
histopathologic evidence of AD.
However, like PET and SPECT, the clinical utility of perfusion MR
in the diagnosis of AD is still undergoing evaluation and is
currently limited to special cases.
Volumetric analysis has undergone more thorough evaluation than
any other imaging technique in AD. Both CT and MRI, as well as
autopsy correlation, show global atrophy in patients with AD. For a
single patient, the amount of atrophy has not been shown to
correlate with clinical dys-function.
However, in longitudinal studies, patients with AD have a higher
rate of global atrophy than do age-matched controls.
MRI offers superior contrast resolution and, in some cases,
higher spatial resolution than does CT and can be used to perform
volumetric analysis of brain subregions. Temporal lobe structures
are found to be more atrophic in patients with AD than in
age-matched controls, with specific involvement of the hippocampus
and entorhinal cortex early in AD.
Multiple studies have reported that volumetric atrophy of the
hippocampus is a sensitive indicator of AD.
However, these examinations are limited in specificity, partly due
to limitations in the ability to discriminate microstructural
elements. Also, they have been difficult to correlate between
centers and, despite the high sensitivity, have not become part of
the routine MRI for dementia. However, Adachi and colleagues
have recently developed a DWI approach to evaluate the hippocampus
that very closely parallels the histopathologic findings at autopsy
The addition of DWI to clinical MRI has transformed the clinical
evaluation of stroke, of which multi-infarct dementia is a part.
The role of DWI in stroke is beyond the scope of this paper and has
been extensively reviewed elsewhere.
In the case of AD, DWI allows for improved MR contrast within the
hippocampus (Figure 2).
Investigational imaging in dementia: Diffusion tensor
Diffusion tensor imaging allows for quantitative evaluation of
the rate and direction of water motion (Figure 3). The axonal
organization of white matter tracts in the brain is such that water
motion is limited in directions perpendicular to the axons. The
rate of diffusion, or diffusivity, is measured by the apparent
diffusion coefficient (ADC) (related to the "trace" image on
clinical scanners). Directionality is measured by relative or
fractional anisotropy (FA). The diffusion tensor is a mathematical
model of diffusion in 3-dimensional (3D) space, where anisotropy is
used to represent the shape of the tensor ellipsoid. To obtain this
3D information, we must apply diffusion gradients in a minimum of 6
-values). Measurements of anisotropy in individual voxels can be
grouped to obtain estimates of the trajectories of fiber tracts, a
process known as tractography (Figure 4).
Tractography has shown losses of tract-count (number of fibers
comprising a tract) and in the FA of the fibers within counts in
frontal-occipi-tal and thalamo-frontal tracts in patients suspected
of early AD (Figure 5).
This corresponds to quantitative electroencephalography (EEG)
findings of a loss of coherence between the frontal and occipital
White matter damage can also be assessed in AD by the placement
of regions-of-interest (ROIs) in specific areas of the brain. White
matter values of ADC and FA in a small cohort of AD patients have
been established by Bozzali and colleagues.
More recently, investigators at the University of Rochester used
diffusion tensor MRI to evaluate the corpus callosum in AD and
found a decrease in FA (Figure 6). Neuropsychological examinations
of the same patients found a direct relationship between decreasing
FA in the corpus callosum and cognitive decline.
A recent, larger study established values of ADC and FA for young
adults, elderly adults without dementia, and AD patients.
In this study, researchers at Washington University found decreased
FA in anterior white matter structures for elderly patients without
dementia but in both anterior and posterior structures for elderly
patients with AD (Figure 7).
The improved resolution and contrast of temporal lobe structures
available with investigational diffusion MR sequences (Figure 2)
also allows for quantitative diffusion tensor analysis of the
hippocampus. In research presented at the 2004 meeting of the
International Society for Magnetic Resonance in Medicine (ISMRM),
Kantarci and colleagues
followed-up 89 elderly individuals over a mean time period of 40
months with serial high-resolution hippocampal MRI and clinical
examinations. They showed that ADC abnormalities in the hippocampus
can be used to predict later development of mild cognitive
impairment and AD. Specifically, those patients with amnestic mild
cognitive impairment and an elevated baseline hippocampal ADC were
more likely to convert to AD during the time of the study protocol,
and baseline hippocampal ADC predicted future progression to AD
with 70% sensitivity and 75% specificity.
Also at the 2004 ISMRM meeting, Samann and colleagues
presented a template-derived histogram approach to this hippocampal
ADC analysis. Using this semiautomated approach, patients with
early AD can be separated from age-matched controls. This method
holds great potential for clinical utility, as it allows for the
type of streamlined analysis necessary in reviewing clinical
Diffusion tensor MR also holds great potential in the diagnosis
of Lewy body dementia. Like AD, Lewy body dementia is a fairly
common form of dementia for which there are no specific imaging
findings. It is very difficult to distinguish from AD by
conventional imaging. SPECT studies of Lewy body dementia have
found diffuse cortical hypoperfusion and hypometabolism involving
the occipital lobe that is not present in AD patients.
Similarly, Bozzali and colleagues
recently used diffusion tensor MR to show abnormal diffusivity and
FA in the occipital lobes of patients with Lewy body dementia,
providing further evidence that this technique will be increasingly
useful in distinguishing among various forms of dementia.
Validation of diffusion tensor approaches to the diagnosis of
dementia still relies upon clinical follow-up information, and
whenever possible, postmortem correlation with imaging findings.
Fortunately, diffusion tensor MR can be performed on fixed tissue
samples. These postmortem correlations of imaging and histologic
features offer ongoing validation of the imaging technique and
remain a current area of research in dementia.
Ongoing research also continues in the imaging of animal models
of AD. Transgenic mouse models of AD are based upon overexpression
of the amyloid precursor protein (APP), that gives rise to the
β-amyloid plaques which, in excess, are pathognomonic for AD. These
mice develop the classic AD plaques and progressive neurological
decline found in AD patients. MR imaging studies of these mice
allow for high-field (typically 4.7T and higher), high-resolution
(<0.5 mm) in vivo and ex vivo imaging with direct
histopathologic correlations. Recently, this approach has been used
to identify age-dependent white matter injury in APP mice,
to identify hippocampal changes in FA that precede the development
of β-amy-loid plaques,
and to correlate changes in cortical diffusively with expansion of
the volume of extracellular space in APP mice.
Other investigational uses of MRI in dementia
Further investigation into the utility of MR volumetrics, MR
spectroscopy (MRS), perfusion MR, and functional MRI (fMRI) is
ongoing. In the area of volumetric analysis, clinical utility has
been limited due to the overlap of normal aging and AD. Recently,
however, it has been reported that in individual patients
followed-up with serial MR examinations, the rate of atrophy
directly correlates with the progression of disease.
This research has direct clinical applications, as whole-brain
volumetric calculations can be easily applied to many of the
clinical MR sequences currently in use. As patients return for MRI
for other indications, a volumetric analysis could be added for
those patients with suspected or diagnosed AD. Further research
continues to establish parameters for volume loss in the
hippocampus, the entorhinal cortex, and other temporal lobe
MR spectroscopy is another technique that may be useful for the
diagnosis of AD in difficult cases. In the setting of
encephalopathy, MRS can be useful in differentiation of hepatic
encephalopathy (elevated levels of glycine and glutamate
and changes in levels of GABA
) and HIV encephalopathy (decreased N-acetylaspartate [NAA] and
levels). More recently, at 3T, MRS has been used to distinguish
between AD, frontotemporal dementia, and progressive supranuclear
palsy based upon ratios of NAA, choline, and myoinositol levels to
Functional MRI utilizes changes in blood oxygenation to identify
areas of brain activation. The difference in the fMRI signal
between activated and nonactivated brain areas at 1.5T is small (2%
to 4%) but significant. There are 3 main types of neuropsychologic
tests currently under investigation with fMRI: Encoding of recent
memory, simple cognitive tasks, and image memory. However, there
are no clear standards available yet for the routine use of fMRI in
patients with dementia.
Finally, a relatively new MR technique, T1ρ (T1-rho) imaging,
offers a new alternative to standard T1, T2, and DWI contrast. T1ρ
gives contrast that is somewhere between T1 and T2. T1ρ contrast is
based upon the spin-lattice relation in the rotating frame of
reference. At the 2004 meeting of the ISMRM, T1ρ was preliminarily
reported to directly identify areas of β-amyloid plaque in an
APP-mouse model of AD (manifested as loss of signal intensity on
Alzheimer's disease presents a diagnostic and prognostic dilemma
for which new imaging techniques will need to emerge as progress on
treatments continues. Current guidelines from the neurologic
literature recommend routine CT and MRI examination of all patients
with newly diagnosed dementia. Although there are many imaging
modalities for AD currently under investigation, MRI offers a
single, noninvasive examination allowing for the measurement of
changes in multiple parameters. These include relaxation, water
diffusivity, volumetric analysis, brain perfusion, metabolic
analysis, and functional imaging. Diffusion tensor MR is emerging
as a sensitive and specific tool for identifying patients with
early and late AD and as a method for evaluating disease
progression-information that will be critical to referring
physicians as new pharmaceuticals become available.