As the U.S. population continues to age, dementia is becoming an
increasingly prevalent problem. After laboratory testing has
excluded the various metabolic and hormonal abnormalities that can
cause dementia, imaging may be required to make a definitive
diagnosis. Many of the possible diagnoses (e.g., frontal lobe
tumors, subdural hematomas, and normal pressure hydrocephalus) are
treatable. Magnetic resonance imaging and spectroscopy are clearly
the preferred techniques for this evaluation, as described in this
issue of Applied Imaging.
--William G. Bradley, Jr., MD, PhD, FACR
MRI of Dementia
The physician's primary role in the evaluation of dementia is to
determine its cause and, if possible, reverse it. Unfortunately,
the most common cause of dementia--Alzheimer's disease or senile
dementia of the Alzheimer's type, which accounts for 50% of
dementia cases-- isn't currently treatable.
While other causes of dementia are treatable (alcoholism, 10% of
cases; metabolic abnormalities, 10% [e.g., vitamin B
deficiency]; and hormonal abnormalities, 10% [e.g.,
hypothyroidism]), this issue of Applied Imaging will focus on those
causes of dementia which are diagnosable by magnetic resonance
imaging (MRI) (20% of cases). The most common causes of dementia in
this category are multi-infarct dementia, normal pressure
hydrocephalus, frontal lobe tumors, and subdural hematomas.
In addition, this issue of
will also briefly address the emerging application of MR
spectroscopy to diagnose certain causes of dementia.
Multi-infarct dementia accounts for approximately 10% of all
cases of dementia.
Multi-infarct dementia appears as multiple areas of infarction
involving the cerebral cortex, deep white matter, and/or basal
On MR images these appear as asymmetric areas of focal cortical
loss, often bordered by high-signal-intensity gliosis on
proton-density weighted images (Figure 1). The subarachnoid space
is widened and the ventricles are comparably enlarged, the classic
(but nonspecific) findings of atrophy. Deep white matter and
lacunar infarcts also appear as areas of cerebrospinal fluid (CSF)
intensity (similar to that in the ventricles) lined by the higher
intensity gliosis (Figure 2). These are the chronic MRI findings of
multi-infarct dementia. As noted in a previous issue of Scan
Digest, acute infarcts
are also bright on diffusion images (Figure 3). Generally,
diffusion imaging can only be performed on the newer generation of
echo planar-capable MR units.
Occasionally, altered levels of consciousness without focal
neurologic findings can be produced by acute frontal lobe infarcts.
Normal Pressure Hydrocephalus
Approximately 6% of all cases of dementia are caused by normal
pressure hydrocephalus (NPH),
which is characterized clinically by a triad of gait disturbance,
dementia, and incontinence.
Typically, these patients are elderly and have ventricles enlarged
out of proportion to any sulcal enlargement (Figure 4), i.e., the
pattern of communicating hydrocephalus instead of atrophy.
With initial ventricular enlargement, tangential shearing forces on
the paracentral corona radiata fibers lead to the gait disturbance.
As the ventricles continue to enlarge and the cortex is pushed
against the inner table of the calvarium, radial shearing forces
lead to dementia.
Ventriculoperitoneal (VP) shunting is a curative procedure in many
patients and at least slows the progression in others.
MR imaging has proven to be extremely useful for the evaluation
of patients with possible NPH who are being considered for VP
When marked CSF signal loss is noted on proton-density weighted
images (Figure 4) extending from the anterior third ventricle to
the bottom of the fourth ventricle, appropriately symptomatic
patients have been shown to have a very good chance of responding
to VP shunting.
This "CSF flow void" (Figure 4B) reflects the increased to-and-fro
motion of CSF in these patients. Since this CSF motion is driven by
the rhythmic expansion and contraction of the brain over the
cardiac cycle, brain atrophy (and its associated decreased blood
supply) decreases the CSF flow void. Unfortunately, modern MRI
techniques (e.g., flow compensation and fast spin echo) also
decrease the magnitude of the CSF flow void. Thus the sign is
specific, but not very sensitive.
More recently, phase contrast flow-quantifying MRI techniques
have been used with even greater success to select appropriately
symptomatic patients for shunting (Figure 5).
These techniques involve positioning an MR slice perpendicular to
the aqueduct to measure the volume of CSF going craniocaudad in
systole or caudocraniad in diastole. The average of these volumes
is known as the "aqueductal CSF stroke volume." In a recent study,
all patients with an aqueductal CSF stroke volume above a certain
value (42 (µL) responded to shunting while only half of those with
an aqueductal CSF stroke volume less than that value responded,
reflecting a degree of concomitant atrophy.
It should be emphasized that shunting is most successful when the
gait disturbance is more pronounced than the dementia.
Such patients should have conventional MRI first, and if a
hyperdynamic CSF flow void is seen, they should undergo shunting.
If the CSF flow void is not increased in appropriately symptomatic
patients, central atrophy may have ensued
or the MRI technique may be insensitive. Such patients should then
undergo phase contrast CSF velocity imaging to ascertain
appropriateness of shunting.
It has been reported that 5% of all cases of dementia are caused
by tumors or subdural hematomas.
Tumors in certain parts of the brain can achieve considerable size,
leading to increased intracranial pressure and dementia without
producing focal neurologic symptoms. These tumors are more common
in the right frontal (Figure 6) and right temporal lobes, away from
the speech centers which are usually in the left hemisphere.
They may involve the genu of the corpus callosum. The classic
lesion is the frontal lobe meningioma, which has a particularly
insidious onset due to its slow growth. Such tumors are diagnosed
easily with gadolinium-enhanced MRI.
Like tumors, subdural hematomas increase the intracranial
pressure, leading to dementia.
With normal aging, the volume of the brain decreases while the
volume of the calvarium obviously does not. This shrinkage
stretches the bridging veins that pass from the brain to the dural
sinuses. With minor trauma, these vessels tear, leading to subdural
hematomas (Figure 7).
One of the advantages of MRI compared to computed tomography
(CT) is the ability to detect and stage hemorrhage. Subacute
subdural hematomas are frequently isodense to brain on CT. Five
distinct stages of hemorrhage can be recognized on MRI,
based on their signal intensity on T
- and T
-weighted images (Table). Thus, not only can the age of the
subdural hematoma be estimated, but when rebleeding occurs, it can
be distinguished from the original hemorrhage based on its T
characteristics (Figure 7B).
Spectroscopy of Alzheimer's Disease
Although it is well known that the temporal horns and parietal
sulci are prominent in Alzheimer's disease, these findings are both
insensitive and nonspecific.
Recently, MR spectroscopy has been used to make this diagnosis.
MR spectroscopy (MRS) is a new high-field MR application that is
being increasingly implemented on a widespread basis. Typically a
2*2*2 cm voxel is positioned over a lesion or in standardized
locations in the gray or white matter. In 3 minutes, MRS provides a
quantitative measure of 7 or so metabolites in the brain.
The relative concentration of these metabolites is often suggestive
of a particular diagnosis. Tumors, for example, have elevated
choline (from membrane turnover), decreased N-acetyl aspartate
(NAA) [which is a neuronal marker], and lipid and lactate peaks
that are not normally present (Figure 8). In the case of
Alzheimer's disease, NAA is decreased and the osmolyte myoinositol
is increased (Figure 9).
In conclusion, MRI is the imaging modality of choice for the
evaluation of dementia, once metabolic and hormonal causes have
been excluded. In addition to being able to diagnose treatable
causes of dementia such as normal pressure hydrocephalus, frontal
lobe tumors, and subdural hematomas, MRI is also useful to
demonstrate certain causes of dementia before they become totally
untreatable, e.g., multi-infarct dementia. The latest MR
application, spectroscopy, is also showing promise in diagnosing
the most common cause of dementia, Alzheimer's disease.
1. Adams RD, Victor M: Principles of Neurology 2nd ed, pp
285-295. New York, McGraw Hill, 1981.
2. Jensen MC, Brant-Zawadzki MN, Jacobs BC: Ischemia. In: Stark
DD, Bradley WG (eds): Magnetic Resonance Imaging 3rd ed, pp
1255-1275. St. Louis, Mosby and Co., 1999.
3. Maeda M, Abe H, Yamada H, Ishii Y: Hyperacute infarction: A
comparison of CT and MRI, including diffusion-weighted imaging.
Neuroradiology 41(3):175-178, 1990.
4. Bradley WG, Chen D-Y, Atkinson D, Edelman RE: Fast spin-echo
and echo-planar imaging. In: Stark DD, Bradley WG (eds): Magnetic
Resonance Imaging 3rd ed, pp 125-158. St. Louis, Mosby and Co.,
5. Adams RD, Fisher CM, Hakim S, et al: Symptomatic occult
hydrocephalus with "normal" cerebrospinal fluid pressure: A
treatable syndrome. NEJM 273(3):117-126, 1965.
6. Hakim S, Venegas JG, Burton JD: The physics of the cranial
cavity, hydrocephalus and normal pressure hydrocephalus: Mechanical
interpretation and mathematical model. Surg Neurol 5:187-209,
7. Bradley WG, Whittemore AR, Kortman KE, et al: Marked
cerebrospinal fluid void: Indicator of successful shunt in patients
with suspected normal-pressure hydrocephalus. Radiology
8. Bradley WG, Scalzo D, Queralt J, et al: Normal-pressure
hydrocephalus: Evaluation with cerebrospinal fluid flow
measurements at MR imaging. Radiology 198(2):523-529, 1996.
9. Graff-Radford NR, Godersky JC: Normal-presure hydrocephalus:
Onset of gait abnormality before dementia predicts good surgical
outcome. Arch Neurol 43:940-942, 1986.
10. Bradley WG, Whittemore AR, Watanabe AS, et al: Association
of deep white matter infarction with chronic communicating
hydrocephalus: implications regarding the possible origin of
normal-pressure hydrocephalus. AJNR 12(1):31-39, 1991.
11. Bradley WG, Quencer RM: Hydrocephalus and cerebrospinal
fluid flow. In: Stark DD, Bradley WG (eds): Magnetic Resonance
Imaging 3rd ed, pp 1483-1507. St. Louis, Mosby and Co., 1999.
12. Bradley WG: MR appearance of hemorrhage in the brain.
Radiology 189(1):15-26, 1993.
13. Barclay LL, Linden C, Murtagh R: Medial temporal atrophy as
a magnetic resonance imaging marker for Alzheimer's disease. J
Neuroimaging 2(3):131-135, 1992.
14. Shonk TK, Moats RA, Gifford P, et al: Probable Alzheimer
disease: Diagnosis with proton MR spectroscopy. Radiology 195:65,
15. Danielsen ER, Ross BD: Neurospectroscopy. In: Stark DD,
Bradley WG (eds): Magnetic Resonance Imaging 3rd ed, pp
Clinical Quiz: True or False
1. Most causes of dementia can be diagnosed by MRI.
2. Any patient needing an imaging study for the evaluation of
dementia should have an MRI instead of a CT.
3. Shunt-responsive normal-pressure hydrocephalus can be
diagnosed reliably with MRI.
4. Frontal lobe brain tumors causing dementia can be diagnosed
as easily with CT as with MRI.
5. Subdural hematomas causing dementia can be diagnosed equally
well by CT and MRI.
1. False. Alzheimer's accounts for 50% of cases of dementia, and
when alcoholism (10%) and metabolic and hormonal abnormalities
(20%) are factored in, imaging can explain only about 20% of all of
the causes of dementia.
2. False. Certain patients are excluded from having an MR scan,
e.g., those with cardiac pacemakers, ferromagnetic intracranial
aneurysm clips, and neurostimulators.
3. True. Some patients with NPH can be diagnosed solely on the
basis of hyperdynamic CSF flow through the aqueduct; others will
require quantitative phase contrast CSF velocity imaging.
4. True. Large supratentorial masses should be equally well
diagnosed with either of the two techniques.
5. False. Subacute subdural hematomas are notoriously isodense
with brain on CT and, therefore, difficult to detect. With both T
- and T
-weighted images, there is no stage of hemorrhage that is
isointense to either brain or CSF on MRI, simplifying the