Dementia Imaging (Alzheimer/FTD)
- Dementia imaging is mostly a rule-out and pattern job: first prove there's no tumor, bleed, or fixable cause, then look at where the brain has shrunk.
- The pattern of atrophy is the clue. Alzheimer disease loves the medial temporal lobe (the hippocampus); frontotemporal dementia (FTD) prefers the frontal and anterior temporal lobes.
- Structural MRI tells you what's gone. Molecular imaging (FDG-PET, amyloid PET) tells you what's not working or what's accumulating, and often shows trouble before the brain visibly shrinks.
- Atrophy is regional, not just "old brain." The trick is comparing one region to its neighbors and to the patient's age.
Here's the uncomfortable truth about imaging dementia: you usually can't see the disease itself. There's no glowing blob to point at. What you can see is the aftermath — the brain quietly losing volume in a telltale pattern, like a beach where the tide went out and never came back. Your job is to read the shape of the shoreline.
First, the boring-but-important part
Before anyone gets fancy about Alzheimer versus frontotemporal dementia, imaging has a humbler first mission: make sure this isn't something treatable masquerading as dementia. A slow-growing tumor, a chronic subdural bleed, hydrocephalus, or a pile of small strokes can all cause cognitive decline that looks a lot like a degenerative disease from the outside.
So the first read of a "rule out dementia" CT or MRI is essentially: is there anything here I'd be embarrassed to miss? Only once the answer is "no" do we move on to the atrophy-pattern detective work.
CT is the workhorse for excluding the scary structural stuff — bleed, mass, big hydrocephalus. MRI is where the real characterization happens, because it shows regional atrophy and small-vessel changes with much more finesse.
Reading atrophy without fooling yourself
Everyone's brain shrinks a little with age — sulci widen, ventricles enlarge, the whole thing looks a touch roomier in its skull. The mistake is calling that "atrophy" and stopping. Real diagnostic value comes from atrophy that's regional and out of proportion: one area noticeably more hollowed-out than its neighbors or than you'd expect for the patient's age.
Think of it like a lawn. Uniformly thinning grass everywhere is just an old lawn. But a bald patch right under the swing set tells you something specific happened there. In dementia, the bald patch is the diagnosis.
Alzheimer disease: the hippocampus takes the hit
Alzheimer disease (AD), the most common cause, has a signature: medial temporal lobe atrophy, centered on the hippocampus. On a coronal MRI through the temporal lobes, you're looking for hippocampi that have shriveled and pulled away from their surroundings, with the adjacent temporal horn of the lateral ventricle ballooning open to fill the gap.
That's the key relationship — when brain volume disappears, the ventricular space next door expands to take its place. Tissue shrinks, fluid moves in. As things progress, atrophy spreads to the parietal lobes too, but the medial temporal lobe is the classic early calling card.
A single shrunken-looking hippocampus on one slice isn't a diagnosis. Atrophy must be judged in context — both sides, several slices, and against the patient's age. And don't forget the mimics: medial temporal atrophy can also appear in other conditions, so the imaging supports a clinical picture rather than replacing it.
Frontotemporal dementia: the front of the house
Frontotemporal dementia (FTD) goes after a different neighborhood: the frontal lobes and the anterior (front) temporal lobes. Clinically these patients often present younger and with personality or language changes rather than the memory-first story of AD — and the imaging mirrors that, with atrophy concentrated up front while the back of the brain stays relatively plump.
The contrast is the teaching point. AD hollows out the medial temporal region first; FTD hollows out the frontal and anterior temporal regions. Same disease category ("the brain is shrinking"), completely different floor plan.
| Feature | Alzheimer disease | Frontotemporal dementia |
|---|---|---|
| Atrophy hotspot | Medial temporal lobe / hippocampus, then parietal | Frontal & anterior temporal lobes |
| Typical onset | Often later age | Often younger |
| Early clinical clue | Memory loss | Behavior or language change |
| FDG-PET pattern | Reduced metabolism in parietotemporal regions | Reduced metabolism frontally / anterior temporally |
When shrinkage isn't the whole story
Two more things belong on every dementia read. First, vascular contribution: widespread small-vessel ischemic disease — those bright white-matter spots and old lacunes — can cause or worsen cognitive decline, and AD and vascular damage frequently coexist in the same brain. It's rarely a tidy single diagnosis.
Second, molecular imaging picks up where anatomy stops. Structural MRI shows tissue that's already gone; metabolism falters earlier. That's where brain PET (FDG, amyloid, DAT) earns its keep — FDG-PET maps regions where the brain is running cold (reduced glucose use) in a disease-specific distribution, and amyloid PET looks directly for the protein buildup of AD.
A helpful mental model: MRI answers "what does the architecture look like, and is there a fixable cause?"; FDG-PET answers "which regions have stopped pulling their weight?"; amyloid PET answers "is the Alzheimer protein actually here?" They're complementary, not competing.
The takeaway
Dementia imaging is reading absence. Exclude the treatable mimics first, then ask the one question that matters most: where is the brain disappearing? Hippocampus-and-temporal-lobe points toward Alzheimer; front-of-the-brain points toward FTD. Anatomy from MRI, function from PET, and a healthy respect for the fact that real brains often have more than one thing going wrong at once.