Basic Neurosciences · Paper A
Alzheimer's disease neuroimaging
Every dementia leaves a different fingerprint on a brain scan, and Alzheimer's disease has a very specific one: the memory-forming hippocampusA brain structure central to forming new memories.Read more → shrinks first, the brain regions that light up during thinking (the posterior cingulate and temporoparietal cortex) go quiet on a metabolic scan, and, if you look directly for the sticky amyloid protein behind it all, a dedicated PET scan can show it accumulating in the cortex. No single one of these findings proves the diagnosis on its own — it is the combination, and the pattern of WHERE the changes are, that tells Alzheimer's disease apart from vascular damage, Lewy body disease, frontal degeneration, or a plumbing problem in the ventricles that only looks like dementia.
When someone develops a memory problem, the honest starting question is not 'which disease is this' but 'is this even a brain-shrinking (neurodegenerative) problem at all, or something else entirely, including something we could actually fix?' A brain scan is how you answer that question with evidence instead of a guess. Understanding what each type of scan actually measures — a shape, an activity level, or a specific sticky protein — is what lets you read a scan report instead of just seeing the word 'abnormal' and stopping there.
'Neuroimaging' here means using a brain scan to look for the specific pattern of change that different causes of dementia leave behind. A structural scan (MRI) shows the physical shape and size of brain structures — has something shrunk? A functional scan (FDG-PET, an imaging test that tracks where the brain is using glucose for energy) shows which regions are metabolically active — is a region 'quiet' even if it hasn't visibly shrunk yet? A molecular scan (amyloid PET) shows where a specific abnormal protein, amyloid-beta, has built up. Alzheimer's disease has a recognisable answer on all three.
A homely way to hold these apart: imagine you suspect a leak somewhere in an old house. One inspector checks the walls and floors for actual physical damage — cracks, sagging, water stains (that's the structural scan). A second inspector checks which rooms have the lights flickering or barely on, even in walls that still look fine (that's the functional scan). A third tests directly for a specific mould known to cause this kind of damage, swabbing the air in each room (that's the molecular scan). All three can be done on the same house, and each answers a genuinely different question about it.
Here is the trap that catches almost everyone at first: assuming that if a scan shows 'Alzheimer's-type' changes, the patient has Alzheimer's disease dementia. That is not automatically true, and nowhere is it truer than for amyloid PET. Amyloid protein starts building up in the brain years, sometimes decades, before any memory symptoms appear, so a real number of cognitively completely normal older adults will have a 'positive' amyloid scan if you test them. A positive amyloid PET tells you the pathology is present. It does not, by itself, tell you the patient has dementia. The scan is always read together with how the person is actually functioning, never in isolation.
Alzheimer's disease is thought to begin with amyloid-beta protein clumping together outside neurons, forming plaques, over many years before anyone notices anything wrong. This process is believed to trigger a cascade, including a second abnormal protein (tau) building up inside neurons, which more directly correlates with neurons losing function and eventually dying. Because neurons in the hippocampus (the structure most responsible for laying down new memories) are hit early and hard, that is where the earliest visible SHRINKAGE (atrophy) shows up on a structural MRI — which is why an amnestic (memory-loss-first) presentation, with hippocampal atrophy on the scan, is the classic Alzheimer's disease picture.
Before a structure visibly shrinks, though, the neurons in it are often already working less well — using less glucose for energy than a healthy brain region would. That is exactly what an FDG-PET scan detects: it uses a radioactive sugar tracer and measures which brain regions 'light up' less than expected. In Alzheimer's disease, this quietening shows up characteristically in the posterior cingulate, precuneus, and the temporoparietal association cortex (areas involved in integrating memory, spatial orientation and attention), and it can appear before the structural MRI looks clearly abnormal — which is why FDG-PET is genuinely useful when the diagnosis is unclear from MRI alone.
Amyloid PET is the most direct of the three: instead of inferring damage indirectly (shrinkage, or reduced activity), it uses a tracer that physically sticks to amyloid plaques and lights them up directly. This is powerful for confirming that amyloid pathology is present, but, as the caveat above explains, presence of amyloid is not the same as presence of dementia — which is exactly why it is only one piece of a bigger picture, never the whole picture by itself.
Imagine three separate patients, each getting one of these three scans, and imagine what a 'normal' result and an 'Alzheimer's pattern' result would each look like.
Patient one gets an MRI. A radiologist looks at coronal slices through the and rates how big the looks on each side, using a scale from 0 (no shrinkage) up to about 4 (severe shrinkage) — this is a visual rating scale, essentially eyeballing volume loss against reference images. A score of 2 or more, especially if similar on both sides, in someone with progressive memory loss, is consistent with Alzheimer's disease.
Patient two gets an FDG-PET. Rather than a shape, the report describes colour-coded activity across the brain surface — warm colours for high activity, cool colours for reduced activity. In Alzheimer's disease, cool patches appear over the back-and-side (temporoparietal) regions and the posterior cingulate, while the very front (frontal motor strip) and the very back (primary visual cortex) stay warm. That specific 'cool zone' distribution, not just 'some areas are cooler than others', is what a reader is trained to recognise.
Patient three gets an amyloid PET. The report is closer to a yes/no: does the tracer show significant binding across the cortex (positive) or not (negative)? A positive result plus a genuinely progressive memory-first clinical picture, plus ideally a supporting MRI or FDG-PET finding, builds a strong case for Alzheimer's disease. A positive result in someone with no cognitive symptoms at all is a research finding about future risk, not a diagnosis to act on today.
Now contrast this with a fourth patient, whose MRI instead shows big, boxy ventricles that look out of proportion to how much the actual brain surface has shrunk, with a gait that looks like his feet are stuck to a magnetic floor and new urinary incontinence. None of the three Alzheimer's-pattern findings above apply here — this is normal pressure hydrocephalus, a mechanical drainage problem, not a protein-accumulation problem, and unlike everything else in this topic, it is one of the few dementia-like pictures that can genuinely improve with treatment (a shunt to drain the excess fluid).
| Looks similar | How to tell it apart |
|---|---|
| Vascular dementia | Imaging shows infarcts or extensive confluent change, with the medial comparatively spared; decline is classically stepwise, not smooth |
| Dementia with Lewy bodies | Hippocampal volume relatively preserved on MRI; FDG-PET shows occipital hypometabolism with posterior cingulate SPARING (the reverse of Alzheimer's disease); dopamine transporter scan is abnormal |
| Frontotemporal dementia | Atrophy and hypometabolism are anterior (frontal/anterior temporal), not posterior — the mirror image of the Alzheimer's disease pattern |
| Normal pressure hydrocephalus | Ventricles enlarged out of proportion to cortical atrophy, relatively preserved — a mechanical, partly reversible problem |
| Normal ageing | Mild, diffuse atrophy with no focal medial temporal or temporoparietal pattern, in someone with normal cognitive testing |
A clinical diagnosis of Alzheimer's disease is made primarily on the history and cognitive testing (a progressive amnestic syndrome, with other causes reasonably excluded); imaging is not required for every patient. Where the picture is atypical, early-onset, or a reversible cause needs excluding, structural MRI is usually the first-line scan, with FDG-PET or amyloid PET reserved for genuinely unclear cases — for example, distinguishing Alzheimer's disease from frontotemporal dementia when the clinical picture overlaps, or checking for normal pressure hydrocephalus or a vascular contribution before committing to a neurodegenerative diagnosis.
There is no drug that reverses the structural or amyloid changes described here, but the imaging still directly changes management. Finding a reversible or partly reversible pattern — normal pressure hydrocephalus's disproportionate ventriculomegaly, or a significant vascular burden where risk-factor control (blood pressure, diabetes, smoking, per standard vascular-risk guidance) can slow further decline — changes what can actually be offered. For confirmed Alzheimer's disease, standard symptomatic management (acetylcholinesterase inhibitors, memantine at a more advanced stage, per NICE guidance on dementia) proceeds regardless of which specific imaging test confirmed the pattern; the scan's job is to get the diagnosis right, not to select which drug is used.
The exam's favourite trap is a positive amyloid PET presented as if it settles the question — it does not, because amyloid accumulation predates symptoms by years and a real proportion of cognitively normal older people are also amyloid-positive. Before you move on: could you explain to a friend, in one sentence, why a scan can show 'Alzheimer's pathology' in someone who does not have dementia?
Alzheimer's disease neuroimaging is read as a converging pattern — hippocampal atrophy on structural MRI, temporoparietal hypometabolism on FDG-PET, amyloid positivity on molecular PET — and each competing dementia (vascular, Lewy body, frontotemporal, or the reversible mimic normal pressure hydrocephalus) is told apart by exactly which region is hit and which modality shows it, not by the word 'atrophy' alone.