PET in Dementia and Memory Evaluation: What Scans Can and Cannot Say

When memory or thinking changes, families often hear that a brain scan may help, and then wonder what a scan could possibly add to a long conversation with a doctor. The honest answer is that PET adds one kind of information, and only one. It shows how brain tissue is behaving at a molecular level: how much sugar it burns, or whether particular proteins have collected there. It does not read thoughts, it does not predict the future with certainty, and it does not replace the person who knows the patient's history. This page explains what the three main brain PET approaches show in dementia evaluation, how they sit alongside cognitive testing and MRI, and where the limits are. Your own neurologist or memory specialist and the imaging team have the final word on any individual case, and protocols differ between centres.

Where PET fits in a memory evaluation

A memory evaluation usually starts with a history from the patient and someone who knows them well, a physical and neurological examination, and bedside or formal cognitive testing. Blood tests look for treatable contributors. Structural imaging, most often MRI, is used to look for other causes. The Alzheimer's Association describes MRI and CT as primarily used to rule out conditions that have different treatments, such as tumours, strokes, injury or fluid build-up, and describes brain imaging as one part of a diagnosis that also includes history, examination and other tests.

PET usually comes later, and only for some people. It tends to be considered when the picture is unclear after that first round, when symptoms start earlier than expected, or when the pattern of symptoms is unusual. A neurologist or memory-clinic specialist weighs whether the answer would change care, and a specialist physician reads the scan and reports back to the referring doctor. If you want to know what the words in that report mean, see reading a PET report.

FDG-PET: patterns of reduced metabolism

FDG is a glucose-like tracer. Active nerve cells and their connections use glucose, so the scan maps where the brain is using less than expected. Reviewers describe this as a proxy for neurodegeneration, a way of seeing damaged or underactive regions, and not as a direct view of any one disease protein. For background on how tracers are chosen and made, see PET radiotracers explained.

What matters is the pattern. A review in Applied Radiology describes Alzheimer's disease as typically showing reduced metabolism in the lateral temporal and parietal cortex, the precuneus, the posterior cingulate and the medial temporal lobes, usually on both sides, with the visual cortex and the sensory-motor strip relatively preserved. The same review describes frontotemporal dementia as depending on the variant: frontal reduction in the behavioural form, and asymmetric, often left-sided, temporal or language-area reduction in the language forms. Dementia with Lewy bodies often shows reduced metabolism at the back of the brain, in the occipital region, while the posterior cingulate is relatively spared, a feature called the cingulate island sign. Vascular patterns vary with the territories affected.

The limits are real, and the sources are open about them. Applied Radiology notes that FDG uptake reflects regional neurodegeneration and is not specific to one disease, that about 70 percent of people with dementia with Lewy bodies also have coexisting Alzheimer's pathology, so patterns overlap, and that sedatives, antipsychotics, corticosteroids and some other medicines can alter the picture. A normal scan substantially lowers the likelihood of a neurodegenerative cause. A review in Practical Neurology adds that, in the United States, FDG-PET was approved for separating Alzheimer's disease from frontotemporal dementia, and that reimbursement at the time of writing was limited to cases where those two overlap and the standard workup had not settled the question. Coverage rules change, so that detail may be out of date where you live.

Amyloid PET: looking for plaques

Amyloid PET uses tracers that bind to beta-amyloid, the protein that forms plaques in Alzheimer's disease. The 2025 appropriate use criteria from the Alzheimer's Association and the Society of Nuclear Medicine and Molecular Imaging, published in the Journal of Nuclear Medicine, note that three fluorine-18 amyloid tracers are approved by the US Food and Drug Administration, and that visual reads of those scans showed 88 to 98 percent sensitivity and 80 to 95 percent specificity against plaques found at autopsy.

The result is read as positive or negative. A positive scan shows that moderate or frequent plaques are probably present. A negative scan makes Alzheimer's disease an unlikely explanation for symptoms, although the same criteria add that a negative result does not equate to no amyloid at all, only to low levels that are less likely to cause impairment. The revised diagnostic criteria published in Alzheimer's & Dementia make the same point from the other side: amyloid PET is not sensitive to low levels of Alzheimer's pathology.

What amyloid positivity does and does not mean

This is the part of the topic most likely to cause fear, so it is worth slowing down. A positive amyloid scan shows that the plaque protein is present. It does not, on its own, say that plaques are the reason for a particular person's memory problems, how fast things will change, or what stage someone is at.

Age matters a great deal. The 2025 appropriate use criteria report that amyloid positivity in people with no cognitive impairment rises from about 10 percent at age 50 to about 45 percent at age 90, and that carriers of the APOE e4 gene variant are two to three times as likely to be positive in each age group. The criteria therefore rate amyloid PET as rarely appropriate in people without cognitive impairment, even when they carry APOE e4, and describe such scans as generally not appropriate for non-medical purposes such as legal, insurance or employment screening.

In people who do have symptoms the picture changes. The revised criteria drew on 4,637 symptomatic people in a US autopsy database who had moderate or frequent plaques, and found that 95 percent also had the tangle pathology that qualifies as intermediate or high Alzheimer's change. On that basis the workgroup concluded that an abnormal amyloid PET scan, or an equivalent fluid test, is sufficient to establish a diagnosis of Alzheimer's disease in a symptomatic person. They also state that the clinical use of these biomarkers is presently intended for symptomatic individuals, that clinical presentation alone is not diagnostic, and that copathology is common, so clinical judgment is always needed to decide whether Alzheimer's disease is the cause, or a dominant part of the cause, of a person's symptoms.

Tau PET: tangles and staging

Tau PET targets the tau protein tangles that spread through the brain as Alzheimer's disease advances. The first approved tracer, flortaucipir, gained US approval in 2020. Per the appropriate use criteria, its visual read showed 92 percent sensitivity and 80 percent specificity against tangle pathology at Braak stage V or higher, but only about 20 percent of people with pathology at Braak stage III to IV read as positive, so a negative scan may not rule out clinically meaningful tau. The workgroup expressed lower confidence in tau PET than in amyloid PET, and many of its ratings rest on expert opinion.

In the revised criteria, amyloid PET sits in a first group of core biomarkers that define the presence of the disease, while tau PET is used with that first group to stage biological severity and to judge how likely it is that symptoms are due to Alzheimer's disease.

How the scans relate to cognitive testing and MRI

Cognitive testing measures what the brain is doing: memory, attention, language, planning. MRI shows structure, including shrinkage patterns and signs of strokes or other lesions. Practical Neurology notes that in mild cognitive impairment and early Alzheimer's disease, shrinkage appears first in the hippocampus and the neighbouring entorhinal cortex, and that grading this shrinkage remains somewhat subjective. PET shows molecular and metabolic information. None of the three can stand in for the others.

The appropriate use criteria put this in a practical form. Amyloid or tau PET should not be used as a substitute for a comprehensive clinical examination. The criteria suggest PET when Alzheimer's disease is a likely cause, when the cause remains uncertain even after a dementia expert has evaluated the person, and when the result is expected to change management. They also call for shared decision-making that takes the patient's wishes and possible out-of-pocket costs into account.

Imaging and anti-amyloid treatment discussions

Some newer medicines target amyloid. The US Food and Drug Administration's announcement of the full approval of lecanemab in July 2023 states that treatment should be started in people at the mild cognitive impairment or mild dementia stage, and that the confirmatory trial enrolled patients with confirmed amyloid beta pathology. It also describes a boxed warning about amyloid-related imaging abnormalities, temporary swelling that is seen on imaging, and says testing for ApoE e4 status should be done before starting treatment.

The appropriate use criteria add that the prescribing information and related use recommendations for aducanumab and lecanemab require biomarker evidence of amyloid pathology before therapy starts, and they rate amyloid PET as appropriate for establishing eligibility. They also note that amyloid PET is not the only route, since cerebrospinal fluid tests and accurate blood tests are alternatives in the revised criteria. A European review in Alzheimer's & Dementia: Diagnosis, Assessment & Disease Monitoring reports that amyloid PET and cerebrospinal fluid results disagree in around 10 percent of clinical cases. Whether a particular treatment is available, suitable or covered is a decision for the patient, the family and the treating specialist; this page cannot answer it.

Access and coverage vary widely

In the United States, the appropriate use criteria record that the Centers for Medicare and Medicaid Services retired its 2013 coverage restriction on amyloid PET in October 2023. The European review mentioned above says that amyloid PET has been reimbursed by Medicare since 2023 and is regularly available, though almost only at major academic centres, whereas access is quite limited across much of Europe; in Germany, despite a well-resourced health system, amyloid PET is rarely reimbursed and performed at low volumes. Elsewhere in the world, the picture depends on local scanner capacity, tracer supply and payer policy, and it keeps changing. The practical step is to ask the referring doctor and the imaging centre what is offered and what it costs.

The day itself is covered in what to expect on scan day, and radiation in radiation dose in context. For other brain uses see PET in movement disorders and PET in epilepsy surgery planning, and for combined scanners see PET-CT and PET-MRI.

Honest limits

FDG patterns suggest which networks are affected, and they overlap between diseases. Amyloid status is a statement about plaques, and plaques are common in older age. Mixed pathology is frequent, and no scan reports how much each contributor matters. Blood tests are being developed that may reduce how often PET or a lumbar puncture is needed, and guidance is moving quickly, so the best route is always the one your own specialist recommends.

The short version

PET adds metabolic and molecular information to a memory evaluation, and it is most useful when the cause stays uncertain after the clinical workup. FDG shows patterns of reduced activity, amyloid PET shows plaques, and tau PET shows tangles, and each has stated limits. A positive amyloid scan needs interpreting in light of age, symptoms and the rest of the evidence, and the decision to scan belongs with a neurologist or memory specialist.