Brain
Dementia: Sort It by Atrophy & Metabolism
Fellowship-level neuroradiology teaching by E. Brooke Schrickel, MD. Open this topic in the interactive reading room →
Core clinical idea
The 'rule out dementia' requisition is really AD vs FTD vs non-neurodegenerative (for cognitive cases) or which PPA subtype (for language cases). Sort the case by atrophy and FDG distribution, and never miss the reversible mimic (NPH) or the fast fatal one (CJD).
Bottom line
Read the dementia brain as a pattern-sort into the neurologist's question — and never let NPH or CJD slip past in a routine 'dementia' study.
Core workstation questions
- Is the dominant pattern posterior/mesial-temporal (AD), frontal (FTD), or asymmetric parieto-occipital (PCA)?
- Are the DLB signs present — cingulate island sign, occipital tunnel sign, abnormal DAT scan?
- Could this be NPH — ventricles out of proportion to sulci, tight high-convexity/vertex, acute callosal angle, relatively spared hippocampi, with the gait/urinary/cognitive triad?
- Any cortical, basal ganglia, or pulvinar restricted diffusion suggesting CJD?
- For language cases: left perisylvian (logopenic), anterior bitemporal (semantic), or frontal/insular (nonfluent)?
- How heavy is the vascular load (white matter burden, lacunes, microbleeds)?
What changes reporting / management
- NPH is one of the few reversible dementias — suggesting it early (out-of-proportion ventricles, tight vertex, acute callosal angle, triad) can lead to symptom-reversing shunting.
- CJD read (cortical ribboning + basal ganglia + pulvinar restriction) changes prognosis, isolation, and counseling immediately; pair with CSF assay.
- Name DLB (FDG signs + abnormal DAT) rather than calling it AD — it carries neuroleptic-sensitivity and faster-decline implications; amyloid PET won't separate DLB from AD.
- Describe white-matter burden as scattered/patchy/confluent plus lacunes and microbleeds to convey vascular contribution, instead of a formal grade.
- Don't overclaim 'consistent with AD' off FDG alone — FDG is far less specific for AD-vs-other dementia than AD-vs-normal.
Practical traps
- PCA mistaken for an eye problem (cataract workup) before the brain is imaged — look for asymmetric parieto-occipital atrophy.
- Trusting amyloid PET to distinguish DLB from AD when DLB is amyloid-positive in over half of cases.
- Reciting named grading scales (Scheltens, Fazekas, ATN) as if clinically used when practice is qualitative mild-vs-severe / scattered-vs-confluent.
- Missing NPH or CJD inside a routine 'involutional change' dementia read.
Teaching pearls
- Atrophy posterior = AD-spectrum; atrophy frontal = FTD.
- DLB: cingulate island + occipital tunnel + abnormal DAT — don't trust amyloid PET to separate it from AD.
- PCA hides behind the eye complaint; asymmetric parieto-occipital atrophy is the tell.
- Cortex + basal ganglia + pulvinar restricting = CJD; get the CSF assay.
- NPH is the reversible dementia — out-of-proportion ventricles, tight vertex, acute callosal angle: say it.
- Logopenic PPA is the AD-type aphasia; semantic and nonfluent are FTLD-type.
Source lectures
- Dementia Imaging: Typical and Atypical Causes
Educational material for radiology residents and neuroradiology fellows. Nothing here drives individual patient care, and it contains no patient data.