Criteria & tools workbench
Ask what the tool was built to do—and what it has actually proved.
Terminology standards, diagnostic criteria, biomarkers, severity scores, and monitoring frameworks solve different problems. Their names should never substitute for their contexts of use.
Three frequently confused tools
Not competitors. Not interchangeable.
A descriptive standard can be excellent without diagnosing etiology. A diagnostic rule can perform well in a referral population without becoming a screening test. A promising score can remain investigational.
| STRIVE-2 | Boston criteria v2.0 | ARTS | |
|---|---|---|---|
| Primary purpose | Standardize imaging feature definitions and reporting | Classify likelihood of sporadic CAA in defined clinical contexts | Estimate pathology-linked arteriolosclerosis risk for research |
| Output | Phenotype language | Diagnostic category | Risk score |
| Reference logic | Consensus plus feature-specific reliability/validity | Neuropathologic CAA assessment in validation studies | Autopsy pathology in development; independent accuracy incomplete |
| Strongest use | Research harmonization | Symptomatic patients resembling validated populations | Hypothesis-driven research and external validation |
| Dangerous misuse | Treating WMH, PVS, or CMB as an etiologic diagnosis | Universal screening or ignoring mixed pathology and MRI quality | Calling repeatability or outcome association diagnostic validation |
| Open the profile | STRIVE-2 record → | Boston profile → | ARTS profile → |
Distribution as probability
Lobar versus deep is useful—not absolute
Location changes the competing probabilities because vessel compartments differ. Mixed disease, lesion-level exceptions, and imperfect imaging keep the inference probabilistic.
Cortical and leptomeningeal pattern
Strictly lobar hemorrhagic lesions, cortical superficial siderosis, and convexity SAH can increase CAA probability in the right context.
Deep perforator pattern
Deep microbleeds, lacunes, and injury in basal ganglia, thalamus, brainstem, and deep white matter favor non-amyloid arteriopathy.
Mixed and ambiguous
Arteriolosclerosis may underlie some lobar lesions; CAA and B-ASC often coexist with AD and other age-related pathologies.
Validation ladder
Seven questions for any biomarker
Do not jump from reproducibility or association to clinical use.
Technical validity
Does it repeat across raters, scans, sites, and software?
Biological validity
Does it correspond to the intended pathology or state?
Diagnostic accuracy
How sensitive, specific, and calibrated is it against an independent reference?
Value and utility
Does it prognosticate, add information, transport, and improve decisions?