- Design
- Diagnostic development and community validation study
- Population
- 117 community screening participants with 153 carotid plaques
- Primary outcome
- Plaque detection and stenosis grading agreement vs portable ultrasound
- Effect
- Detection 94.8% vs 87.6%; weighted κ 0.836 vs 0.492
This study, published in Annals of Family Medicine, developed an AI model to enhance images from handheld ultrasound devices. After validation in 127 hospital patients, it was tested in community screening of 117 participants with 153 plaques, using a standard portable ultrasound as reference.
AI-enhanced handheld imaging detected 94.8% of carotid plaques against 87.6% for standard handheld imaging. Agreement on stenosis grade with the reference improved from a weighted κ of 0.49 to 0.84, and sensitivity for vulnerable-looking plaques rose from 47.4% to 63.2%, with specificity staying high.
The technical improvement is real. Whether carotid screening in people without symptoms should be done at all is a separate question: major bodies such as the USPSTF recommend against screening the general population for carotid stenosis. The more realistic use is better imaging when a primary care clinician already has a reason to look.
- Do not screen asymptomatic adults for carotid stenosis on the strength of this study.
- Where carotid imaging is clinically indicated and formal ultrasound is distant, AI-enhanced handheld devices may help triage.
- Manage cardiovascular risk with risk scores, statins and blood pressure control regardless of plaque imaging.
- Refer patients with TIA or stroke symptoms urgently; handheld imaging must not delay that.
Why it matters
It shows point-of-care imaging in primary care is improving fast, which makes the question of when to use it more pressing.
Don't overread it
Better plaque detection does not show that screening improves outcomes, and asymptomatic carotid screening is not recommended.
The statistics, in plain English
A weighted κ measures agreement beyond chance: 0.49 is moderate, 0.84 is strong. The comparison used portable ultrasound, not formal duplex or angiography, as the reference, and the samples were small.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for family medicine, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free