AI Boosts Handheld Ultrasound Detection of Carotid Plaques to 94.8%
A new AI model improves handheld ultrasound clarity, catching nearly 15% more carotid plaques than standard devices in community screening tests.
Key points
- AI enhancement increased carotid plaque detection in handheld ultrasounds from 87.6% to 94.8% in a study of 117 adults.
- The system improved the identification of unstable plaques to 63.2% but still missed a significant portion of high-risk cases.
- Researchers recommend using the tool to triage patients for further imaging rather than as a standalone diagnostic method.
Handheld ultrasound devices are becoming common in primary care, but their lower image quality often renders small or faint carotid plaques invisible. To address this gap, researchers in Hunan Province, China, developed an artificial intelligence model that sharpens these images after they have been captured. The goal was to make portable devices more reliable for detecting stroke risk factors outside of specialized hospitals.
The team tested the system in a community screening program involving adults aged 40 and older. By applying the AI enhancement to existing handheld ultrasound data, they aimed to see if the technology could identify risks that standard equipment typically misses. This approach could help clinicians decide who needs further, more intensive imaging or closer follow-up.
Detection rates rise significantly
In a study of 117 participants, the AI-enhanced images identified 94.8% of the 153 carotid plaques present. In contrast, standard handheld ultrasound images only detected 87.6% of them. The additional 11 plaques found through enhancement were primarily small or low-contrast lesions with mild vessel narrowing, which are often difficult to spot without digital aid.
Improved risk assessment accuracy
The technology also performed better in distinguishing plaque stability. The enhanced images correctly flagged 63.2% of plaques that appeared unstable, a notable increase from the 47.4% accuracy of the standard images. Additionally, the system correctly ruled out stable-appearing plaques about 96% of the time, helping to reduce unnecessary anxiety or over-treatment for patients with benign findings.
Limitations remain for unstable cases
Despite these improvements, the tool is not a perfect diagnostic solution. The enhanced images still missed more than one-third of plaques with unstable-appearing features. As reported by news-medical.net, this suggests that the AI should not be used in isolation to rule out high-risk conditions. Clinicians must still rely on confirmatory imaging and clinical judgment to manage stroke risk effectively.
