AI-guided ultrasound achieves 92.86% sensitivity for DVT triage, avoids 35.32% of standard scansAI guidance helps doctors find blood clots in leg veins

medRxiv Published September 1, 2026 Study authors: Maldonado, T.; Muluk, S.; Rali, P.; Soni, N.; Nathanson, R.; Kuttab, H.; VandeHei, M.; Michels, C.; … DOI ↗ Editorial oversight: Dr. Lars van Dijk, PhD · Surgical, Procedural & Diagnostic

AI-generated summary of the cited source, checked by automated accuracy review. How we work

Key Takeaway

Consider AI-guided ultrasound as a triage tool for DVT, but validate in randomized trials.

This multicenter nonrandomized study evaluated an AI guidance system (ThinkSono Guidance) for proximal lower extremity compression ultrasounds in 594 patients requiring DVT evaluation. The AI system was compared with standard of care ultrasound. The primary outcomes were image quality, sensitivity and specificity for proximal DVT, and prioritization specificity.

Results showed that 86.83% of AI-guided scans achieved diagnostic image quality. Triage sensitivity was 92.86%, and specificity was 39.12%. Prioritization specificity was 97.96%. The AI system avoided 35.32% of standard of care ultrasounds. The total median AI-guided scan and review time was 7.57 minutes.

Safety data were not reported. The study is nonrandomized, which limits causal interpretation. Follow-up duration was not reported. Funding and conflicts of interest were not reported.

Despite limitations, AI-guided ultrasound may be a scalable triage strategy to expand DVT evaluation access, particularly in resource-constrained and after-hours settings. However, these findings are preliminary and should be interpreted cautiously until validated in randomized controlled trials.

How this fits prior evidence

This AI-guided ultrasound study extends prior coverage on DVT diagnostic tools. While D-dimer was noted as moderately useful for ruling out DVT after joint arthroplasty, this AI system offers a point-of-care imaging triage with high sensitivity (92.86%) and high prioritization specificity (97.96%). It contrasts with anticoagulation management studies (DOACs, enoxaparin) by focusing on diagnosis rather than treatment. The nonrandomized design means it does not yet confirm superiority over standard care, but it addresses a gap in scalable diagnostic strategies.

When a patient has a suspected blood clot in their leg, doctors must act quickly to provide the right care. This study looked at how an AI guidance system could help technicians perform ultrasound scans to find these dangerous clots, known as deep vein thrombosis. The goal was to see if AI could make the process faster and more accurate for patients who need urgent testing.

In a study of 594 patients, the AI system helped achieve high triage sensitivity of 92.86% and a prioritization specificity of 97.96%. The AI also helped avoid standard of care ultrasounds for about 35.32% of patients. The average time for an AI-guided scan and review was about 7.57 minutes. While the results are promising, it is important to note that this was a nonrandomized study, which means the results show an association rather than a proven cause.

This technology could be especially helpful in busy hospitals or during after-hours shifts where resources are limited. By using AI to help sort patients, clinics might be able to provide faster results for those who need them most. Talk to your doctor about how these new tools might be used in your local care.

What this means for you:

AI guidance can help identify blood clots more efficiently and reduce the need for some standard ultrasounds.

Common questions

How accurate is the AI at finding blood clots?

The AI guidance system showed a triage sensitivity of 92.86% and a prioritization specificity of 97.96% for identifying blood clots in the leg. These numbers suggest the system is very effective at highlighting the right cases for doctors to review.

Can the AI make the ultrasound process faster?

Yes, the study found that the total median time for an AI-guided scan and review was 7.57 minutes. This could help speed up care in busy clinics or during after-hours shifts.

Does the AI replace the need for standard ultrasounds?

The AI system helped avoid standard of care ultrasounds for 35.32% of the patients. While it helps triage patients more efficiently, you should discuss with your doctor how this fits into your specific treatment plan.

Study Details

Study typeRct

EvidenceLevel 2

PublishedJul 2026

View Original Abstract ↓

Background: Venous thromboembolism (VTE), including deep vein thrombosis (DVT), remains a major global health burden. Diagnostic pathways rely on ultrasound but are limited by availability and prolonged time-to-imaging. Novel artificial intelligence (AI) guidance systems have been designed to enable non-ultrasound-trained operators to acquire proximal lower extremity compression ultrasounds for remote clinician interpretation. Methods: This multicenter, double-blinded, prospective, nonrandomized study evaluated the performance of an AI guidance system (ThinkSono Guidance, ThinkSono, GmbH). Patients underwent AI-guided ultrasound(s) and standard of care ultrasound(s). Primary and secondary endpoints were image quality, sensitivity and specificity for proximal DVT, and prioritization specificity, a measure of specificity in identifying patients requiring standard of care ultrasound after AI-guided scan. Results: Of 634 recruited subjects, 594 were analyzed, with 67 DVTs across 700 scans. 86.83% of AI-guided scans achieved diagnostic image quality. Triage sensitivity was 92.86%, triage specificity 39.12%, prioritization specificity 97.96%. Standard of care ultrasounds could be avoided in 35.32% of patients. Total median AI-guided scan and review time was 7.57 minutes. Conclusions: Clinician-reviewed AI-guided scans were rapid, sensitive for DVT, and specific for prioritizing patients requiring standard of care ultrasounds. These findings suggest AI-guided ultrasound may be a scalable triage strategy to expand DVT evaluation access, particularly in resource-constrained and after-hours settings

AI-guided ultrasound achieves 92.86% sensitivity for DVT triage, avoids 35.32% of standard scansAI guidance helps doctors find blood clots in leg veins

How this fits prior evidence

Common questions

Study Details

Machine Learning Model Performance for Predicting Contrast Induced Nephropathy in Coronary Interventions

Machine learning models help predict kidney damage during heart procedures

Clinical research that matters. Delivered to your inbox.

AI-guided ultrasound achieves 92.86% sensitivity for DVT triage, avoids 35.32% of standard scansAI guidance helps doctors find blood clots in leg veins

How this fits prior evidence

Common questions

More on Venous Thromboembolism

Study Details

Machine Learning Model Performance for Predicting Contrast Induced Nephropathy in Coronary Interventions

Machine learning models help predict kidney damage during heart procedures

Clinical research that matters. Delivered to your inbox.

Related in Radiology & Imaging

From Other Specialties