Scholarlink

Background <p>Deep vein thrombosis (DVT) is a&#xa0;common and potentially life-threatening condition that requires prompt diagnosis to prevent thrombus progression and pulmonary embolism (PE). Compression ultrasound is the diagnostic standard, but its accessibility is limited by operator proficiency and workforce availability. Artificial intelligence (AI)-guided ultrasound systems have emerged as a&#xa0;promising solution for enhancing accessibility and standardization in DVT diagnostics.</p> Methods <p>This review synthesizes the current evidence on AI-guided compression ultrasound for DVT, with particular emphasis on the ThinkSono guidance system. We integrated epidemiological data, clinical validation studies, ongoing trials, and emerging applications, taken from published literature and registered studies.</p> Results <p>ThinkSono features real-time AI-guided image acquisition, performs automatic quality checks, and includes reviews by qualified clinicians. The machine learning approach relies on deep learning models that are trained using ultrasound cine-loops to recognize anatomical landmarks and evaluate venous compressibility. Clinical validation demonstrated that AI-guided combined with remote clinician evaluation can achieve a&#xa0;diagnostic performance close to that of traditional ultrasound. Preliminary health-economic analyses suggest possible cost savings, given the significant financial impact of venous thromboembolism.</p> Conclusion <p>The use of AI-guided compression ultrasound represents a&#xa0;promising addition to current DVT diagnostic pathways. By incorporating a&#xa0;human-in-the-loop approach with established escalation protocols, the accessibility and efficiency of vascular ultrasound can be enhanced while maintaining diagnostic safety.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

ThinkSono artificial intelligence (AI)-guided compression ultrasound in deep vein thrombosis

  • 期刊论文

  • 发表日期：

    2026年5月7日

查看全文
  • Aristotelis G. Yfantis,
  • Efthymios D. Avgerinos

摘要

Background

Deep vein thrombosis (DVT) is a common and potentially life-threatening condition that requires prompt diagnosis to prevent thrombus progression and pulmonary embolism (PE). Compression ultrasound is the diagnostic standard, but its accessibility is limited by operator proficiency and workforce availability. Artificial intelligence (AI)-guided ultrasound systems have emerged as a promising solution for enhancing accessibility and standardization in DVT diagnostics.

Methods

This review synthesizes the current evidence on AI-guided compression ultrasound for DVT, with particular emphasis on the ThinkSono guidance system. We integrated epidemiological data, clinical validation studies, ongoing trials, and emerging applications, taken from published literature and registered studies.

Results

ThinkSono features real-time AI-guided image acquisition, performs automatic quality checks, and includes reviews by qualified clinicians. The machine learning approach relies on deep learning models that are trained using ultrasound cine-loops to recognize anatomical landmarks and evaluate venous compressibility. Clinical validation demonstrated that AI-guided combined with remote clinician evaluation can achieve a diagnostic performance close to that of traditional ultrasound. Preliminary health-economic analyses suggest possible cost savings, given the significant financial impact of venous thromboembolism.

Conclusion

The use of AI-guided compression ultrasound represents a promising addition to current DVT diagnostic pathways. By incorporating a human-in-the-loop approach with established escalation protocols, the accessibility and efficiency of vascular ultrasound can be enhanced while maintaining diagnostic safety.