Artificial intelligence in the diagnosis of acute deep vein thrombosis.

Authors

Abstract

Venous thromboembolism (VTE) is a common and potentially life-threatening condition associated with high morbidity rates. Diagnosing deep vein thrombosis (DVT) involves a complex clinical pathway that includes risk assessment tools like the Wells Score, D-dimer testing, and Duplex ultrasound, often resulting in extended waiting periods. Traditionally, compression ultrasound performed by skilled radiologists is the preferred imaging modality, with sensitivity and specificity rates above 90%. The integration of artificial intelligence (AI) into diagnostic imaging, particularly point-of-care ultrasound, has emerged as a promising approach to address these challenges. AI-guided ultrasound devices, such as the ThinkSono Guidance System (ThinkSono GmbH), are being studied and used to assist non-ultrasound-trained healthcare professionals in acquiring DVT ultrasound images, a compelling and rapidly evolving innovation. Based on several pilot clinical studies and a few thousand patients enrolled, when AI guided ultrasound is combined with expert review, it can achieve diagnostic accuracy comparable to gold-standard duplex ultrasound and can offer significant cost savings. There is potential to transform how medical imaging is being conducted in a busy tertiary hospital or even in primary care and diminish pressure on overloaded radiology departments. The opportunities can be endless such as in remote environments lacking radiologists, intensive care units, following-up endovenous procedures and a lot more. As research continues, efforts are now focusing on validation across heterogeneous populations to establish clear ethical and regulatory frameworks to support its safe and responsible use.

Tags

Topics