New AI technique could make minimally invasive surgeries safer and more precise (news.mit.edu)

Researchers at MIT have developed a groundbreaking AI technique that significantly enhances the safety and precision of minimally invasive surgeries by quickly and accurately aligning real-time X-ray images with preoperative 3D scans of patients. This method, known as xvr (X-ray volume registration), adapts to individual patient anatomy in just five minutes, providing sub-millimeter precision that enables clinicians to steer surgical instruments like catheters and endoscopes with greater confidence and clarity. The approach addresses the long-standing challenge in surgery where clinicians struggle to interpret flat X-ray images, which can lead to complications. What sets xvr apart from existing AI tools is its ability to generate patient-specific synthetic X-rays using physics-based simulations from individual 3D scans. This advanced model significantly reduces the time taken for registration, performs reliably across a diverse patient cohort, and can seamlessly integrate into emergency surgical scenarios. The promise of this technology lies not only in its ability to improve surgical outcomes and accessibility—especially critical in urgent care settings—but also in its potential application in robotic surgery and more complex medical processes, with ongoing collaborations expected to refine and deploy these innovations effectively.