Researchers at MIT developed xvr, an AI technique that rapidly and precisely matches real-time X-rays with preoperative 3D medical scans to guide minimally invasive surgical tools. The system adapts to individual patients in five minutes and achieves sub-millimeter accuracy, potentially making complex procedures like stroke interventions more accessible and safer by reducing complications from imprecise tool positioning.
RADAR is a generalist vision-language model trained on over 400,000 contrast-enhanced abdominal CT examinations with 15 million image-text pairs, achieving expert-level performance for radiology diagnosis without manual annotation. The model is available with pre-trained checkpoints on HuggingFace and is released under the Apache License 2.0.