Open-source AI aligns live X-ray with 3D CT at submillimeter accuracy, Nature paper
Dr_Alex_Crimi · x · 2026-09-29
A team including MIT researchers published an open-source framework, xvr, in Nature: it rapidly and precisely registers 2D intraoperative X-ray fluoroscopy to preoperative 3D CT/MRI volumes at submillimeter accuracy — the missing transform surgeons never had.
- Problem: intensity-based optimizers need per-patient hyperparameter tuning; prior deep-learning methods need large labeled datasets and only work on the anatomy they were trained on.
- Approach: xvr is a self-supervised framework combining patient-specific neural networks with gradient-based optimization, generalizing across fluoroscopy-guided procedures without manual labels.
Paper and code are both public.
Related event: Open-Source AI Achieves Sub-millimeter X-ray to 3D CT Registration(2 posts)→
More from Research
- Alignment researcher points to 'self-undermining unilateral optimization' classics — edelwax · 2026-09-29
- World's first live AI-assisted brain surgery removes tumour using real-time DINOv3 inference — TimDarcet · 2026-09-29
- Open-source project trains an LLM from scratch in PyTorch, 13M params on free Colab — thisguyknowsai · 2026-09-29
- 8 parallel AI societies run for weeks: agents evade isolation, invent uninterpretable language flagged as suicidal ideation — Slight-Box-2890 · 2026-09-29
- Quantum computer simulates matter 'popping into existence' from the vacuum — Sauerkrautkid7 · 2026-09-29
- Adobe's FlowTool frames image retouching as flow matching, cutting latency 50x — adobe-research · 2026-09-29