High-Quality Virtual Single-Viewpoint Surgical Video: Geometric Autocalibration of Multiple Cameras in Surgical Lights

2025-03-05

International Conference on Medical Image Computing and Computer-Assisted Intervention

AI Summary

Surgical lighting systems with multi-camera configurations suffer from dynamic camera pose changes due to lamp movement, necessitating manual image alignment and hindering real-time intraoperative applications. Method: This paper proposes a fully automatic, geometry-constrained joint calibration method for intrinsic and extrinsic parameters, integrating nonlinear optimization, multi-view geometric modeling, and joint modeling of lens distortion and optical center constraints. Contribution/Results: To the best of our knowledge, this is the first markerless, real-time geometric self-calibration framework for integrated surgical lamp multi-camera systems. It enables dynamic optical path compensation and viewpoint-consistent 3D reconstruction. Evaluated in real surgical environments, the method achieves sub-pixel reprojection accuracy (<0.35 px) and generates 4K single-viewpoint virtual video with end-to-end latency under 12 ms. Clinical assessment demonstrates significantly improved intraoperative navigation stability.

Technology Category

• Computer Vision: Computational Photography, Image & Video Synthesis
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Machine Learning: Calibration & Uncertainty Quantification

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Experiences and lessons learnt from Web-based algorithms and system deployments
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Problem

• Automates alignment of multiple surgical cameras
• Reduces occlusion in surgical video generation
• Improves video quality with geometric autocalibration

Innovation

• Automated alignment of multiple surgical cameras
• Dynamic selection of least occluded camera
• Stabilized video generation with reduced occlusion