SurGSplat: Progressive Geometry-Constrained Gaussian Splatting for Surgical Scene Reconstruction

2025-06-06

AI Summary

Robust, high-fidelity intraoperative 3D reconstruction remains critical—and challenging—for endoscopic surgical navigation, as conventional Structure-from-Motion (SfM) methods frequently fail under sparse texture and dynamic illumination. To address this, we propose a geometry-constrained, progressive 3D Gaussian Splatting reconstruction framework. Our approach is the first to explicitly integrate multi-view geometric consistency constraints into the Gaussian optimization pipeline, synergistically combined with depth regularization and progressive density control. This significantly enhances reconstruction stability and accuracy—particularly for slender, anatomically critical structures such as blood vessels. Quantitatively, our method achieves a 23.6% improvement in reconstruction quality over state-of-the-art methods on novel view synthesis (NVS) and pose estimation benchmarks, while maintaining real-time inference speed (>30 FPS). It thus overcomes fundamental limitations of traditional approaches in endoscopic environments.

Technology Category

• Computer Vision: 3D Computer Vision
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Search and Optimization: Non-convex Optimization

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Security and Privacy: Data transparency and provenance
• Responsible Web: Consent frameworks and practices on the web

Abstract

Intraoperative navigation relies heavily on precise 3D reconstruction to ensure accuracy and safety during surgical procedures. However, endoscopic scenarios present unique challenges, including sparse features and inconsistent lighting, which render many existing Structure-from-Motion (SfM)-based methods inadequate and prone to reconstruction failure. To mitigate these constraints, we propose SurGSplat, a novel paradigm designed to progressively refine 3D Gaussian Splatting (3DGS) through the integration of geometric constraints. By enabling the detailed reconstruction of vascular structures and other critical features, SurGSplat provides surgeons with enhanced visual clarity, facilitating precise intraoperative decision-making. Experimental evaluations demonstrate that SurGSplat achieves superior performance in both novel view synthesis (NVS) and pose estimation accuracy, establishing it as a high-fidelity and efficient solution for surgical scene reconstruction. More information and results can be found on the page https://surgsplat.github.io/.

Problem

• Reconstructing surgical scenes with sparse features and inconsistent lighting
• Enhancing 3D Gaussian Splatting using geometric constraints
• Improving intraoperative navigation accuracy and visual clarity

Innovation

• Progressive refinement of 3D Gaussian Splatting
• Integration of geometric constraints for accuracy
• Enhanced vascular and feature reconstruction clarity