TR-GS: High-Fidelity Sparse-View CT Volumetric Rendering via t-Distribution Gaussian Splatting and Ray-Confidence Modeling

2026-08-16

AI Summary

This study addresses structural artifacts and uncertainty arising from insufficient observations in sparse-view CT reconstruction by proposing the TR-GS framework. This method innovatively introduces projectable t-distribution primitives to replace standard Gaussians, integrating an adaptive ray confidence adjustment mechanism with confidence-guided wavelet regularization to achieve high-fidelity volume rendering while effectively suppressing artifacts. Experimental results demonstrate that TR-GS significantly outperforms mainstream baselines on both synthetic and real-world datasets, successfully preserving fine anatomical details. Consequently, this framework provides reliable technical support for downstream medical applications such as XR visualization, offering a robust solution to the challenges inherent in limited-angle tomographic imaging.

Abstract

High-fidelity 3D medical visualization supports applications such as clinical assessment and surgical planning. Sparse-view computed tomography (CT) can reduce projection requirements and associated radiation exposure, but limited observations may introduce structural artifacts and reconstruction uncertainty. Although 3D Gaussian Splatting (3DGS) provides an efficient explicit representation for volumetric rendering, existing CT methods based on standard Gaussian primitives may be sensitive to unreliable observations under sparse-view acquisition. We present TR-GS, a Gaussian-splatting framework for sparse view CT volumetric rendering. TR-GS replaces standard Gaussian primitives with projectable Student's t-distribution primitives and introduces a ray-confidence model that regulates their degrees of freedom according to local ray observability. Confidence-guided 3D wavelet regularization is further used to balance high-frequency detail preservation and noise suppression. This work is licensed under a Creative Commons Attribution 4.0 International License. Experiments on synthetic and real-world datasets show that TR-GS improves over representative baselines in most evaluated settings and remains competitive in the remaining cases. The resulting volumetric representations may support downstream medical multimedia applications, including XR-based visualization and interactive clinical rendering.

Problem

• Sparse-view CT
• Volumetric Rendering
• 3D Gaussian Splatting
• Reconstruction Uncertainty
• Structural Artifacts

Innovation

• Student's t-distribution primitives
• Ray-confidence modeling
• Sparse-view CT
• 3D Gaussian Splatting
• Wavelet regularization