X$^{2}$-Gaussian: 4D Radiative Gaussian Splatting for Continuous-time Tomographic Reconstruction

2025-03-27

AI Summary

Existing 4D-CT reconstruction relies on discrete respiratory phase binning, causing motion misalignment and requiring specialized gating hardware. To address this, we propose a hardware-free, continuous-time 4D-CT reconstruction framework. Our method introduces dynamic radial Gaussian splatting into tomographic reconstruction—enabling differentiable, spatiotemporally continuous deformation modeling of anatomical structures. We further design a physiology-driven projection-domain periodic consistency loss to enable self-supervised learning of patient-specific respiratory dynamics. A spatiotemporal encoder-decoder network jointly optimizes the reconstructed image sequence and the underlying motion field. Experiments demonstrate a 9.93 dB PSNR improvement over conventional binning-based methods and a 2.25 dB gain over prior Gaussian splatting approaches. The framework significantly enhances motion fidelity and clinical practicality while eliminating hardware dependencies.

Technology Category

• Computer Vision: Motion & Tracking
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Four-dimensional computed tomography (4D CT) reconstruction is crucial for capturing dynamic anatomical changes but faces inherent limitations from conventional phase-binning workflows. Current methods discretize temporal resolution into fixed phases with respiratory gating devices, introducing motion misalignment and restricting clinical practicality. In this paper, We propose X$^2$-Gaussian, a novel framework that enables continuous-time 4D-CT reconstruction by integrating dynamic radiative Gaussian splatting with self-supervised respiratory motion learning. Our approach models anatomical dynamics through a spatiotemporal encoder-decoder architecture that predicts time-varying Gaussian deformations, eliminating phase discretization. To remove dependency on external gating devices, we introduce a physiology-driven periodic consistency loss that learns patient-specific breathing cycles directly from projections via differentiable optimization. Extensive experiments demonstrate state-of-the-art performance, achieving a 9.93 dB PSNR gain over traditional methods and 2.25 dB improvement against prior Gaussian splatting techniques. By unifying continuous motion modeling with hardware-free period learning, X$^2$-Gaussian advances high-fidelity 4D CT reconstruction for dynamic clinical imaging. Project website at: https://x2-gaussian.github.io/.

Problem

• Enables continuous-time 4D-CT reconstruction without phase discretization
• Eliminates dependency on external respiratory gating devices
• Improves reconstruction accuracy over traditional and prior methods

Innovation

• Dynamic radiative Gaussian splatting for 4D-CT
• Self-supervised respiratory motion learning
• Physiology-driven periodic consistency loss