Discretized Gaussian Representation for Tomographic Reconstruction

2024-11-07

AI Summary

Existing deep learning-based CT reconstruction methods suffer from heavy reliance on large-scale annotated datasets and high computational costs, while NeRF and 3D Gaussian Splatting (3DGS) struggle to directly model 3D volumetric structures. To address these limitations, we propose Discrete Gaussian Representation (DGR): a differentiable, end-to-end 3D volume reconstruction framework built upon learnable discrete Gaussian functions as voxel primitives. Our method integrates physics-driven forward projection modeling, fast parallel voxel accumulation (Fast Volume Reconstruction), and parametric Gaussian rendering—jointly reducing data and compute dependencies. Evaluated on both synthetic and real CT datasets, DGR achieves PSNR gains of 2.1–3.8 dB over state-of-the-art deep learning reconstruction (DLR) and NeRF/3DGS approaches, accelerates inference by 3.2×, and reduces GPU memory consumption by 45%. To our knowledge, DGR is the first method enabling efficient, accurate, and fully differentiable direct 3D volumetric reconstruction.

Technology Category

• Computer Vision: 3D Computer Vision
• Machine Learning: Deep Generative Models & Autoencoders
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

Computed Tomography (CT) is a widely used imaging technique that provides detailed cross-sectional views of objects. Over the past decade, Deep Learning-based Reconstruction (DLR) methods have led efforts to enhance image quality and reduce noise, yet they often require large amounts of data and are computationally intensive. Inspired by recent advancements in scene reconstruction, some approaches have adapted NeRF and 3D Gaussian Splatting (3DGS) techniques for CT reconstruction. However, these methods are not ideal for direct 3D volume reconstruction. In this paper, we propose a novel Discretized Gaussian Representation (DGR) for CT reconstruction, which directly reconstructs the 3D volume using a set of discretized Gaussian functions in an end-to-end manner. To further enhance computational efficiency, we introduce a Fast Volume Reconstruction technique that aggregates the contributions of these Gaussians into a discretized volume in a highly parallelized fashion. Our extensive experiments on both real-world and synthetic datasets demonstrate that DGR achieves superior reconstruction quality and significantly improved computational efficiency compared to existing DLR and instance reconstruction methods. Our code has been provided for review purposes and will be made publicly available upon publication.

Problem

• Enhance CT image quality and reduce noise efficiently
• Direct 3D volume reconstruction using discretized Gaussians
• Improve computational efficiency in tomographic reconstruction

Innovation

• Discretized Gaussian Representation for 3D volume
• Fast parallelized volume reconstruction technique
• End-to-end CT reconstruction with Gaussians