SplineSplat: 3D Ray Tracing for Higher-Quality Tomography

TL;DR

This paper introduces SplineSplat, a novel 3D ray-tracing method utilizing B-splines and neural networks to improve tomographic reconstruction quality over traditional voxel-based approaches.

Contribution

It presents a new efficient ray-tracing algorithm for 3D volumes represented by B-splines, incorporating neural networks for basis function contribution computation.

Findings

Achieves higher reconstruction quality than traditional voxel-based methods

Handles arbitrary projection geometries effectively

Operates efficiently in well-posed cases without regularization

Abstract

We propose a method to efficiently compute tomographic projections of a 3D volume represented by a linear combination of shifted B-splines. To do so, we propose a ray-tracing algorithm that computes 3D line integrals with arbitrary projection geometries. One of the components of our algorithm is a neural network that computes the contribution of the basis functions efficiently. In our experiments, we consider well-posed cases where the data are sufficient for accurate reconstruction without the need for regularization. We achieve higher reconstruction quality than traditional voxel-based methods.