Four-dimensional tomographic reconstruction by time domain decomposition

TL;DR

This paper introduces a novel 4D tomographic reconstruction method that allows for dynamic samples by decomposing data in the temporal domain with basis functions and L1 regularization, enabling practical GPU implementation.

Contribution

It presents a new reconstruction approach that relaxes the static sample requirement in tomography by incorporating temporal decomposition and regularization techniques.

Findings

Successful implementation on GPU systems

Effective decomposition of dynamic data sets

Relaxation of static sample constraint

Abstract

Since the beginnings of tomography, the requirement that the sample does not change during the acquisition of one tomographic rotation is unchanged. We derived and successfully implemented a tomographic reconstruction method which relaxes this decades-old requirement of static samples. In the presented method, dynamic tomographic data sets are decomposed in the temporal domain using basis functions and deploying an L1 regularization technique where the penalty factor is taken for spatial and temporal derivatives. We implemented the iterative algorithm for solving the regularization problem on modern GPU systems to demonstrate its practical use.