A proton simulator for testing implementations of proton CT reconstruction algorithms on GPGPU clusters

TL;DR

This paper introduces a proton CT simulator designed to generate realistic data sets for testing and improving reconstruction algorithms on GPGPU clusters, aiming to enhance clinical efficiency.

Contribution

It presents a novel simulator for realistic pCT data generation to evaluate and optimize reconstruction algorithms on GPGPU hardware.

Findings

Simulator produces realistic pCT data sets

Enables testing of string-averaging and block-iterative algorithms

Supports development of faster, clinically practical reconstruction methods

Abstract

Proton computed tomography (pCT) is an image modality that will improve treatment planning for patients receiving proton radiation therapy compared with the current treatment techniques, which are based on X-ray CT. Reconstruction of a pCT image requires solving a large and sparse system of linear equations, which should be accomplished within a few minutes in order to be clinically practical. Analyzing the efficiency of potentially clinical reconstruction implementations requires multiple quality pCT data sets. The purpose of this paper is to describe the simulator that was developed to generate realistic pCT data sets to be used in testing the efficiency of reconstruction algorithms, in particular string-averaging and block-iterative projection algorithms using sparse matrix formats on General Purpose Graphics Processing Units (GPGPU)s.