A Real-time Image Reconstruction System for Particle Treatment Planning Using Proton Computed Tomography (pCT)

TL;DR

This paper presents a real-time image reconstruction system for proton computed tomography (pCT), enabling rapid, accurate proton therapy planning through advanced distributed computing techniques.

Contribution

It introduces a fast, GPU-accelerated image reconstruction software for pCT, improving speed and accuracy over traditional methods.

Findings

Reconstruction software achieves near real-time performance.

System provides RSP images within minutes.

Validated with phantom data from prototype scanner.

Abstract

Proton computed tomography (pCT) is a novel medical imaging modality for mapping the distribution of proton relative stopping power (RSP) in medical objects of interest. Compared to conventional X-ray computed tomography, where range uncertainty margins are around 3.5%, pCT has the potential to provide more accurate measurements to within 1%. This improved efficiency will be beneficial to proton-therapy planning and pre-treatment verification. A prototype pCT imaging device has recently been developed capable of rapidly acquiring low-dose proton radiographs of head-sized objects. We have also developed an advanced, fast image reconstruction software based on distributed computing that utilizes parallel processors and graphical processing units. The combination of fast data acquisition and fast image reconstruction will enable the availability of RSP images within minutes for use in clinical settings. The performance of our image reconstruction software has been evaluated using data collected by the prototype pCT scanner from several phantoms.