Computational methods for proton imaging: FRM funded project

Computational methods to optimize proton radiography and tomography for improved proton therapy.

Research project funded by the Fondation pour la Recherche Médicale

Objective

Range uncertainties currently represent the biggest caveat for the exploitation of the full potential of proton therapy. To better predict the proton range in tissues and produce more conformal treatment plans, we propose to directly derive the proton range from proton computed tomography (pCT). To address the issue of poor spatial resolution of pCT, we propose optimization algorithms to correctly model the multiple Coulomb scattering, even in heterogeneous materials, and we will further investigate methods to merge proton and X-ray tomographic data. We will solve the inverse reconstruction problem for both of the two currently worldwide developed scanners: single proton detectors (list-mode) and integrating detectors (integral mode). With the eventual aim to bring pCT into clinical practice, we will define a new treatment work-flow in which range information in the patient will be obtained from the optimized pCT and not solely from the conversion of the planning X-ray CT.

People involved in the project

  • Jean Michel Létang, lecturer-researcher (enseignant-chercheur)

  • Simon Rit, researcher (chercheur CNRS)

  • Nils Krah, post-doc researcher

  • Feriel Khellaf, PhD student 

  • Ilaria Rinaldi, researcher from partner institution

Background

Proton therapy has rapidly grown in the past thirty years and it has become a superior alternative to conventional radiotherapy for certain clinical indications. In France, there are two clinical centers, in Orsay and Nice, and several projects to build new centers.

Proton therapy offers high dose selectivity due to the protons' distinct depth dose profile. However, appropriate management of treatment uncertainties is required to fully exploit this advantage. Precise knowledge of the relative stopping power (RSP) of the patient tissues is needed to correctly predict the proton range in the treatment planning. Currently, this prediction is approximated from X-ray computed tomography (CT) and the associated uncertainties require additional safety margins. Proton radiography (pR) and/or CT (pCT) [1,2] could improve or even bypass this approximation by directly measuring the RSP. Additionally, it could be used to verify and monitor the positioning of the patient prior to or in-between the treatment, potentially in presence of motion.  

Proton imaging suffers from limited spatial resolution, e.g., with respect to X-ray imaging, due to multiple Coulomb scattering (MCS). Most investigations concentrate on hardware improvements to cope with this limitation. We propose to develop complementary algorithmic solutions which will significantly improve the image quality. We have recently shown that MCS has an edge-enhancing effect in list-mode proton imaging which could be used to improve spatial resolution [3]. Similar effects have been investigated with integrating detectors [4].

So far, statistical descriptions of MCS only exist for the simplified case of homogeneous media. The purpose of this project is to develop a computationally efficient model of MCS inheterogeneous media and to integrate it into a typical tomographic reconstruction workflow. Both list-mode and integrated scenarios will be investigated. We will further investigate methods to merge proton and X-ray tomographic data and define a new treatment work-flow in which range information in the patient will be obtained from the optimized pCT and not solely from the conversion of the planning X-ray CT. 

----
[1] Rinaldi et al 2014b. http://doi.org/10.1088/0031-9155/59/12/3041
[2] Arbor et al 2015. http://doi.org/10.1088/0031-9155/60/19/7585
[3] Quinones et al. 2016. http://doi.org/10.1088/0031-9155/61/9/3258
[4] Krah et al 2015. http://doi.org/10.1088/0031-9155/60/21/8525
 

Project details

Duration: 3 Years
Start: Septembre 2017
Budget: 287 000 euros for one PhD student (Feriel Khellaf) and a post-doc researcher (to be recruted), plus computing equipment