Rescigno Regina, Bopp Cécile, Rousseau Marc, Brasse David
Université de Strasbourg, IPHC, 23 rue du Loess, Strasbourg 67037, France and CNRS, UMR7178, Strasbourg 67037, France.
Med Phys. 2015 Nov;42(11):6610-24. doi: 10.1118/1.4933422.
A new approach to proton computed tomography (pCT) is presented. In this approach, protons are not tracked one-by-one but a beam of particles is considered instead. The elements of the pCT reconstruction problem (residual energy and path) are redefined on the basis of this new approach. An analytical image reconstruction algorithm applicable to this scenario is also proposed.
The pencil beam (PB) and its propagation in matter were modeled by making use of the generalization of the Fermi-Eyges theory to account for multiple Coulomb scattering (MCS). This model was integrated into the pCT reconstruction problem, allowing the definition of the mean beam path concept similar to the most likely path (MLP) used in the single-particle approach. A numerical validation of the model was performed. The algorithm of filtered backprojection along MLPs was adapted to the beam-by-beam approach. The acquisition of a perfect proton scan was simulated and the data were used to reconstruct images of the relative stopping power of the phantom with the single-proton and beam-by-beam approaches. The resulting images were compared in a qualitative way.
The parameters of the modeled PB (mean and spread) were compared to Monte Carlo results in order to validate the model. For a water target, good agreement was found for the mean value of the distributions. As far as the spread is concerned, depth-dependent discrepancies as large as 2%-3% were found. For a heterogeneous phantom, discrepancies in the distribution spread ranged from 6% to 8%. The image reconstructed with the beam-by-beam approach showed a high level of noise compared to the one reconstructed with the classical approach.
The PB approach to proton imaging may allow technical challenges imposed by the current proton-by-proton method to be overcome. In this framework, an analytical algorithm is proposed. Further work will involve a detailed study of the performances and limitations of this approach in terms of image quality. The paper shows how to account for the MCS in the reconstruction step with this new approach when an analytical reconstruction algorithm is used.
提出一种质子计算机断层扫描（pCT）的新方法。在这种方法中，质子不是逐个追踪，而是考虑一束粒子。基于这种新方法重新定义了pCT重建问题的要素（剩余能量和路径）。还提出了一种适用于此场景的解析图像重建算法。
利用费米 - 艾格斯理论的推广来考虑多次库仑散射（MCS），对笔形束（PB）及其在物质中的传播进行建模。该模型被整合到pCT重建问题中，允许定义类似于单粒子方法中使用的最可能路径（MLP）的平均束路径概念。对该模型进行了数值验证。沿MLP的滤波反投影算法被应用于逐束方法。模拟了完美质子扫描的采集，并使用数据通过单质子和逐束方法重建体模相对阻止本领的图像。对所得图像进行了定性比较。
将建模的PB参数（均值和展宽）与蒙特卡罗结果进行比较以验证模型。对于水靶，分布均值方面有良好的一致性。就展宽而言，发现深度相关的差异高达2% - 3%。对于非均匀体模，分布展宽的差异范围为6%至8%。与用经典方法重建的图像相比，逐束方法重建的图像显示出高水平的噪声。
质子成像的PB方法可能允许克服当前逐个质子方法带来的技术挑战。在此框架内，提出了一种解析算法。进一步的工作将涉及详细研究这种方法在图像质量方面的性能和局限性。本文展示了在使用解析重建算法时，如何用这种新方法在重建步骤中考虑MCS。
