Compton cameras for medical applications (COMMA): Image reconstruction and corrections

这一后续建议的重点是图像重建和补偿图像退化现象的康普顿相机（CC）。它建立在一个致力于基于cc的快速伽马成像（PGI）的项目上，用于粒子治疗的范围验证。在这个项目中，我们开发了先进的重建算法，并提出了新的方法来处理基于cc的PGI和截断伪影的低计数水平。在未来的项目中，除了PGI之外，还将探索将CC用于放射性核素治疗的规划和控制的可能性。首先，我们将改进和深入研究我们现有的方法，包括基于hodoscope提供的信息的具有专用先验的贝叶斯重建（也称为惩罚或正则化函数）。这些算法旨在减少噪声贡献以及图像退化，并允许对距离偏移进行鲁棒和准确的识别。建议的惩罚函数将集成到正则化（贝叶斯）重建方案中。此外，我们将通过先验将正则化应用于Origin Ensembles （OE）算法，这是一种尚未探索的有前途的方法。在此基础上，进一步解决了PG辐射未知能量引起的降解问题。稍后，将不再假设hodoscope数据的可用性，并从图像中提取所需的信息。为此，将考虑备选优化程序、图像处理和机器学习技术。将特别关注降解现象及其补偿，因为这个主题迄今为止几乎没有解决，尽管对某些现象的补偿是临床应用所必需的。为了在粒子治疗中的实际应用，需要考虑高PG发射率。因此，将开发基于GATE和定制程序的专用仿真管道。目的是重现治疗光束的时间结构，PG率的调制，以及随后的偶然巧合检测。这些信息是设计随机贡献的估计和补偿方法所必需的，这是CC成像尚未考虑的问题。为此目的，将开发能够估计随机贡献的模型。为了范围验证和治疗，还将分析由患者散射和衰减引起的退化。对这些现象的补偿方法将在重建框架内发展和整合。设想的重建算法并不局限于特定的康普顿相机，应该很容易适应任何配置。同样，我们的方法也可以用于其他应用和模式，例如多同位素成像，甚至用于束内PET。

This follow-up proposal focuses on image reconstruction and compensation of image degradation phenomena for Compton cameras (CC). It builds on a project devoted to CC-based prompt-gamma imaging (PGI) for range verification in particle therapy. Within this project, we have developed advanced reconstruction algorithms and proposed novel approaches to cope with the very low-count levels typical of CC-based PGI and truncation artefacts. In the future project, in addition to PGI, the possible use of CC for planing and control of radionuclide-based therapy will be also explored. Firstly, we will improve and investigate in depth our current approaches, which include Bayesian reconstruction with dedicated priors (also called penalty or regularization functions) based on the information provided by a hodoscope. These algorithms aim to reduce the noise contribution as well as image degradation and allow for a robust and accurate identification of range shifts. The suggested penalty functions will be integrated into regularized (Bayesian) reconstruction schemes. Additionally, we will apply regularization via priors to the Origin Ensembles (OE) algorithm, a promising approach not explored yet. The degradation caused by the unknown energy of the PG radiation will be further tackled. Later, the availability of hodoscope data will not be longer assumed and the required information will be extracted from the images. To this end, alternative optimization procedures, image processing and machine learning techniques will be considered. Special attention will be put on degradation phenomena and their compensation, as this topic has been hardly addressed so far, although compensation of certain phenomena is mandatory towards clinical application. For a realistic application in particle therapy, the high PG emission rates need to be considered. Therefore, a dedicated simulation pipeline based on GATE and custom-made programmes will be developed. The objective is to reproduce the time structure of the therapeutic beams, the modulation of the PG rates, and the subsequent detection of accidental coincidences. This information is required to design estimation and compensation methods for the random contribution, an issue which has not been contemplated for CC imaging yet. For this purpose, models able to estimate the random contribution will be developed. The degradation caused by patient scatter and attenuation will be also analyzed, for both range verification and theranostics. Compensation methods for these phenomena will be developed and integrated within the reconstruction framework. The envisioned reconstruction algorithms are not restricted to a particular Compton camera, and should be easily adaptable to any configuration. In the same vein, our approaches could be also of use for other applications and modalities, such as multi-isotope imaging or even for in-beam PET.