Exploiting photon energy information in Positron Emission Tomography

正电子发射断层扫描(PET)成像是管理癌症和许多其他疾病患者治疗途径的关键部分。然而，由于对患者的放射剂量的限制，PET图像可能会有噪声，并且图像分辨率相对较低。正如初步的科学论文所显示的那样，现代PET扫描仪具有比以前更详细地测量伽马光子能量信息的实验能力，这有可能提高信噪比。然而，在这项技术可以发展到商业系统之前，还需要进一步的工作。本项目旨在评估利用额外光子能量信息的各种方法对图像质量的影响，并优化最有希望的候选方法。要包括的方面包括使用能量窗调谐的散射抑制、光子散射估计、偶然巧合以及图像重建策略。我们设想将已建立的PET数据处理与使用机器学习的新技术相结合。该项目将扩展先前在PET和SPECT中使用光子能量的工作。以前的出版物已经研究了降低噪声的可能性，或者联合估计衰减和发射图像。我们打算通过研究最优能源窗口来结合这一点，在结果估计中提供准确性和精确度之间的权衡。该项目主要适合EPSRC卫生技术主题的第二个挑战：改变早期预测和诊断。

Positron Emission Tomography (PET) imaging forms a crucial part of managing the patient's treatment pathway in cancer and many other diseases. However, due to limitations on radioactive dose to the patient, PET images can be noisy and with relatively low image resolution. Modern PET scanners have experimental capabilities to measure gamma photon energy information in much more detail than before, which has the potential to increase signal to noise (SNR) ratio, as shown in preliminary scientific papers. However, further work is needed before this can be progressed to commercial systems.This project aims to evaluate the impact of various methods to exploit additional photon energy information on image quality, as well as optimise the most promising candidates. Aspects to be included are scatter rejection by using energy window tuning, photon scatter estimation, accidental coincidences, as well as image reconstruction strategies. We envisage combining established PET data processing with novel techniques using Machine Learning.This project will extend previous work on using photon energy in PET and SPECT. Previous publications have investigated the potential for a decrease in noise or the joint estimation of attenuation and emission images. We intend to combine this by investigating optimal energy windows providing a trade-off between accuracy and precision in the resulting estimates.This project fits primarily in the second Challenge of the EPSRC health technologies theme : Transforming early prediction and diagnosis.