Making cone beam CT imaging fit for aggressive targeting & adaptive re-planning of photon and proton radiotherapy

放射治疗是一种重要的癌症治疗方法，每年约有12.5万名患者接受放射治疗。它通常是在几周内使用多个高能x射线（现在是质子）“束”以每日剂量（分次）递送。这些光束是针对每个病人单独形成的，并设计成在目标疾病的精确位置重叠。目的是给癌细胞最大剂量，同时尽量减少对附近健康组织的剂量。通常的做法是根据治疗开始前拍摄的CT图像来规划这些治疗光束的排列和形状。确保患者和他们的肿瘤靶点在治疗的每一天都处于正确的治疗位置是一项挑战。微小的变化（超过几毫米）可能使治疗前计划失效，导致目标接受过低剂量的辐射（从而减少治愈的机会）或健康组织接受过高剂量的辐射（从而增加副作用的机会）。在治疗室内使用锥束CT （CBCT）成像检查病人的体位、姿势和解剖结构，就在辐射光束打开之前，最近已经变得普遍。然而，患者形状的变化可能是复杂的，这使得很难计算所接受的辐射剂量的变化是否显著——也就是说，是否有必要改变预先计划的治疗来考虑这种变化？我们的目标是简化这一决策过程。我们将开发一种计算机化的方法，使用患者的CBCT图像来计算处方剂量和计划剂量的变化。目前这是不可能的，因为计算辐射剂量需要病人体内组织密度的准确数据，以便确定x射线（或质子）如何与他们的解剖结构相互作用。与用于生成初始治疗计划的CT图像不同，CBCT图像不能提供关于组织密度的准确信息。该项目将开发一种“校正”CBCT图像的方法，以便它们所包含的组织密度信息可用于直接计算释放剂量。这将对放疗患者有很大的好处，因为工作人员可以快速检查是否给予了正确的剂量，或者是否有必要采取措施避免错误的剂量。目前，这一过程非常耗时——组织边界必须手动绘制到CBCT图像上，并假定每个区域的密度值。我们建议开发的技术将加速这种评估，估计大约五分之一的CBCT图像是必要的。另一个好处是，我们的校正方法不仅恢复了准确的CBCT密度值，而且显著提高了视觉图像质量。这使得图像更容易解释，更适合自动分析，有可能进一步节省时间。该项目建立在我们之前的工作基础上，我们已经开发了一种对骨盆或头颈部图像有效的校正方法。我们已经获得了这项发明的英国专利，确保NHS的利益和价值可以最大化。在这个项目中，我们建议将我们的方法扩展到肺部图像中。由于存在组织密度（肺、软组织、骨骼）的巨大差异以及固有的呼吸运动，该部位具有挑战性。我们还将研究校正后的CBCT图像对质子放疗计划的适用性，这是我们在英国开设第一家高能质子治疗中心时面临的一个迫在眉睫的挑战。

Radiotherapy is an important cancer treatment given to about 125,000 patients each year. It is typically delivered in daily doses (fractions) over a period of several weeks using multiple high energy X-ray (and now proton) "beams". The beams are individually shaped for each patient and designed to overlap at the precise location of the target disease. The intention is to give maximum dose to the cancer cells while minimising dose to nearby healthy tissues. Usual practice is to plan the arrangement and shape of these treatment beams based on CT images taken before treatment begins. Ensuring that the patient and their tumour target are in the correct position for treatment on each day of their therapy is challenging. Small changes (more than a few millimetres) could invalidate the pre-treatment planning leading to the target receiving too low a dose of radiation (and hence reduced chance of cure) or healthy tissues receiving too high a dose of radiation (and hence increased chance of side effects). The use of cone-beam CT (CBCT) imaging within the treatment room to check patient position, pose, and anatomy just before the radiation beams are switched on has recently become widespread. However, changes in patient shape can be complex, making it difficult to calculate whether the resulting change in radiation dose received will be significant - that is, will it be necessary to alter the pre-planned treatment to take account of the change?Our aim is to simplify this decision process. We will develop a computerised method that uses a patient's CBCT image to calculate changes from their prescribed and planned dose. Currently this is not possible because calculation of radiation dose requires accurate data on tissue density within the patient, in order to determine how X-rays (or protons) will interact with their anatomy. Unlike CT images, which are used to generate the initial treatment plan, CBCT images do not give accurate information on tissue density.This project will develop a method to "correct" the CBCT images so that the tissue density information that they contain can be used to directly compute delivered doses. This will be of significant benefit to radiotherapy patients since staff we be able to quickly check that the correct dose will be delivered, or if it is necessary to take action to avoid incorrect doses. Currently this process is very time consuming - tissue boundaries have to be manually drawn onto CBCT images and assumed density values assigned to each region. The technology we propose to develop will accelerate such assessments, estimated to be necessary for about one fifth of CBCT images. A further benefit is that our correction method not only restores accurate CBCT density values, but also markedly improves visual image quality. This makes images easier to interpret and more suitable for automatic analysis, with potential for further time savings.The project builds on our previous work, where we have developed a correction method that appears to be effective for pelvic or head and neck images. We have acquired a UK patent for this invention, ensuring that benefits and value to the NHS can be maximised. In this project we propose to extend our method for use in lung images. This site is challenging due to the large differences in tissue densities present (lung, soft-tissue, bone), and the inherent respiratory motion. We will additionally investigate the suitability of corrected CBCT images for the planning of proton radiotherapy, a looming challenge as we move towards the opening of the first high-energy proton therapy centres in the UK.