Advancing Personalized Cancer Care with an Automated Radiomics-Based Radiation Therapy Method

放射治疗（RT）是治疗癌症的主要方法之一。RT的提供是一个复杂的过程，需要临床和技术专业知识。制定放射治疗计划的过程依赖于复杂的技术和大量的人工工作，通常需要数小时到数天的专门时间来计划每个病人，该计划规定了如何将治疗的辐射剂量传递给病人。我们提出了一种自动生成RT计划的方法，该计划可以很容易地集成到现有的临床RT过程中。该研究的目标是开发并向其他机构提供一种自动化的RT治疗计划方法，该方法将i)在几分钟内快速生成RT计划，加快RT过程，ii)根据加拿大已建立的癌症机构的专业知识，允许患者更多地获得RT，以及iii)为每个特定患者量身定制高质量的RT计划。自动化计划方法应用最先进的机器学习、图像配准和优化算法来学习RT图像和RTplan数据中的哪些关系和模式最适合决定在RT计划中剂量应该放置在哪里以及剂量应该如何传递。该方法基于先前治疗过的患者的放疗计划数据自动学习，并生成个性化的放疗计划，无需任何人工干预。我们正在与加拿大的多家癌症机构合作，对自动化规划方法的开发进行评估和贡献，以确保该研究具有广泛的临床适用性，并得到专家的一致审查。拟议的研究适用于所有接受放射治疗的患者，作为其癌症管理的一部分。该研究将改变现有的放射治疗过程，以更好地利用有限的资源，同时仍然确保所有放射治疗患者都能获得高质量、个性化和具有成本效益的医疗保健。

Radiation therapy (RT) is one of the main methods of treating cancer. The delivery of RT is acomplicated process that requires both clinical and technical expertise. The process ofgenerating an RT treatment plan, which specifies how the radiation dose for treatment is to bedelivered to the patient, relies on complex technology and considerable manual effort oftenrequiring hours to days of dedicated time to plan each patient. We are proposing a method toautomatically generate an RT plan that can readily be integrated into the existing clinical RTprocess.The goal of the research is to develop and make available to other institutions an automatedRT treatment planning method that will i) rapidly generate RT plans in minutes, expediting theRT process, ii) allow patients greater access to RT based on the expertise of establishedcancer institutions in Canada, and iii) produce high quality RT plans that are tailored to eachspecific patient.The automated planning method applies state-of-the-art machine learning, image registration,and optimization algorithms to learn which relationships and patterns in RT image and RTplan data are best for deciding where dose should be placed and how dose should bedelivered in an RT plan. The method automatically learns based on RT plan data frompreviously treated patients and generates a personalized RT plan without requiring anymanual intervention.We are engaging multiple cancer institutions in Canada to evaluate and contribute to thedevelopment of the automated planning method in order to ensure the research has wideclinical applicability and has consensus expert review.The proposed research is applicable to all patients receiving RT as part of their cancermanagement. The research will transform the existing RT process to make better use oflimited resources, while still ensuring high quality, personalized, and cost-effective healthcareis accessible for all RT patients.