Multi-Criteria Intensity-Modulated Radiotherapy Optimization

多标准调强放射治疗优化

• 批准号: 7237949
• 负责人: THOMAS R. BORTFELD
• 金额: $ 23.73万
• 依托单位: MASSACHUSETTS GENERAL HOSPITAL
• 依托单位国家: 美国
• 财政年份: 2004
• 资助国家: 美国
• 起止时间: 2004-07-01 至 2008-04-30
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/7237949
• 关键词: Algorithms; Characteristics; Charge; Clinic; Clinical; Communities; Complex; Databases; Detection; Development; Dimensions; Disorder by Site; Dose; Dose-Limiting; Feedback; Goals; Head and neck structure; Human; Individual; Intensity-Modulated Radiotherapy; Lead; Licensing; Manuals; Methodology; Methods; Monitor; Organ; Outcome; Patients; Plant Leaves; Procedures; Process; Prostate; Radiation therapy; Range; Research; Score; Shapes; Software Tools; Solutions; Standards of Weights and Measures; Structure; Surveys; System; Techniques; Time; Tissues; Weight; Work; base; cost; design; experience; graphical user interface; improved; mathematical algorithm; novel strategies; programs; prototype; software development; tool; treatment planning; tumor

DESCRIPTION (provided by applicant): The overall long-term objective of this project is to incorporate clinical decisions through interactive feedback into the inverse treatment planning process for intensity-modulated radiotherapy. The hypotheses are that this will (i) make the inverse planning process more effective and (ii) increase the clinical relevancy of optimized plans, introducing better tradeoffs between target coverage and sparing of healthy tissues. The new approaches to tackle this problem include multi-criteria optimization techniques and an interactive plan navigation tool for searching a pre-calculated treatment plan database. The idea of multi-criteria optimization in treatment planning is that multiple planning criteria in different critical structures and in the target volume can be controlled simultaneously. In contrast, in current inverse planning algorithms a single objective (score function) is maximized or minimized. This conventional optimization gives only limited control of the planning result, and major manual plan tweaking using trial and error is often necessary. For the new multi-criteria optimization, we will use both dose-based criteria and equivalent uniform doses (EUDs) as costlets. In addition, hard physical constraints will be considered in order not to deviate too much from the dose range for which clinical experience exists. An efficient algorithm will be developed to calculate a set of "Pareto" treatment plans, which are defined as treatment plans in which one cannot improve the dose in one organ without compromising at least one other organ. All Pareto solutions for an individual patient will be stored as a "Pareto front" in a treatment plan database. Pareto fronts generated with physical dose and EUD criteria will be compared and we will search for stable regions that are Pareto optimal with respect to both. With the help of a navigation tool, the clinician and treatment planner can then interactively and quickly find the most suitable plan in the database. The navigation will be monitored in order to get a better understanding of the clinical decision process. The proposed mathematical algorithm for the multi-criteria optimization is an adapted Newton barrier algorithm. The navigation tool will be based on a prototype that was used for a preliminary approach with linear objective functions. The new approach will be evaluated using challenging clinical cases.

描述（由申请人提供）：本项目的总体长期目标是通过互动反馈将临床决策纳入调强放疗的逆向治疗计划过程。 假设这将（i）使逆向计划过程更有效，以及（ii）增加优化计划的临床相关性，从而在目标覆盖和健康组织的保留之间引入更好的权衡。 解决这一问题的新方法包括多标准优化技术和用于搜索预先计算的治疗计划数据库的交互式计划导航工具。 治疗计划中的多标准优化的思想是可以同时控制不同关键结构和靶体积中的多个计划标准。 相反，在当前的逆规划算法中，单个目标（得分函数）被最大化或最小化。这种传统的优化只能对规划结果进行有限的控制，并且通常需要使用试错法进行主要的手动规划调整。 对于新的多标准优化，我们将使用基于剂量的标准和等效均匀剂量（EUD）作为costlet。 此外，将考虑严格的物理限制，以免偏离临床经验的剂量范围太多。 将开发一种有效的算法来计算一组“帕累托”治疗计划，这些治疗计划被定义为在不损害至少一个其他器官的情况下无法提高一个器官中的剂量的治疗计划。 个体患者的所有帕累托解将作为“帕累托前沿”存储在治疗计划数据库中。 将比较物理剂量和EUD标准生成的帕累托前沿，我们将搜索帕累托最优的稳定区域。 在导航工具的帮助下，临床医生和治疗计划者可以在数据库中交互式地快速找到最合适的计划。 将监测导航，以便更好地了解临床决策过程。 所提出的多目标优化的数学算法是一个适应牛顿障碍算法。 导航工具将以用于线性目标函数初步方法的原型为基础。新方法将使用具有挑战性的临床病例进行评估。