Treatment Planning Validation

In this chapter, we will discuss the use of machine learning to detect errors in radiotherapy plans and charts. We will cover some general principles and established techniques for detecting errors in radiotherapy. We will discuss the rationale for using machine learning to detect large errors or outliers in radiotherapy treatment plans. As a concrete example, an automated error detection system for radiation treatment plans will be described. The technique was based on unsupervised machine learning, i.e., data clustering, and achieved over 90 % success rates in detecting outliers in over 1,000 treatment plans. Finally, future research directions in the clinical applications of machine learning for treatment planning validation will be briefly discussed.