Automation and AI Tools

Abstract

Radiation therapy treatment planning is naturally a large-scale mathematical optimization process. The goal of treatment planning is to customize the dose distribution toward patient-specific anatomy. Over the last few decades, we saw explosive technology development in radiation therapy delivery. Starting from conventional 2D/3D treatment, intensity-modulated radiation therapy (IMRT) and volumetric-modulated arc therapy (VMAT) were later invented to better customize dose gradient around the target with better organ-at-risk (OAR) sparing. Recent advancement of onboard imaging modalities, including but not limited to kV, MR, and PET, allows more confident target localization. This in return pushes the limit of more aggressive treatment planning goals, trying to maximize control while minimizing toxicity. Artificial intelligence (AI) has been no stranger to radiation therapy in the last couple of decades. Thanks to the ever-growing computation power, more complex AI models are achievable for high-dimensional predictions. Various efforts have been invested in using machine learning and AI algorithms to streamline the treatment planning process, inspect plan quality, and improve overall plan quality and consistency. A few early successful AI tools have been made into the clinic and affected patient care. More AI development is expected to improve the landscape of treatment planning.

Ähnliche Arbeiten