Recent Advancements of Artificial Intelligence in Particle Therapy

TL;DR

This paper surveys recent AI advancements in particle therapy, highlighting its unique challenges and applications across treatment planning, dose verification, image guidance, and adaptive replanning.

Contribution

It provides a comprehensive overview of AI applications specifically in particle therapy, emphasizing differences from photon therapy and discussing future perspectives.

Findings

AI enhances accuracy and efficiency in particle therapy

Distinct AI strategies are needed for photon vs. particle therapy

Identifies challenges and potential pitfalls in AI deployment

Abstract

We are in a golden age of progress in artificial intelligence (AI). Radiotherapy, due to its technology-intensive nature as well as direct human-machine interactions, is perfectly suited for benefitting from AI to enhance accuracy and efficiency. Over the past few years, a vast majority of AI research have already been published in the field of photon therapy, while the applications of AI specifically targeted for particle therapy remain scarcely investigated. There are two distinct differences between photon therapy and particle therapy: beam interaction physics (photons vs. charged particles) and beam delivery mode (e.g. IMRT/VMAT vs. pencil beam scanning). As a result, different strategies of AI deployment are required between these two radiotherapy modalities. In this article, we aim to present a comprehensive survey of recent literatures exclusively focusing on AI-powered particle therapy. Six major aspects are included: treatment planning, dose calculation, range and dose verification, image guidance, quality assurance and adaptive replanning. A number of perspectives as well as potential challenges and common pitfalls, are also discussed.