Research progress of dosiomics in precision radiotherapy

Abstract

Radiotherapy is a conventional method that plays an important role in the comprehensive treatment of tumors. However, it has inevitable side effects that may affect prognosis. Therefore, increasing attention has been paid to radiotherapy-related side effects and prognosis after radiotherapy. With the development of artificial intelligence, high-throughput extraction of quantitative features and correlation analysis of medical images have rapidly developed to improve tumor diagnosis, staging, grading, and personalized treatment. In recent years, there has been growing interest in the use of machine learning models to predict the effects of radiotherapy based on three-dimensional dose distribution maps generated by optimizing radiotherapy plans, which contain dose features or dosiomics that reveal the dose-response relationship of organs and treatment. The use of machine learning modeling to describe the advantages and accuracy of dosiomics in predicting the toxicity and prognosis of radiotherapy has laid a foundation for personalized radiotherapy. This paper aimed to review the achievements of past dosiomics research, introduce the latest advancements in clinical radiotherapy, and discuss the value and future direction of dosiomics in personalized radiotherapy.