Radiation therapy is one of the main treatment methods for cancer. Machine learning can be used in all aspects of clinical practice in radiation therapy, including clinical decision support, automatic segmentation of target volumes, prediction of treatment efficacy and side effects. Despite the challenges of lacking structured data and poor interpretability of models, the application of machine learning in radiotherapy will become increasingly profound and extensive. This review contains three aspects: introduction of machine learning, the clinical application of machine learning in radiotherapy, challenges and solutions.

Objective:To establish the mouse model with radiation-induced pulmonary fibrosis, and to identify and analyze it from the aspects of function, imaging and pathology.Methods:Thirty C57BL/6 mice were randomly divided into the control group, 16 Gy irradiation group and 20Gy irradiation group. The mice in the irradiation groups received a single 16 Gy or 20 Gy chest X-ray irradiation, and underwent functional examination, imaging examination and pathological examination at 3 and 6 months after irradiation.Results:At 6 months after irradiation, hair on the chest and back of the mice turned white and fell off, and the airway resistance was increased significantly. CT images showed extensive patch shadows and consolidation in the lung. Three dimensional reconstruction suggested that the lung of mice was distorted and deformed, and the volume was decreased significantly. Pathological examination confirmed that there was extensive pulmonary fibrosis.Conclusions:Significant pulmonary fibrosis occurs after 6 months of chest irradiation in mice. The animal model of radiation-induced pulmonary fibrosis in C57BL/6 mice was successfully established.

Neoadjuvant chemoradiotherapy combined with surgery is the standard treatment for patients with locally advanced esophageal cancer. However, there is significant variability in how patients respond to neoadjuvant chemoradiotherapy. The value of existing conventional diagnostic methods in predicting the effectiveness of neoadjuvant chemoradiotherapy is limited. Image-based artificial intelligence (AI), particularly radiomics and deep learning technologies, have shown great potential in predicting the efficacy of neoadjuvant chemoradiotherapy by automatically quantifying and analyzing a vast amount of information in medical images. This review summarizes AI research based on CT, positron emission computed tomography (PET-CT), and other imaging modalities, highlighting the current limitations and future directions of the research.