Posture Monitoring System Based on Deep Learning for Predicting Patient's Sudden Movement During Radiotherapy

In the past few decades radiation therapy is one of the important methods for cancer treatment. Although human posture recognition has been widely used in fields such as intelligent security, the elderly indoor fall monitoring, and safe driving, few relevant studies focus on the intelligent monitoring of patients during clinical radiotherapy. Improper postures of patients due to physical exhaustion or nervous tension during radiotherapy may affect the precision for radiotherapy and even result in injury to patients. Therefore, it is necessary to propose an intelligent system for monitoring patient's postures during clinical radiotherapy. In this study, we developed a patient's postures recognition system using deep learning with ResNet-18 to detect patient's incorrect postures and sending warnings to technicians to remind patients of maintaining the correct posture, or taking possible emergent actions. The dataset was collected by installing a camera in the treatment room. In addition, a user interface used in the treatment control room was implemented to monitor real-time patient's postures in the treatment room. From the results, it is feasible and reliable to apply ResNet-18 in patient's postures monitoring during clinical radiotherapy. Moreover, the proposed system can avoid the harm to the patient during radiotherapy, which is of great importance for the future development of safe radiotherapy. Therefore, it will be helpful in maintaining patient's setup and improving the radiotherapy accuracy for the patients using this system.