Ruijie Yang, PhD | Peking University Thrid Hospital;
Competition Sponsor: Chinese Academy of Medical Sciences
Awardee Year: 2023
At present, the total number of cancer annual incidence is about 20 million, 10 million for mortality globally. As the second modality for cancer treatment, radiotherapy is beneficial for 60- 70% cancer patients. Precise tumor targets and organs at risk (OARs) contouring, treatment planning and quality assurance are the most critical factors for the success of radiotherapy, also the most important and time-consuming tasks for radiation oncologists and medical physicists. Nowadays, these tasks are performed manually, with significant variance among different countries, regions, hospitals, radiation oncologists and medical physicists, which undermine severely the safety and quality, availability and consistency of radiotherapy.
The main objective of this project is to develop an interactive cervical cancer radiotherapy structure segmentation foundation model based on prompt technology. The model will achieve real-time interactive structure automatic segmentation for different institutions, different disease stages, multimodality images, and multiple clinical scenarios, without the need for model retraining. By using prompt technology combined with clinical information prior and updated clinical practice, the accuracy and robustness of radiotherapy structure segmentation algorithms will be improved, thereby solving the problem of complex structure and tumor target segmentation and breaking through the key bottleneck of clinical application. This will promote the precision, intelligence, and process automation of radiotherapy, improve quality and efficiency, safety and efficacy. The project aims to improve the homogeneity and accessibility of radiotherapy, contribute to standardizing and guiding clinical practice.