Implementation of AI-based contouring and treatment planning for daily personalized RT treatments for lung cancer

The general research objective is to investigate the potential of a CBCT-only workflow for online adaptive radiation therapy for lung cancer. Radiotherapy is the standard treatment for (inoperable) non-small cell lung tumors. Thanks to radio(chemo)therapy, it can be shown that such tumors shrink by 2,5% per day. This regression also means that higher doses are delivered to surrounding organs-at-risk. In addition, it was shown that the highest possible radiation dose upon the lesion is necessary for improved locoregional control and overall survival. Consequently, periodic adaptations of the radiotherapy treatment plan are necessary to reduce toxicity and induce survival. Since several years the use of convolutional neural networks (CNNs) has been investigated to improve the quality of the preparation of a radiotherapy treatment plan. Besides an improvement in quality, it is seen that AI also decreases the time needed to prepare a radiation treatment plan, making it suitable to use it in adaptive radiotherapy, where the plan is renewed during the course of the treatment. This current project aims at studying the advantages and challenges of the clinical implementation of AI in the daily routine of the treatment of lung cancer in the Radiation-Oncology department of the University Hospital of Leuven.