The aim of this project is to automatically and accurately detect and segment organs at risk (OAR) relevant in radiation therapy planning both in MRI and CT and PET which are the most used image modalities for this task. The approach taken to tackle this complex problem was broken down into two parts. An overview of the project pipeline is illustrated in Fig. 1.

Part 1: Organs at risk localization:

The OAR are first localized leveraging artificial intelligence models based on deep learning driven solely by data.

Part 2: Segmentation of organs at risk once an organ of interest is localized, implement, develop and evaluated the deep learning-based models for OAR automatic segmentation.