MRIMath technology uses Artificial Intelligence (AI) to deliver pixel-level contouring of brain tumors from surrounding healthy tissue to shorten the time it takes to plan radiation treatment, reduce variability among oncologists, and boost patient outcomes. Tumor delineation and quantification remains the weakest link in the search for accuracy in radiotherapy. Inaccuracy and variation in defining critical volumes could compromise treatment outcomes. This critical task is generally performed manually in the clinic. MRIMath is developing an interpretable and trustworthy Physician-in-the-loop AI platform that: i) performs 3D delineation of tumors and organs at risk within the brain with accuracy higher than 90%; ii) outputs an uncertainty map of its prediction that reflects self-assessment of the AI; iii) registers CT-MRI and MRI-MRI pair and group of images; and iv) tracks individual metastases longitudinally. The usability of the software will be evaluated with at least 100 users. Physicians will finish their work in less than 10 minutes and the variability between users will be less than 20%. The project will also perform a large-scale validation study with human medical image data. The MRIMath software - using longitudinal volumetric analysis - detects tumor growth at least 3 months earlier than visual inspection by radiologists.