Abstract: Gliomas are highly invasive brain tumors whose cells infiltrate even beyond the tumor regions visible on medical scans. The visible tumor is usually resected, while radiotherapy is used to target the invisible tumor residuals. The standard-of-care radiotherapy plans irradiate the region of the visible tumor, including the resection cavity, extended by a uniform margin. However, the uniform margin does not account for the patient-specific conditions and might not provide optimal dose distribution. We developed a multimodal framework that integrates complementary information from high-resolution MRI scans and highly specific FET-PET metabolic maps to infer tumor infiltration pathways in the patient-specific brain anatomy. An initial clinical population study showed that the radiotherapy plans generated from the inferred tumor cell infiltrations spare more healthy tissue, thus reducing radiation toxicity while yielding comparable accuracy with standard radiotherapy protocols. The proposed framework has potential to provide complementary information to standard radiotherapy plans and assisting personalized radiotherapy design.

About the speaker: Jana is currently a postdoc fellow in the lab of Prof. Faisal Mahmood, in the department of pathology at Brigham and Women’s Hospital. Her work focus on development of AI methods for diagnosis and prognosis, using patient medical data. Jana completed her PhD work at Technical University in Munich, focused on computer vision and mathematical modeling in neuro-oncology. See more info at http://jana-research.org/