Selene obtained a Master of Science in Biomedical Engineering at the Vrije Universiteit Brussel (VUB) in 2019. At the end of the same year, she started her PhD at the Department of Electronics and Informatics (ETRO). Her research focusses on the application of artificial intelligence on medical images. More specifically, she investigates the possibilities of AI on neuroimaging for clinical decision support in glioma management. 

Spatial tumor tissue region differentiation: revealing glioma heterogeneity and assessing response early-on 

Gliomas account for the majority of malignant primary tumors of the brain. In addition to a poor prognosis, these tumors are associated with a management that is extremely cumbersome. To this day, it is not possible to conclusively obtain accurate characterization of the tumor at diagnosis, nor to reliably assess the response of the tumor to therapy in early follow-up stages through the use of medical imaging alone. Confirmation is only obtainable through biopsy or further follow-up. However, time is of great essence in pathologies with this kind of poor prognosis and invasive biopsies are associated with substantial risks for the patient.

Therefore, we aim to tackle these challenges by assessing the glioma’s heterogeneity on non-invasive, medical imaging alone through the use of advanced image analysis methods, as well as unsupervised machine and deep learning. This assessment allows us to (i) better distinguish tumor infiltration boundaries from healthy tissue, which could improve therapy planning and possibly prevent recurrence; (ii) differentiate therapy-related changes from true progression, which aids earlier determination of the patient’s response and (iii) allow the exploration of yet unknown regions within the active tumor.