Computer Vision News - March 2023

40 Congrats, Doctor Chiara! Magnetic Resonance Imaging (MRI) is a technique for image acquisition which is extensively used in neuro-clinical practice, to support clinicians in making diagnosis and planning treatments. In the last decades, researchers started to develop computational methods to make automatic predictions of variables of interest, such as a subject's diagnosis or prognosis, based on brain MRI scans. Since MRI is able to detect very subtle changes in brain anatomy, these methods have a huge potential in many clinical applications, such as providing early diagnosis, better therapy planning and monitoring, paving the way to personalized treatments.  Current image-based prediction methods mainly focus on deep discriminative learning techniques. While these methods are able to produce accurate results, they have been proven to be difficult to interpret. This may be problematic, since, in many neuroimaging tasks, it is important not only to predict well, but also to interpret morphological changes underlying predictions. Given the limitations of existing techniques, during my PhD I developed a novel method for image-based predictions [ springer ] . Thismethod is based on a lightweight generativemodel, Chiara Mauri recently finished her PhD with the Computational Neuroimaging group at Technical University of Denmark. Her research aimed to develop a novel way of making accurate and interpretable predictions based on brain images. She is soon starting a postdoc at Harvard Medical School, where she will continue her research on computational imaging methods. Congrats, Doctor Chiara!