AIML Research Seminar: Anatomically Aware Brain MRI Segmentation of the Cerebral Vasculature
As humans, we learn to know what is abnormal by establishing an understanding of what the ‘norm’ actually is. Medical tasks require specific knowledge (e.g. radiologists need to be highly trained), and much is known about healthy, ‘normal’ brain anatomy- documented by anatomical atlases and brain models. This knowledge can then be applied to identify abnormal brain structure from medical images, to identify pathology.
This presentation outlined Georgia's approach to synthetic data utilisation in network training, for cerebral vessel segmentation in brain MRI. She discussed the process of using unlabelled brain MRI and cerebral vasculature literature to produce a method that has the potential to not only enhance segmentation performance, but also reduce the need for large amounts of labelled, training data and improve model explainability and reliability, making networks more suitable for real-world clinical applications.
