bidac6

Collaborators: Scott McNally
Departments: Dept. Radiology and UCAIR (Utah Center for Advanced Imaging Research)

Project:
We aim to automatically identify and segment intracranial aneurysms on 3D time of flight MR angiography acquisitions. We leveraged DeepMedic, a tensorflow-based multi-scale CNN developed for brain lesion segmentation. Its architecture contains one main pathway and two parallel subsampled pathways to efficiently incorporate both local and contextual information to improve segmentation results.


Screenshot DeepMedic Architecture