Brain Strokes are primarily of 2 kinds: Cerebral Ischemia and Hemorrhage.

Picture below shows how the 2 types of strokes (on either side) differ from the normal brain (middle) as seen through CT scans

HEMORRHAGE | NORMAL | ISCHEMIA

Cerebral Hemorrhage is due to clotting and thickening of the blood inside the brain causing white patches on the CT and MRI brain scans indicating higher density of blood than normal.

Cerebral Ischemia is due to the lack of blood supply in certain regions of the brain causing dark patches on the CT and MRI scans of the brain indicating lower density of materal.

Concept Idea: This asymmetric black and white patches distinct from the normal brain CT due to cerebral ischemia and hemorrhage respectively, makes diagnosing the type of brain stroke a visual classification task for the radiologists. In order to speed up this process of diagnosis, our project aims to help the radiologists by automatically suggesting the type of brain stroke utilizing medical-imaging techniques.

I have personally collaborated with the radiologists from the top medical-research institute in my hometown to prepare a stroke-dataset of about 500 brain-scans. Pre-processing this dataset using various image-processing techniques, I have conducted a survey of different deep learning architectures to classify the brain-scans to ischemia and hemorrhage, and further segment them.