We develop image algorithms to support diagnosis and treatment decisions in stroke. We further aim to simplify the CT imaging protocol by developing new techniques to enable single 4D CT scanning, saving radiation dose, contrast agent and time to diagnosis.
The first algorithms have focused on segmentation of the main cerebral structures, including the intra-cavity space, blood vessels, white matter, gray matter and cerebrospinal fluid. The image below won the 'Best Medical Image' vote in the RSNA image contest of 2016. See also this interview.
Aunt Minnie has extensively covered our work: on intra-cavity segmentation, on wm/gm segmentation and on color-mapping to visualize vascular flow disturbances. The work on intra-cavity segmentation was also covered by SPIE Newsroom.
We have fully switched to deep learning, for good reasons.