Abstract #1449

A probabilistic framework to learn average shaped tissue templates and its application to spinal cord image segmentation

Claudia Blaiotta¹, Patrick Freund^1,2, Armin Curt², Jorge Cardoso³, and John Ashburner¹

¹Wellcome Trust Centre for Neuroimaging, University College London, London, United Kingdom, ²Spinal Cord Injury Center Balgrist, University of Zurich, Zurich, Switzerland, ³Centre for Medical Image Computing, University College London, London, United Kingdom

Magnetic resonance imaging of the spinal cord has a pre-eminent role for understanding the physiopathology of neurological disorders; nevertheless it is confronted with numerous technical challenges, which currently limit its applicability. In this work we focus on the problem of automatically extracting and segmenting the cord, a crucial processing step for neuroimaging studies. We present a novel computational framework that allows delineating structures within the cord, thus providing a reliable and fast alternative to manual segmentation. We test the method on a data set of high-resolution cervical scans and demonstrate the consistency of our results with expert manual annotation.

This abstract and the presentation materials are available to members only; a login is required.

Join Here