Meeting Banner
Abstract #1825

QuaSI-MTR (qualitative scans for imaging MTR): deep-learned MTR from routine scans using U-nets.

Antonio Ricciardi1,2, Francesco Grussu1,3, Ferran Prados1,2,4, Baris Kanber2, Rebecca S Samson1, Daniel C Alexander3, Declan T Chard1,5, and Claudia A M Gandini Wheeler-Kingshott1,6,7
1NMR Research Unit, Queen Square MS Centre, Department of Neuroinflammation, UCL Queen Square Institute of Neurology, Faculty of Brain Sciences, University College London, London, United Kingdom, 2Centre for Medical Image Computing, Department of Medical Physics and Biomedical Engineering, University College London, London, United Kingdom, 3Centre for Medical Image Computing, Department of Computer Science, University College London, London, United Kingdom, 4Universitat Oberta de Catalunya, Barcelona, Spain, 5National Institute for Health Research (NIHR) University College London Hospitals (UCLH) Biomedical Research Centre, London, United Kingdom, 6Department of Brain and Behavioral Sciences, University of Pavia, Pavia, Italy, 7Brain MRI 3T Research Center, IRCCS Mondino Foundation, Pavia, Italy

Magnetisation transfer ratio (MTR) is a popular MR-modality for the identification of brain anomalies in multiple sclerosis due to its sensitivity to myelin changes. It however requires dedicated sequences with long acquisition times, which make its applicability in clinics less feasible. In this work, deep learning U-net architectures have been used to extract MTR information directly from routine qualitative images, bypassing the need for specialised acquisitions. Results show strong correlation with MTR and agreement between regional distributions in normal appearing tissues, both in healthy controls and multiple sclerosis patients.

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

Join Here