Meeting Banner
Abstract #0466

Magnitude versus complex-valued images for spinal cord diffusion MRI: which one is best?

Francesco Grussu1,2, Jelle Veraart3, Marco Battiston1, Torben Schneider4, Julien Cohen-Adad5,6, Manuel Jorge Cardoso7,8, Claudia Angela Gandini Wheeler-Kingshott1,9,10, Els Fieremans3, Daniel C. Alexander2, and Dmitry S. Novikov3

1Queen Square MS Centre, UCL Institute of Neurology, Faculty of Brain Sciences, University College London, London, United Kingdom, 2Centre for Medical Image Computing, Department of Computer Science, University College London, London, United Kingdom, 3Center for Biomedical Imaging, Department of Radiology, New York University School of Medicine, New York, NY, United States, 4Philips UK, Guildford, Surrey, United Kingdom, 5NeuroPoly Lab, Institute of Biomedical Engineering, Polytechnique Montréal, Montréal, QC, Canada, 6Functional Neuroimaging Unit, CRIUGM, Université de Montréal, Montréal, QC, Canada, 7Centre for Medical Image Computing, Department of Medical Physics and Biomedical Engineering, University College London, London, United Kingdom, 8Dementia Research Centre, UCL Institute of Neurology, Faculty of Brain Sciences, University College London, London, United Kingdom, 9Brain MRI 3T Research Centre, C. Mondino National Neurological Institute, Pavia, Italy, 10Department of Brain and Behavioural Sciences, University of Pavia, Pavia, Italy

Advanced diffusion imaging of the spinal cord is hampered by low signal-to-noise ratio, leading to strong Rician bias in magnitude images. Here, we investigate how to mitigate such bias studying complex-valued 3T diffusion scans of the cervical cord. We test two approaches, based on decorrelated phase (DP) and total variation (TV) filtering, corroborating results with simulations. The DP and TV methods, proposed for the brain, can be applied successfully also in the cord. Moreover, they appear useful pre-processing tools for image denoising, as state-of-the-art noise removal based on Marčenko-Pastur principal component analysis (MP-PCA) performs better on complex-valued as opposed to magnitude data.

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

Join Here