Meeting Banner
Abstract #0774

Deep neural network based framework for in-vivo axonal permeability estimation

Ioana Diana Hill1, Marco Palombo1, Mathieu David Santin2,3, Francesca Branzoli2,3, Anne-Charlotte Philippe2,3, Demian Wassermann4,5, Marie-Stephane Aigrot2, Bruno Stankoff2,6, Hui Zhang1, Stephane Lehericy2,7,8, Alexandra Petiet2,7, Daniel C. Alexander1, Olga Ciccarelli9, and Ivana Drobnjak1

1Centre for Medical Image Computing and Dept of Computer Science, University College London, London, United Kingdom, 2CENIR, ICM, Paris, France, 3Inserm U 1127, CNRS UMR 7225, Sorbonne Universités, UPMC Univ Paris 06 UMR S 1127, Institut du Cerveau et de la Moelle épinière, ICM, Paris, France, 4INRIA, Université Côte d'Azur, Sophia-Antipolis, France, 5Parietal, CEA, INRIA, Saclay, Sophia-Antipolis, France, 6AP-HP, Hôpital Saint-Antoine, Paris, France, 7Hôpital de la Pitié Salpêtrière, Sorbonne Universités, UPMC Paris 06 UMR S 1127, Inserm UMR S 1127, CNRS UMR 7225, Institut du Cerveau et de la Moelle épinière, Paris, France, 8AP-HP, Hôpital de la Pitié Salpêtrière, Paris, France, 9Department of Brain Repair and Rehabilitation, Institute of Neurology, University College London, London, United Kingdom

This study introduces a novel framework for estimating permeability from diffusion-weighted MRI data using deep learning. Recent work introduced a random forest (RF) regressor model that outperforms approximate mathematical models (Kärger model). Motivated by recent developments in machine learning, we propose a deep neural network (NN) approach to estimate the permeability associated with the water residence time. We show in simulations and in in-vivo mouse brain data that the NN outperforms the RF method. We further show that the performance of either ML method is unaffected by the choice of training data, i.e. raw diffusion signals or signal-derived features yield the same results.

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

Join Here