Abstract #0696

Automatic MS lesion segmentation in the spinal cord using deep learning

Charley Gros¹, Atef Badji^1,2, Josefina Maranzano³, Ren Zhuoquiong⁴, Yaou Liu^4,5, Elise Bannier^6,7, Anne Kerbrat^7,8, Gilles Edan⁸, Pierre Labauge⁹, Virginie Callot^10,11, Jean Pelletier^11,12, Bernard Audoin^11,12, Henitsoa Rasoanandrianina^10,11, Paola Valsasina¹³, Massimo Filippi¹³, Rohit Bakshi¹⁴, Shahamat Tauhid¹⁴, Ferran Prados¹⁵, Marios Yiannakas¹⁵, Hugh Kearney¹⁵, Olga Ciccarelli¹⁵, Sridar Narayanan³, and Julien Cohen-Adad^1,16

¹NeuroPoly Lab, Polytechnique Montreal, Montréal, QC, Canada, ²Faculty of Medicine, University of Montreal, Montreal, QC, Canada, ³Montreal Neurological Institute, Montreal, QC, Canada, ⁴Department of Radiology, Beijing Tiantan Hospital, Capital Medical University, Beijing, China, ⁵Department of Radiology, Xuanwu Hospital, Capital Medical University, Beijing, China, ⁶Department of Radiology, University Hospital of Rennes, Rennes, France, ⁷Univ Rennes, Inria, CNRS, Inserm, IRISA UMR 6074, VISAGES ERL U-1228, F-35000, Rennes, France, ⁸Department of Neurology, University Hospital of Rennes, Rennes, France, ⁹University Hospital of Montpellier, Montpellier, France, ¹⁰CRMBM, CNRS, Aix-Marseille University, Marseille, France, ¹¹CEMEREM, Hôpital de la Timone, AP-HM, Marseille, France, ¹²Department of Neurology, CHU Timone, APHM, Marseille, France, ¹³Neuroimaging Research Unit, INSPE, Division of Neuroscience, San Raffaele Scientific Institute, Vita-Salute San Raffaele University, Milan, Italy, ¹⁴Brigham and Women's Hospital, Harvard Medical School, Boston, MA, United States, ¹⁵Queen Square MS Centre, UCL Institute of Neurology, Faculty of Brain Sciences, University College London, London, United Kingdom, ¹⁶Functional Neuroimaging Unit, CRIUGM, Université de Montréal, Montreal, QC, Canada

Detection of multiple sclerosis (MS) lesions in the spinal cord is clinically important for diagnosis and disease progression assessment. Although several automatic segmentation methods have been proposed for brain lesions, these methods cannot be directly applied to spinal lesions. We propose a fully automatic pipeline based on deep learning to segment the spinal cord and spinal MS lesions, and validate it against a dataset of T₂-w images (265 patients from 5 centers). The proposed cord segmentation achieved better results than the current state-of-the-art, and lesion segmentation yielded a median Dice of 63.4%. The pipeline will be available as open-source.

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

Join Here