Intensity based Regions Of Interest (ROIs) preselection followed by Convolutional Neuronal Network (CNN) based segmentation for new lesions detection in Multiple Sclerosis

Mariem Hamzaoui (1) , Théodore Soulier (1) , Arya Yazdan-Panah (1) , Marius Schmidt- Mengin (2) , Olivier Colliot (2) , Nicholas Ayache (3) , Bruno Stankoff (1)

1 ICM - Institut du Cerveau = Paris Brain Institute
2 ARAMIS - Algorithms, models and methods for images and signals of the human brain = Algorithmes, modèles et méthodes pour les images et les signaux du cerveau humain [ICM Paris]
3 EPIONE - E-Patient : Images, données & mOdèles pour la médeciNe numériquE

Detecting new lesions is a key aspect of the radiological follow-up of patients with Multiple Sclerosis (MS), leading to eventual changes in their therapeutics. Our pipeline for new lesion detection based on two consecutive FLAIR MRIs consists in two steps. We start by a detection of potential Region Of Interest (ROI) containing new lesions using a sensitive classical image processing procedure. It detects the connected voxels which intensity increases in the second visit compared to the first one. We then apply a filtering procedure to segment voxels corresponding to new lesions in those potential ROI by applying Convolutional Neuronal Networks (CNN).

Mots clés: Classical image processing; CNNs; Hybrid approach

Domaines: Neurobiologie; Imagerie; Traitement des images [eess.IV]; Imagerie médicale; Vision par ordinateur et reconnaissance de formes [cs.CV]; Traitement du signal et de l'image [eess.SP]

Mariem Hamzaoui, Théodore Soulier, Arya Yazdan-Panah, Marius Schmidt- Mengin, Olivier Colliot, et al.. Intensity based Regions Of Interest (ROIs) preselection followed by Convolutional Neuronal Network (CNN) based segmentation for new lesions detection in Multiple Sclerosis. MICCAI 2021 MSSEG2 - 24th International Conference on Medical Image Computing and Computer Assisted Intervention - Challenge on multiple sclerosis new lesions segmentation challenge using a data management and processing infrastructure — MICCAI-MSSEG-2, Sep 2021, Strasbourg, France. ⟨hal-03826791⟩