Title:Multi Modal Convolutional Neural Networks for Brain Tumor Segmentation

Abstract:In this work, we propose a multi-modal Convolutional Neural Network (CNN) approach for brain tumor segmentation. We investigate how to combine different modalities efficiently in the CNN this http URL adapt various fusion methods, which are previously employed on video recognition problem, to the brain tumor segmentation problem,and we investigate their efficiency in terms of memory and this http URL experiments, which are performed on BRATS dataset, lead us to the conclusion that learning separate representations for each modality and combining them for brain tumor segmentation could increase the performance of CNN systems.