Brain white matter lesion segmentation with 2D/3D CNN

Abstract

Automated detection of white matter hyperintensities (WHM) may have a broad clinical use, because WHM appear in several brain diseases. Deep learning architectures have been recently very successful for the segmentation of brain lesions, such as ictus or tumour lesions. We propose a Convolutional Neural Network composed of four parallel data paths whose input is a mixture of 2D/3D windows extracted from multimodal magnetic resonance imaging of the brain. The architecture is lighter than others proposed in the literature for lesion detection so its training is faster. We carry out computational experiments on a dataset of multimodal imaging from 18 subjects, achieving competitive results with state of the art approaches.