White Matter, Gray Matter and Cerebrospinal Fluid Segmentation from Brain Magnetic Resonance Imaging Using Adaptive U-Net and Local Convolutional Neural Network

Abstract According to the World Alzheimer Report 2015, 46 million people are living with dementia in world. The diagnosis of diseases helps doctors treating patients better. One signs is related white matter, grey matter and cerebrospinal fluid. Therefore, automatic segmentation three tissues brain imaging especially from magnetic resonance (MRI) plays an important role medical analysis. In this research, we proposed effective approach segment automatically these three-dimensional (3D) MRI. First, a deep learning model used sure unsure regions. region, another classify each pixel. experiments, adaptive U-net regions, Local Convolutional Neural Network (CNN) multiple inputs pixel only region. Our method was evaluated real image database, Internet Brain Segmentation Repository 18 persons (IBSR 18) (https://www.nitrc.org/projects/ibsr) compared state art methods being results very promising.