Abstract
Segmentation has been a major task in neuroimaging. A large number ofautomated methods have been developed for segmenting healthy and diseased braintissues. In recent years, deep learning techniques have attracted a lot ofattention as a result of their high accuracy in different segmentationproblems. We present a new deep learning based segmentation method, DeepMRSeg,that can be applied in a generic way to a variety of segmentation tasks. Theproposed architecture combines recent advances in the field of biomedical imagesegmentation and computer vision. We use a modified UNet architecture thattakes advantage of multiple convolution filter sizes to achieve multi-scalefeature extraction adaptive to the desired segmentation task. Importantly, ourmethod operates on minimally processed raw MRI scan. We validated our method ona wide range of segmentation tasks, including white matter lesion segmentation,segmentation of deep brain structures and hippocampus segmentation. We providecode and pre-trained models to allow researchers apply our method on their owndatasets.