Axon segmentation with microscopic images and based on deep convolutional neural networks. The U-Net from ( Ronneberger et al.) is used as a pixel classifier model. Then an MRF is applied for the post-processing. The resulting axon segmentation mask feeds a myelin detector (Zaimi et al.)
