Myelin sheath, wrapped around axons, allows rapid neural signal transmission, and degeneration of myelin causes various neurodegenerative diseases, such as, Multiple Sclerosis (MS). For candidate drug discovery, it is essential to quantify myelin. This requires tedious expert labor comprising myelin labelling on microscopic fluorescence images, usually acquired by confocal microscopes. In this study, semantic segmentation based automatic myelin segmentation on fluorescence microscopy images was introduced. Three-channel and three-dimensional fluorescence images of mouse stem cell derived neuron and oligodendrocyte co-cultures were labeled by an expert. The images were divided into patches for training and the labels corresponded to each patch were acquired. A data set of 11552 patches was used for training to identify myelin and non-myelin regions. In the data set, myelin detection performances of semantic segmentation technique were evaluated using 3 different learning algorithms. The highest accuracy value of 97.32 percent was achieved by using "RMSprop" learning algorithm with a group size of 8 and after 250 epochs. Results suggested that the proposed myelin segmentation method was suitable for detecting myelin. Thus, the outlined myelin segmentation method has the potential to be incorporated into remyelination drug screens.