NetSeg is a software that I developed as part of my honors thesis project at UNC-Chapel Hill, department of Computer Science. It allows the user to use a Convolutional Neural Network (CNN) to segment subcortical brain structures in MR Images (at least 12 months of age). Nick_Honors_Thesis_final.pdf is the written thesis accompanying this project--it contains an in-depth description of the methods used here.
To use this software, you must be in an environment with python 3 installed. Additionally, you will need the SimpleITK and TensorFlow 2 python packages.
For GUI, simply execute the "NetSeg" binary inside your environment. Then you will select a T1 image and a T2 image. Then choose your output type (segmentation and/or probability map), and click "Compute Segmentation". If you chose both output types, you will need to specify save paths for the segmentation file and the probability map file, in that order.
For no GUI, simply use the python script "segment.py". Execute "python3 segment.py -h" for details.