lucasplagwitz / rm_masking

Rat-Mouse Brain Masking with U-Net

This project demonstrates the application of the U-Net architecture for masking mouse and rat brains. It focuses on:

1. the possibility/necessity of transfer-learning from mouse to rat,
2. an analysis of required training data size, and
3. providing a Nifti in-out brain masker.

The code and net structure is inspired by a PyTorch U-Net implementation. The training process of the model will be added in one of the next versions.

Demo

Input	U-Net mask	U-Net Output

Installation

• Activate the Anaconda environment if available.
• If git is missing: install it via conda install -c anaconda git
• Install the package with pip install git+https://github.com/lucasplagwitz/rm_masking.git.
• Start the transformation with python -c "from rm_masking import predict; predict.run(r'/path/to/input_niis/', r'/path/to/output_niis/')"

References

[1] O. Ronneberger, P. Fischer, T. Brox: U-Net: Convolutional Networks for Biomedical Image Segmentation. Medical Image Computing and Computer-Assisted Intervention – MICCAI 2015 pp 234-241.