pykao/brats2018 repository overview

This repository contains the brain tumor segmentation algorithm used in Brain Tumor Segmentation and Tractographic Feature Extraction from Structural MR Images for Overall Survival Prediction⁠. This is a GPU ONLY version and the minimum requirement for the GPU memory is 11 GB. This may take around 20 minutes to process a BraTS subject. The input data files, i.e., the ‘flair.nii.gz’, ‘t1c.nii.gz’, ‘t1.nii.gz’, ‘t2.nii.gz’ volumes, will be linked to /data and your segmentations are placed in /data/results. The outputs including UCSBVRL_HarvardOxford-sub.nii.gz, HarvardOxford Subcortical atlas in subject space, and tumor_UCSBVRL_class.nii.gz, the predicted tumor mask.

Please cite [1] if you find this repository is helpful to your research.

[1] Kao, Po-Yu, Thuyen Ngo, Angela Zhang, Jefferson W. Chen, and B. S. Manjunath. "Brain tumor segmentation and tractographic feature extraction from structural mr images for overall survival prediction." In International MICCAI Brainlesion Workshop, pp. 128-141. Springer, Cham, 2018.

How to run:

docker run --runtime=nvidia -e NVIDIA_VISIBLE_DEVICES=0 --shm-size 11G -v <host_directory>:/data -it pykao/brats2018:latest