bet2

standalone Brain Extraction Tool (bet2) with python wrapper, released by http://fsl.fmrib.ox.ac.uk/

The following features are added:

• reconfigured with CMake (I suggest using CMake version > 3.15.)
• support windows platform
• python wrapper based on pybind11, allow to install with pip (new)
• use 4D numpy array as input and output for multi-echo GRE images (new)
• enable openmp (new)

To install:

git clone --recursive https://github.com/TripleCoenzyme/bet2.git

pip install ./bet2

To use:

import bet2

mask = bet2.run_bet(image_array, voxel_size, f_th, g_th)

where

• image_array is the 4D float numpy image array in (x, y, z, echo)
• voxel_size is a list containing spatial resolution in (x, y, z)
• f_th and g_th is OPTIONAL fractional_threshold and gradient_threshold in original BET (with -f and -g). Default set to 0.4 and 0.

a test.py and a tmp.nii test data can be found in the path.

Known BUG:

1. Output array dtype will be numpy.float32 but not int as given. Please convert output to numpy.uint8 to save space.
2. Only support CMake<3.19 in Ubuntu 18.04, no matter with gcc/g++ 7, 8 or 9.

Thank twice to Liangfu Chen for his contribution.