DeepFixel: Crossing white matter fiber identification through spherical convolutional neural networks Testing data is stored in `fissile_data.tar.gz`. Each .mat file contains 10 examples of crossing fibers separated using FISSILE. `total_fod` is the total fiber orientation distribution function represented as 28 spherical harmonic coefficients in mrtrix format (https://mrtrix.readthedocs.io/en/latest/concepts/spherical_harmonics.html). `separate_fod` is the separated fiber orientation distribution functions found by FISSILE in the frame of the fiber `rotate_separate_fod` is the separated fiber orientation distribution functions found by FISSILE in the frame of the image `est_theta`, `est_phi`, `est_v` are the estimated polar angle, azimuthal angle, and volume fraction of the separated fiber orientation distribution functions `true_theta`, `true_phi`, `true_v` are the simulated angles and volume fractions `true_fod` is the simulated fiber orientation distribution function based on `est_theta`, `est_phi`, `est_v` `best_model_mlp.pth` is the multilayer perceptron model trained to separate 2 and 3 fibers `best_model_scnn.pth` is the spherical CNN model trained to separate 2 and 3 fibers (recommended) See https://github.com/MASILab/spherical_deep_fixel for more info and code to use the data.