BdPy

Python package for brain decoding analysis

Requirements

• Python 3.8 or later
• numpy
• scipy
• scikit-learn
• pandas
• h5py
• hdf5storage
• pyyaml

Optional requirements

• dataform module
  • pandas
• dl.caffe module
  • Caffe
  • Pillow
  • tqdm
• dl.torch module
  • PyTorch
  • Pillow
• fig module
  • matplotlib
  • Pillow
• bdpy.ml module
  • tqdm
• mri module
  • nipy
  • nibabel
  • pandas
• recon.torch module
  • PyTorch
  • Pillow

Optional requirements for testing

• fastl2lir

Installation

Latest stable release:

$ pip install bdpy

To install the latest development version ("master" branch of the repository), please run the following command.

$ pip install git+https://github.com/KamitaniLab/bdpy.git

Installation (Python 3.6~3.8)

For Python 3.6, use the following commands:

$ pip install "Cython==0.29.36" "numpy"
$ pip install bdpy

Packages

• bdata: BdPy data format (BData) core package
• dataform: Utilities for various data format
• distcomp: Distributed computation utilities
• dl: Deep learning utilities
• feature: Utilities for DNN features
• fig: Utilities for figure creation
• ml: Machine learning utilities
• mri: MRI utilities
• opendata: Open data utilities
• preproc: Utilities for preprocessing
• recon: Reconstruction methods
• stats: Utilities for statistics
• util: Miscellaneous utilities

BdPy data format

BdPy data format (or BrainDecoderToolbox2 data format; BData) consists of two variables: dataset and metadata. dataset stores brain activity data (e.g., voxel signal value for fMRI data), target variables (e.g., ID of stimuli for vision experiments), and additional information specifying experimental design (e.g., run and block numbers for fMRI experiments). Each row corresponds to a single 'sample', and each column representes either single feature (voxel), target, or experiment design information. metadata contains data describing meta-information for each column in dataset.

See BData API examples for useage of BData.

Developers

• Shuntaro C. Aoki (Kyoto Univ)