DingkunLiu / MI-BCI

MI-BCI is the acronym for minimal invasive brain-computer interface (BCI). The repository is the sample code for the paper "Intracranial brain-computer interface spelling using localized visual motion response."

MI-BCI

MI-BCI means minimal invasive brain-computer interface (BCI). The repository is the sample code for the paper "Intracranial brain-computer interface spelling using localized visual motion response."

Content

• Overview
• System Requirements
• Installation Guide
• Tutorial
• Example Dataset
• Support

Overview

The code demonstrates the basic analysis pipeline of this paper, including fMRI localization and visualization, and SEEG processing.

The fMRI localizer:

The activated MT area is obtained for choosing the optimal implantation spots.

The BCI paradigm:

System Requirements

This package required the following softwares to function properly:

• FreeSurfer 7.1.1
• Brainnetome atlas freesurfer
• python 3.6 or higher

Installation Guide

Python

Note!!! Installation using pip may face a problem of dependency package version conflict! The newest 'mayavi' version has a conflict with the newest 'vtk' package, so in requirement.txt we define the downloading versions to be 'vtk==8.1.2' and 'mayavi==4.6.2'. If you have installed a later version of mayavi or vtk before, please notice the version changing issue.

 git clone https://github.com/HongLabTHU/MI-BCI.git
 cd MI-BCI
 pip install -r requirements.txt

Brainnetome atlas

1. Download the atlas from the official website.
2. Copy the files lh.BN_Atlas.gcs, rh.BN_Atlas.gcs, BN_Atlas_subcortex.gca, BN_Atlas_246_LUT.txt, to the FreeSurfer SUBJECTS_DIR folder.

Tutorial

MRI Preprocessing

To run this demo, you need to recon the surface using your local FreeSurfer:

cd ./MT-Localizer
./recon.sh

Wait until the recon-all command has finished. Then, run fsfast_mt_analysis.sh to analyze fMRI

./fsfast_mt_analysis.sh -p ../data/fsfast --all

Then, perform parcellation based on Brainnetome atlas:

./BN_atlas_mapping.sh -s S1

Electrode mark

Enter module localizer-viz

cd ..
cd ./localizer-viz

Project the marked electrodes to the pial or inflated surface.

python elecs_projection.py -s S1 --fsfast ../data/fsfast --hemi lh -eloc ../data/fsfast/AnatDir/S1_Electrode_locations.dat
# for detailed explaination of the input arguments, please type ``python elecs_projection.py --help ``

Now, we can visualize the result

python viz.py -s S1 --fsfast ../data/fsfast --hemi lh -eloc ../data/fsfast/AnatDir/S1_Electrode_locations.dat

SEEG processing

cd ..
cd SEEG-proc
python seeg_processing.py

ERP waveforms:

HG power:

Example Dataset

For example data, please click here to download. Unzip the file and put the two folders SEEG and fsfast into the data directory.

Support

If you have a question or feedback, or find any problems, please feel free to open a git issue or pull request. Thanks!