Access &
Workflow
A guide to downloading the datasets, setting up the analysis environment, and understanding the ethical use requirements.
Data Acquisition
Datasets are hosted on public data repository such as OpenNeuro and FigShare. OpenNeuro provides web-based access and AWS S3 downloads.
pip install openneuro-py
openneuro-py download --dataset ds00XXXX --target ./data/
Workflow
for Utilization
Step 1
Download & Validate
Download the dataset from OpenNeuro. Verify BIDS compliance using the BIDS Validator.
pip install bids-validator
bids-validator ./data/ds00XXXX/
Step 2
Preprocessing
Use ArKairos to apply the standardized preprocessing pipeline used in our publications:
- 50 Hz notch filter (+ harmonics)
- Common average reference (CAR)
- 2–120 Hz bandpass filter
- Adaptive EMG artifact removal (for EEG)
- Resampling to 240 Hz
bmica preprocess --config preprocess_default.yaml --input ./data/ds00XXXX/
Step 3
Analysis & Decoding
Load data with our PyTorch dataloaders and train or evaluate decoding models. We provide pre-trained weights corresponding to each published result.
from bmica.datasets import SpeechEEGDataset
from bmica.models import EEGConformer
dataset = SpeechEEGDataset("ds00XXXX", task="word", split="test")
model = EEGConformer.from_pretrained("araya/silent-speech-word-all")
results = model.evaluate(dataset)
print(f"Word Classification Accuracy: {results['wca']:.1%}")
Ethical Use Policy
This dataset contains EEG and EMG recordings from human participants. Although fully anonymized, users must adhere to the following terms by downloading the data:
- Do not attempt to re-identify any participant.
- Data is for non-commercial research purposes unless separately licensed.
- All academic publications using this dataset must cite the associated paper (see Publications).
- The data is provided as-is. Data providers accept no liability for direct or indirect damages arising from its use.
This study was approved by the Shiba Palace Clinic Ethics Review Committee and conducted in compliance with the Declaration of Helsinki.