bciflow 1.0.0.dev8

A package for BCI-related tasks, including dataset loading, preprocessing, feature extraction, and classification.

BCI Flow

Bciflow is a Python package focused on Brain-Computer Interface (BCI)-related work. It provides tools for loading pre-established datasets, performing analysis, pre-processing, filtering, feature extraction, and classification, covering the entire process of creating BCI models. Installation

You can install bciflow directly from PyPI using pip:

pip install bciflow 

Features

• Dataset Loading: Support for popular BCI datasets, such as BCI Competition datasets, OpenBMI, among others.
• Pre-processing: Filtering, artifact removal, normalization, and other data preparation techniques.
• Feature Extraction: Methods for extracting relevant features from EEG signals, such as frequency bands, CSP (Common Spatial Patterns), etc.
• Classification: Implementation of classification algorithms such as SVM, LDA, Neural Networks, among others.
• Complete Pipeline: Facilitates the creation of complete pipelines for processing and analyzing BCI data.

Loading a Dataset

The datasets must be loaded by the user, and the data path must be provided by the user in the path parameter. For example, to load the CBCIC dataset:

from bciflow.datasets import cbcic
# Load CBCIC dataset
data = cbcic(
    subject=1,  # Subject number
    session_list=None,  # List of sessions to load (default: all sessions)
    labels=['left-hand', 'right-hand'],  # Labels to include
    path='data/cbcic/'  # Path to the dataset files
)

Implemented Methods

Method	Category	Implemented
Temporal Filtering
Bandpass Filtering	Temporal Filtering	Yes
Chebyshev II Filter	Pre-processing	Yes
Signal Convolution	Pre-processing	Yes
Empirical Mode Decomposition (EMD)	Temporal Filtering	Yes
Filter Bank (ChebyshevII)	Temporal Filtering	Yes
Filter Bank (Convolution)	Temporal Filtering	Yes
Cubic Resampling	Temporal Filtering	No
FFT-based Resampling	Temporal Filtering	No
Wavelet	Temporal Filtering	Yes
Short-Time Fourier Transform (STFT)	Temporal Filtering	No
Spatial Filtering
Common Spatial Patterns (CSP)	Spatial Filtering	Yes
Euclidean Alignment (EA)	Spatial Filtering	Yes
Feature Extraction
Average Power Spectral Density (APSD)	Feature Extraction	Yes
Curve Length	Feature Extraction	Yes
Log Power	Feature Extraction	Yes
Nonlinear Energy	Feature Extraction	Yes
Welch Periodogram	Feature Extraction	Yes
Canonical Correlation Analysis (CCA)	Feature Extraction	No
Logarithmic PSD (LPSD)	Feature Extraction	No
Feature Selection
Mutual Information Best Individual Features (MIBIF)	Feature Selection	Yes
Classification
Naive Bayes Parzen Window (NBPW)	Classification	Yes
Data Augmentation
EMD-based Augmentation (emdaug)	Data Augmentation	No
Gaussian Noise (gaussiannoise)	Data Augmentation	No
Time-frequency Gaussian Noise (tfgaussiannoise)	Data Augmentation	No
Time-frequency Recombination (tfrecombination)	Data Augmentation	No