(G-255) A Comparison Between MATLAB and Python EEG Analysis and Classification

Introduction:: EEG (Electroencephalogram) signal denoising and frequency analysis is an important step for feature extraction and diagnostics of different diseases. Currently, MATLAB remains a popular choice for EEG signal processing. However, Python, an open-source programming language offers many versatile libraries for signal processing. The objective of this study is to compare the CPU time of Python and MATLAB EEG signal analysis and classification and check the accuracy of the Python code.

Materials and Methods:: All 52 samples from the public EEG Dataset Left/Right Hand MI [1] were used to compare the MATLAB and Python codes. The MATLAB code used was Dudhal's EEG Analysis and Classification from Mathworks file exchange [2], which uses MATLAB's signal processing and wavelet toolboxes. The Python code was adapted to use the same algorithm as Dudhal's code and uses the pywt, scipy, numpy, and pandas libraries, along with the wavelet-wrcoef gitHub [3]. CPU time was clocked for the following: time to calculate reconstructed coefficients (t1), frequency analysis of gamma (t2), beta (t3), alpha (t4), theta (t5), and delta (t6) components. A paired t-test was used to compare each CPU time and to compare each component's maximum frequency.

Results, Conclusions, and Discussions:: It was found that MATLAB has a significantly lower CPU time (p< . 01). However, the Python code calculated the same maximum frequency for each component (p >0.99).

For EEG analysis and classification, MATLAB has a faster CPU time and trusted results. The benefit of using Python's open-source code is that it allows for low-cost collaboration. However, more contribution is needed to optimize the code.

Acknowledgements:: This research was partially funded by the SCIO Supporting Structures grant.

References::

1. Dr.Vijay Dudhal (2023). EEG ANALYSIS AND CLASSIFICATION(https://www.mathworks.com/matlabcentral/fileexchange/55112-eeg-analysis-and-classification), MATLAB Central File Exchange. 

2.  Cho H; Ahn M; Ahn S; Kwon M; Jun SC (2017): Supporting data for "EEG datasets for motor imagery brain computer interface" GigaScience Database. https://dx.doi.org/10.5524/100295

3. Ilya Zlotnik. wavelet-wrcoef. https://github.com/izlotnik/wavelet-wrcoef