Effective EEG Feature Selection for Interpretable MDD (Major Depressive Disorder) Classification

In this work, we propose an interpretable electroencephalogram (EEG)-based solution for the diagnostics of major depressive disorder (MDD). The acquisition of EEG experimental data involved 32 MDD patients and 29 healthy controls. A feature matrix is constructed involving frequency decomposition of EEG data based on power spectrum density (PSD) using the Welch method. Those PSD features were selected which were statistically significant. To improve the interpretability, the best features are first selected from features space via the non-dominated sorting genetic (NSGA-II) evolutionary algorithm. The best features are utilized for support vector machine (SVM), and k-nearest neighbors (k-NN) classifiers, and the results are then correlated with features to improve the interpretability. The results show that the features (gamma bands) extracted from the left temporal brain regions can distinguish MDD patients from control significantly.

Reference

If you use this work, please refernce a following paper, where you can find more details.

Vojtech Mrazek, Soyiba Jawed, Muhammad Arif, and Aamir Saeed Malik.
2023. Effective EEG Feature Selection for Interpretable MDD (Major Depressive Disorder) Classification. In Genetic and Evolutionary Computation
Conference (GECCO ’23), July 15–19, 2023, Lisbon, Portugal. ACM, New York,
NY, USA, 9 pages. https://doi.org/10.1145/3583131.3590398

Environment

We used Anaconda to setup the enviroment. All required modules are located in environment.yml

conda env create -f environment.yml

Basically, this project uses MNE, SciPy and Py-ParetoArchive libraries.

Data preparation

Download all Eyes-open data (EO suffix) data from https://figshare.com/articles/dataset/EEG_Data_New/4244171

to the folder data/MDD. Keep only the files ending by _EO.edf. The labels of the subject are determined by the filename.

Running

The scripts are run consequently. Hereby we provide a short description of the scripts.

01_create_data.py

Runs the Welsh filter and stores the numpy arrays to the cache folder

02_standard_classifier.py

Example of standard classifier on the selected data. Just for testing puposes

03_ga_features_selector.py

The core script that uses NSGA-II algorithm to find a correct feature extractors.

05_* and 06_*

Analysis scripts, just as example.

galib library

This library provides an abstraction of chromosome. Chromose class provides a function extract, that selects the precalculated Welsh-filter outputs from the cache. It also provides visulazation functions

Visualize

chrom = ChromosomeChannels(ga_ops)
chrom.from_str('(EEG O2-LE, 30.0, 49.75, AGG)(EEG F8-LE, 8.0, 12.0, DS8)(EEG T6-LE, 8.0, 12.0, DS8)(EEG P4-LE, 12.0, 20.0, DS2)(EEG Fz-LE, 30.0, 49.75, AGG)(EEG P4-LE, 0.0, 4.0, DS8)')
ax = chrom.vizualize() # axis object

plt.savefig("plot.svg")

Topological view

from galib import ChromosomeChannels
chrom = ChromosomeChannels(ga_ops)
chrom.from_str('(EEG O2-LE, 30.0, 49.75, AGG)(EEG F8-LE, 8.0, 12.0, DS8)(EEG T6-LE, 8.0, 12.0, DS8)(EEG P4-LE, 12.0, 20.0, DS2)(EEG Fz-LE, 30.0, 49.75, AGG)(EEG P4-LE, 0.0, 4.0, DS8)')

head = chrom.plot_head("tab:blue") # returns pyplot axis
ax.savefig("head.svg")

or you can sen your own color.

from galib import EEGschema
sch = EEGschema()
cm = plt.get_cmap("viridis")
for ch, c in zip(chrom1.ops.channels, stat_chan):
    #print(ch, c)
    sch.set_channel_color(ch.replace(
        "EEG ", "").replace("-LE", "").lower(), cm(c))

sch.savefig(f"usage_{cl}.svg")

ehw-fit / eeg-mdd