lkorczowski/tinnituseeg

Find EEG markers of Tinnitus using Riemannian Geometry Classification

TinnitusEEG

Toolbox for EEG analysis for finding markers of Tinnitus

Author: Louis Korczowski

Ownership: Robin Guillard EIRL (all rights reserved), previously owned by Zeta Technology

• Code is shared as is for public research purpose and shouldn't be reproduced without the direct consent of the authors.
• Use/reproduction by commercial companies is strictly prohibited (contact us if you want to use/understand that code).
• Some of the EEG Data were shared to us for research purpose, we don't own them (Thanks to S. Vanneste, D. De Ridder and M. Congedo)

Goals

• Be able to differentiate EEG pattern of Tinnitus subjects from non-Tinnitus subjects (spatial, temporal and frequency patterns)
• Maximize accuracy, robustness and "efficiency" of Tinnitus versus non-Tinnitus EEG classification using Riemannian Geometry
• Find EEG markers of residual inhibition

Results

• ROC AUC about 0.926 using only 2 seconds of EEG epochs (Tinnitus versus non-Tinnitus) using simple RG classif
• increasing the number of epochs/length of epochs didn't increase significantly the AUC

Bibliography

• Vanneste et al. 2010 The neural correlates of tinnitus-related distress https://pubmed.ncbi.nlm.nih.gov/20417285/
• De Ridder et al. 2011 The Distressed Brain: A Group Blind Source Separation Analysis on Tinnitus https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0024273