Intelligent diagnosis of tinnitus using electroencephalography

Abstract

Tinnitus is a global challenge with high prevalence and low cure rates, and its clinical examination process is complex and extremely difficult, and there is a lack of a quick and easy way to diagnose objectively. In this study, we found the difference in the mean power spectral density (mPSD) of electroencephalography (EEG) signals between tinnitus patients and healthy people, and realized the classification of tinnitus locations by extracting the connectivity features of the brain. The feature factor with the best classification accuracy was the connectivity feature the Pearson correlation coefficient (PCC), with an accuracy of 99.42%, and the Phase locking value (PLV) also performed well. The experimental conclusions demonstrate that EEG signals can be used as biomarkers to identify the location of tinnitus, and can provide clinicians with a new objective diagnostic strategy.