Shannon entropy measures for EEG signals in tinnitus

Tinnitus is known as a common clinical symptom, and it comprehensively is essential to understand the underlying mechanisms. Time-varying EEG is considered an appropriate technique to explore brain regions and related activities, and nonlinear methods may extract the irregularities in the EEG signal and gather more expanded information. Therefore, we studied Shannon Entropy in EEG raw data obtained from normal subjects and compared it to data from chronic tinnitus sufferers before and after an intervention.

From the qEEG database of Iran University of Medical Sciences, the School of Rehabilitation Sciences, we have selected 23 healthy and 24 chronic tinnitus subjects. Nineteen subjects of the tinnitus group had benefited from a complete course of sound therapy (binaural beat for a month). QEEGs were measured with a 27-channel EEG amplifier in the sitting and eye-closed position for 3 min. Shannon entropy was investigated for all electrodes separately and compared among groups using a one-way ANOVA statistical test.

Our results revealed a significant difference between healthy and tinnitus subjects (p < 0.05). Post-hoc comparisons using the Bonferroni test showed increased entropy in the tinnitus group for all electrodes (p < 0.05) at low frequencies and most electrodes at mid frequencies. In addition, after the intervention, paired t-test showed a reduction in entropy to somehow above normal control levels for all electrodes at low-frequencies. Such results were accompanying clinical improvement after the intervention.

The increased entropy in tinnitus patients might reflect the chaotic behavior of the brain. Nonlinear methods in EEG studies (Entropy) could be of great importance in understanding tinnitus neurophysiology and might potentially be a suitable criterion for clinical practice.