Quick and objective hearing threshold estimation with semi-dry EEG

Authors: Timothe Guyonnet-Hencke1, Eva Koderman1, Marten de Vries1, Michael Schulte2, Alexandra Winkler2, Peter Desain1, Jordy Thielen3, Andy Beynon3, Oscar Portoles1, Mark Kesteren1

2Hörzentrum Oldenburg gGmbH
3Radboud University Medical Centre

Background: The current ‘gold standard’ of audiometry relies on subjective behavioral responses, which are impractical and unreliable for certain groups such as children, individuals with disabilities, or the elderly. Here we present a new electroencephalography (EEG) method with a quick and easy set-up to objectively estimate audiometric thresholds from 32 semi-dry electrodes and short testing time.
Methods: Hearing thresholds at 250, 500, 1000, 2000, 4000, and 8000Hz with 5dB-HL volume resolution were estimated with the proposed EEG system, pure-tone audiometry (PTA), and an automatized behavioral test with the same stimuli as the EEG test (autoBEH). EEG stimuli were 30ms pure-tones presented at 5Hz with a random volume, frequency, and lateralization combination. In 15 participants (10 elderly with asymmetric sensorineural hearing impairment, 5 young normal hearing), EEG, PTA, and autoBEH hearing thresholds were compared using Pearson’s correlation and linear regression analyses.
Results: EEG and PTA had a Pearson’s correlation and regression coefficients (β0,intercept;β1,slope) of 0.8 (β0=-14.6;β1=0.9), 0.9 (β0=-18.5;β1=1.1), 0.9 (β0=-10.2;β1=1.0), 0.9 (β0=-16.2;β1=1.1), 0.9 (β0=-6.6;β1=1.0), and 0.7 (β0=7.9;β1=1.0) for 250, 500, 1000, 2000, 4000, and 8000 Hz, (p<.001). Correlation and regression coefficients between EEG and autoBEH were 0.7 (β0=8.0;β1=0.9), 0.9 (β0=-1.2;β1=1.1), 0.9 (β0=5.2;β1=0.9), 0.9 (β0=-1.3;β1=1.0), 1.0 (β0=0.4;β1=1.1), and 0.7 (β0=12.4;β1=1.0), (p<.001). The absolute error between PTA and EEG remained constant after 5 min of EEG recording.
Conclusion: The strong correlations and the regression slopes between EEG and PTA, and EEG and autoBEH show the validity of the EEG thresholds. The intercepts of regressions show a systematic offset between EEG and PTA, but not between EEG and autoBEH, suggesting different temporal integration between short and long stimuli. This study opens the door to short objective hearing threshold estimation using EEG.