Evaluation of a posteriori Wiener filtering applied to frequency-following response extraction in the auditory brainstem

• A posterior Wiener filtering was applied to the FFR extraction, as well as sub-band optimal weighted averaging and Median averaging method.
• Wiener achieved the best performance in improving FFR quality and experiment efficiency, thus it is the most suitable method for FFR preprocessing.
• Comprehensive and quantitative quality indices were described to evaluate FFR qualities.

The main goal of the present study was to determine the best preprocessing method for extracting the frequency-following response (FFR) in the auditory brainstem. The a posteriori Wiener filtering (APWF) method was first applied in FFR preprocessing and then compared with the standard method of conventional averaging with artifact rejection (MeanAR). Two other methods, sub-band optimal weighted averaging (SubBand) and median averaging (Median), were also investigated. FFRs were recorded from 10 normal-hearing subjects. A harmonic complex tone with a missing fundamental frequency was used as the sound stimulus. Comprehensive and quantitative indices were constructed to evaluate the quality of FFRs processed by the four methods. The indices in the time domain included the root mean square (RMS) of the residual background noise, RMS of the FFR, and autocorrelation function, and the indices in the frequency domain included the signal-to-noise ratios (SNRs) of the harmonics. The results revealed that the APWF method achieved the best performance in FFR extraction. Additionally, the effect of sweep number on FFR quality was studied. Paired t-tests indicated that APWF required far fewer sweeps compared with other methods in obtaining equivalent high-quality FFRs. In conclusion, APWF is a more suitable method for FFR preprocessing than the existing methods because of its advantages in improving SNR and experiment efficiency.