Abstract
Objective: To evaluate and compare the specificity, sensitivity and detection time of various time-domain and multi-band frequency domain methods when detecting the auditory brainstem response (ABR). Design: Simulations and subject recorded data were used to assess and compare the performance of the Hotelling’s T2 test (applied in either time or frequency domain), two versions of the modified q-sample uniform scores test and both the Fsp and Fmp, which were evaluated using both conventional F-distributions with assumed degrees of freedom and a bootstrap approach. Study sample: Data consisted of click-evoked ABRs and recordings of EEG background activity from 12 to 17 normal hearing adults, respectively. Results: An overall advantage in sensitivity and detection time was demonstrated for the Hotelling’s T2 test. The false-positive rates (FPRs) of the Fsp and Fmp were also closer to the nominal alpha-level when evaluating statistical significance using the bootstrap approach, as opposed to using conventional F-distributions. The FPRs of the remaining methods were slightly higher than expected. Conclusions: In this work, Hotelling’s T2 outperformed the alternative methods for automatically detecting ABRs. Its promise as a sensitive and efficient detection method should now be tested in a larger clinical study.
Acknowledgements
The authors would like to thank Debbie Cane for collecting the subject recorded ABR data and Sara M. K. Madsen for collecting the recordings of EEG background noise. The data used throughout this study is openly available at the University of Southampton repository at http://doi.org/10.5258/SOTON/D0168.
Declaration of interest
The authors report no conflicts of interest.
This work was supported by the Oticon Fonden and the Engineering and Physical Sciences Research Council ([EPSRC, grant No. EP/M026728/1)].
Supplementary material available online.