Non-linear maximum length sequences for the acquisition of the auditory brainstem response

• Main Authors: Bradley, Andrew P., Smith, Andrew, Petoe, Matthew, Dzulkarnain, Ahmad Aidil Arafat, Wilson, Wayne J.
• Format: Conference or Workshop Item
• Language: English; English
• Published: 2014
• Subjects: RF Otorhinolaryngology
• Online Access: http://irep.iium.edu.my/36594/1/XXXII_WCA_2014_Posters.pdf; http://irep.iium.edu.my/36594/5/aidil.pdf; http://irep.iium.edu.my/36594/; http://www.wca2014.com/

A number of quasi-periodic and pseudo-random acoustic stimuli, such as the Q-sequence (QSD) and maximum length sequences (MLS), have been proposed to improve the acquisition efficiency of the auditory brainstem response (ABR). However, these methods are only optimal under the assumption that the ABR is acquired from a linear, time-invariant system, which the auditory system is not. The aim of this work was to investigate whether the MLS reconstruction algorithm could be improved by assuming these non-linearities are both static and invertible. The expected amplitude and latency variation of ABR wave V was modelled using conventional linear reconstruction techniques on data acquired from a single neonate subject at click rates of: 23, 33, 66, 90, 133, 282 and 481 clicks per second (cps) and stimulus intensities of: 35, 45, 55 and 65 dBnHL. MLS ABRs, reconstructed using this model, were then acquired from 16 neonates at a stimulus intensity of 35dBnHL and rates of 90, 133, 282 and 481 cps. The signal-to-noise-ratios of both the linear and non-linear MLS reconstructions were compared using the single point statistic (Fsp). Results showed that, over all rates tested, there was little benefit from the non-linear reconstruction, aside from a marginal improvement in Fsp at the lowest two rates. This would indicate that the non-linearities observed in ABR waveforms are not well modelled as static and invertible and that it is yet to be determined how the ABR adapts, dynamically, to the changes in ISI implicit to pseudo-random sequences.