A study of adaptation mechanisms based on ABR recorded at high stimulation rate

Auditory brainstem and middle latency responses recorded at fast rates with randomized stimulation

“…The real ABR signals obtained with RSA for a jitter of 0.6 ms at rates of 71, 83, 100, and 125 Hz show additional components similar in morphology to ABR signals appearing with latencies of 10-25 ms. At rates higher than 100 Hz these additional components (which are not part of the response) appear within the first 10 ms of the averaging window and contaminate the response, producing, for example, overestimation of wave I at 125 Hz, of wave II at 167 Hz, and of wave IV at 250 Hz. In contrast, the ABR signals obtained with the I-RSA method present no additional components, and the changes in the morphology of the ABR signals across stimulation rates is consistent with previous literature: As the stimulation rate increases, the amplitude of the components decreases and the latency increases (to a greater extent the more central the components) as a consequence of neural adaptation (Lasky, 1997;Burkard et al, 1990;Valderrama et al, 2014a).…”

“…The nature of De-Bruijn sequences imposes on MLS the restriction of a very high jitter (Jewett et al, 2004;Ozdamar et al, 2007). Some authors believe that high-jittered stimulation sequences are a disadvantage in recording AEPs, because the morphology of the response associated with a stimulus not only depends on the averaged stimulation rate, but also on the preceding ISI; therefore, high-jittered stimulation sequences could evoke auditory responses of different morphology (especially at high rates), and the assumption of a timeinvariant response would not be accomplished (Jewett et al, 2004;Ozdamar and Bohorquez, 2006;Valderrama et al, 2014a). The ADJAR technique obtains the transient AEP iteratively in the time domain by convolving the AEP estimate in each iteration with the statistical ISI distribution of the stimulation sequence to estimate the separate effects of preceding and succeeding stimuli on the averaged response.…”

“…However, the recording of AEPs at higher rates presents certain advantages, as several authors have reported. First, the recording of AEPs at high rates allows the study of neural adaptation (Lasky, 1997;Burkard et al, 1990;Valderrama et al, 2014a). Other authors state that high stimulation rates may improve accuracy in estimating the hearing threshold of a subject (Leung et al, 1998).…”

“…The main advantages of the RSA method are that (a) it facilitates precise control of the jitter of the stimulation sequence, (b) randomized stimulation sequences are easy to generate, since they are not subject to restrictions in the frequency domain, because RSA does not perform deconvolution, and (c) it allows sweeps to be processed separately. These particular advantages were used in Valderrama et al (2014a) to carry out a study of the fast and slow mechanisms of neural adaptation in humans through the separated responses method, which is based on the categorization of sweeps according to their preceding ISI. The categorization of responses according to their preceding ISI may be accomplished because of the individual processing of sweeps that is allowed by RSA.…”

“…Accurate estimation of b H(f) requires (a) the reduction of the power of the noise distribution N(f) by averaging, and (b) the selection of a stimulation sequence s(t) whose frequency components S(f) are not close to zero, otherwise the noise at that frequency would be amplified, leading to instability. Finally, estimation of the AEP in the time domain is the inverse Fourier transform ( Jewett et al, 2004;Ozdamar and Bohorquez, 2006;Valderrama et al, 2014a).…”