Abstract
Auditory evoked potentials (AEPs) are commonly acquired to evaluate the structure and integrity of the ascending auditory pathway. Historically, AEPs are either acquired as transient or steady state responses, where recording protocols for each are typically mutually exclusive. Transient response waveforms contain rich information pertaining to neuroanatomical correlates, but their evaluation contains some degree of subjectivity. Steady state responses have gained clinical favorability due to the ease of objective evaluation. However, high stimulation rates used to acquire steady state responses result in significant waveform overlap that obscures waves corresponding to specific neuroanatomical structures. Deconvolution methods have been applied to overcome this restriction imposed by stimulation rate using jittered stimulation sequences that allow for the recovery of the transient (or unitary) response. However, application of deconvolution methods, derived from linear systems theory, to a non-linear and interactive system such as the human auditory system raises concern. The violation of linearity and time-invariance assumptions may result in a distortion of the derived transient response or may introduce nonlinear artifacts. While several authors have speculated about the severity of these issues, the extent of their severity has not been evaluated per application to acquiring the auditory evoked potential. In order to address these concerns, high fidelity deconvolvable sequences were used to obtain a multicomponent AEP transient response rate profile over a wide range of stimulation rates (0.3 - 40/sec) while maintaining uniform recording parameters across rates (stimulus, bandwidth, SNR). This range spans rates conventionally used for acquiring transient (≤2 Hz) and steady state responses (≥20 Hz). Such a rate profile is generated to facilitate investigation of the effects of stimulation rate (or adaptation) at the level of individual neuroanatomical components, as well as to provide a method to evaluate relationships between simultaneously acquired brainstem and cortical auditory processing that is potentially of clinical interest. Additionally, highly jittered sequences are acquired to explore potential deconvolution artifacts and waveform distortions introduced by applying linear systems deconvolution theory to the human auditory system. Synthetic responses for isochronic and jittered stimulation are generated using the deconvolved rate profile and compared to conventionally acquired responses. If the synthetic responses faithfully predict conventionally acquired responses in temporal and spectral domains, we can infer that: a) simple superposition theory is capable of explaining the steady state response for rates between 3.5-40/sec, and b) the sacrifice made in the form of jittered stimulation does not significantly distort the actual response content of the AEP. Results indicate that responses obtained by deconvolution with narrowly jittered sequences are capable of explaining the steady state response using simple superposition theory, and thus are faithful to the underlying neural generators under high rate stimulation. Deconvolution of broadly jittered sequences results in transient responses with minimal effects to early and middle latency responses, and provide a reasonable approximation of the cortical AEPs when considering the large rate span within the sequences. Additionally, use of highly jittered sequences may result in the introduction of predictable artifacts that can be mitigated with judicious sequence design.