Issues In The Deconvolution Techniques Of High-order Auditory Evoked Potential

Auditory evoked potentials (AEPs) are successfully applied to clinical diagnosis of auditory diseases, such as evaluation of infant hearing and identifying the auditory system pathology. The conventional recording paradigm applying an equal stimulus onset asynchrony (SOA) and ensemble averaging method is perfectly used in clinic. As the stimulus rate increases to a degree that an overlapping response arises on account of the superposition of transient AEPs to the successive stimuli, this overlapped AEP is called high stimulus rate AEP (HSR-AEP), and its transient AEP can be named as high-order AEP (HO-AEP). The study of HO-AEP is promising in clinic application. High stimulus rates can exert strong stresses on the auditory system which might reveal the pathology in the auditory system at an earlier stage. Moreover, different reactions of neurons under low and high stimulus rates conditions will allow a more complete evaluation of auditory adaptation.One also expects that using HRS paradigm to shorten AEP recording time since it will take less time when delivering the same number of stimulus events.From the engineering point of view, the HSR-AEP results from the convolution of a HO-AEP and a binary stimulus sequence. Based on this model, a SOA-jittering technique is applied--an unequal SOAs sequence rather than conventional constant SOAs sequence is proposed to implement the deconvolution operation. There are three main deconvolution techniques available at present--Maximum length sequence (MLS) technique, Continuous loop averaging deconvolution (CLAD) technique and Quasi-periodic sequence deconvolution (QSD) technique. The studies of HO-AEP are all based on the convolution model, and many researches are focusing on the characteristics and applications of deconvolution techniques, new processing methods, et al. The major works of this thesis include four projects as follows:1. To investigate the restoring efficiency of evoked responses based on a simulation strategy. Simulated HSR-AEPs were derived from three paradigms--conventional ensemble averaging, CLAD and MLS. By comparing with performance of conventional recording, performances of the other two methods were evaluated in terms of correlation coefficients and Euclidean distances between the ideal and the derived responses. The results showed that MLS can only slightly enhance the performance at the cost of larger stimulus jitter and much more stimulus numbers; while CLAD method with lower jittering even degenerates the quality, suggesting that in practical use the recording efficiency will not be significantly improved by simply using high rate stimulation with overlapping responses introduced.2. Using Wiener filter approach developed by Wang et al, a supplementary method was proposed to carry out the power spectrum estimation of the response using a long-term memory iterative strategy without the need of a prior knowledge of the transient response itself. Validation is accomplished on the simulated data as well as the real AEP data, which yields the performance close to the ideal and theoretical solution.3. To propose and verify a formula for signal to noise ratio (SNR) changes of HO-AEP elicited by MLS, and discuss the recovery quality and the SNR changes of MLS-HO-AEP obtained in different experimental condition. It concludes that the SNR change is proportional to the MLS order with certain noise environment and stimulus rates, while it is inversely proportional to stimulus rates by keeping the same noise environment and MLS order. Using the proposed formula we can estimate the SNR changes of MLS-HO-AEP accurately, which is helpful to provide reference information on MLS selection under different experimental conditions.4. Based on the session-jittering deconvolution technique proposed by Gutschalk et al, the work of this section is to record HSR-AEP with conventional AEP recording method, and applied Tikhonov regularization method to solve the ill-condition problem brought by synthesis matrix. With regularization process, the noise affection to HO-AEP reconstruction is reduced a lot, so the HO-AEP can be recovered by general deconvolution. Using this method, we tested on the deconvolution of auditory brainstem responses (ABR) and middle latency responses (MLRs). It showed that session-jittering deconvolution with Tikhonov regularization method can minimize the noise disturbing during recovering process and make the reconstruction of HO-AEP more reliable.