The use of QSD (q-sequence deconvolution) to recover superposed, transient evoked-responses

Abstract

Objective: We describe q-sequence deconvolution (QSD), a new data acquisition/analysis method for evoked-responses that solves the problem of waveform distortion at high stimulus repetitionrates, due to response overlap. QSD can increase the sensitivity of clinically useful evoked-responses because it is well known that high stimulus repetition-rates are better for detecting pathophysiology.Methods: QSD is applicable to a variety of experimental conditions. Because some QSD-parameters must be chosen by the experimenter, the underlying principles and assumptions of the method are described in detail. The theoretical and mathematical bases of the QSD method are also described, including some equivalent computational formulations.Results: QSD was applied to recordings of the human auditory brainstem response (ABR) at stimulus repetition-rates that overlapped the responses. The transient ABR was recovered at all rates tested (highest 160/s), and showed systematic changes with stimulus repetition-rate within a single subject. Conclusions: QSD offers a new method of recovering brain evoked-response activity having a duration longer than the time between stimuli.Significance: The use of this new technique for analysis of evoked responses will permit examination of brain activation patterns across a broad range of stimulus repetition-rates, some never before studied. Such studies will improve the sensitivity of evoked-responses for the detection of brain pathophysiology. New measures of brain activity may be discovered using QSD. The method also permits the recovery of the transient brain waveforms that overlap to form 'steady-state' waveforms. An additional benefit of the QSD method is that repetition-rate can be isolated as a variable, independent of other stimulus characteristics, even if the response is a nonlinear function of rate.

Some thoughts on the interpretation of steady-state evoked potentials

“…Deconvolution is possible, however, under certain conditions if the stimuli are not equally spaced in time (e.g., [19,20]). This would not be a steady-state response in the traditional sense, and the straightforward frequency-space analysis becomes inaccurate.…”

EMD-based EEG signal enhancement for auditory evoked potential recovery under high stimulus-rate paradigm

“…The basic arithmetic operations that involves and any real number r is defined as r ± = r. Such calculation only occurs when we synthesize the filtered EEG sweeps by summing up IMFs as formulated in Eq. (7). Artifact rejection on the fluctuating layers using Eq.…”

“…To obtain the transient AEP from such EEG recordings, an unwrapping process called deconvolution becomes mandatory. Since the deconvolution process is very sensitive to noises (for reasons detailed in Section 2.2), quality requirement of EEG for HSR-AEP study is much higher than that for conventional paradigm [7].…”

“…Since the large jitter of interstimulus interval (ISI) with an m-sequence imposes an obstacle for practical applications, continuous loop averaging deconvolution (CLAD) paradigms have been developed by Delgado and Özdamar [10]. And Jewett et al [7] also provided an approach to selecting the appropriate stimulus loop (named q-sequence) making the HSR paradigm more practical. However, the CLAD method, inverse filtering in essence, can amplify such irrelevant components as random noise and artifacts within specific frequency bins determined by the properties of the stimulus sequences in frequency domain [1,11,12], introducing a new issue of low AEP recording efficiency [13].…”

Wiener filter deconvolution of overlapping evoked potentials

“…The importance of the stimulus sequence has been intuitively demonstrated recently by Jewett et al (2004). However, these methods did not take into account the filtering characteristics of the deconvolution filter.…”