Adaptive noise cancellation in a multimicrophone system for distortion product otoacoustic emission acquisition.

This study focuses on adaptive noise cancellation (ANC) techniques for the acquisition of distortion product otoacoustic emissions (DPOAEs). Otoacoustic emissions (OAEs) are very low level sounds produced by the outer hair cells of normal cochleas, spontaneously or in response to sound stimulation as a byproduct of a frequency and threshold sensitivity increasing mechanism. Current OAE recording systems rely on test probe noise attenuation and synchronous ensemble averaging for increasing signal-to-noise ratios (SNRs). The efficiency of an ANC algorithm for noise suppression was investigated using three microphones: one placed in the test ear, one in the nontest ear for internal noise reference; one near the subject's head for external noise reference. The system proposed was tested with simulations, off-line averaging and real-time implementation of the ANC algorithm. Simulation results showed that the technique had a potential noise reduction capability of 24 dB for complex multifrequency noise signals. Off-line results were positive, with a mean SNR improvement of 4.9 dB. Real-time results indicated that the use of an ANC algorithm in combination with standard averaging methods can reduce noise levels by as much as 10 dB beyond that obtained with standard noise reduction methods and probe attenuation alone.

Adaptive Wiener filtering for improved acquisition of distortion product otoacoustic emissions.

An innovative acoustic noise canceling method using adaptive Wiener filtering (AWF) was developed for improved acquisition of distortion product otoacoustic emissions (DPOAEs). The system used one microphone placed in the test ear for the primary signal. Noise reference signals were obtained from three different sources: (a) pre-stimulus response from the test ear microphone, (b) post-stimulus response from a microphone placed near the head of the subject and (c) post-stimulus response obtained from a microphone placed in the subject's nontest ear. In order to improve spectral estimation, block averaging of a different number of single sweep responses was used. DPOAE data were obtained from 11 ears of healthy newborns in a well-baby nursery of a hospital under typical noise conditions. Simultaneously obtained recordings from all three microphones were digitized, stored and processed off-line to evaluate the effects of AWF with respect to DPOAE detection and signal-to-noise ratio (SNR) improvement. Results show that compared to standard DPOAE processing, AWF improved signal detection and improved SNR.

Measurement of signal and noise characteristics in ongoing auditory brainstem response averaging.

Although diagnostic testing with auditory evoked potentials (EPs) has become routine, quantitative measurements of signal and noise are still lacking. In this study, current signal, power, noise power, and signal-to-noise ratio (SNR) estimation formulas are reviewed and applied to auditory brainstem response averaging. Single-sweep responses to individual sound stimuli are recorded and estimation formulas are evaluated during off-line averaging under various sound level and noise conditions. The results show that the quality of the averaged EP can be quantitatively assessed by the continuous display of the SNR and residual noise estimates during the averaging process. This method also allows the study of different types of averaging techniques to improve EP response acquisition.

Two-dimensional spectral processing of sequential evoked potentials.

The processing of sequential evoked potentials (EPs) is investigated using two-dimensional processing techniques. Two-dimensional EP arrays or images are formed by stacking sequential recordings. Processing is accomplished in the frequency domain by 2-D low-pass filtering using Gaussian filters. Auditory brainstem responses (ABRs), which are the early auditory EPs, are used to investigate the effects of the 2-D filtering on real data. Gaussian filtering improves signal-to-noise ratios by reducing high frequency noise effectively in both intra-EP and inter-EP dimensions. Applications to intra-operative monitoring are simulated with real ABR data.

Automated electrophysiologic hearing testing using a threshold-seeking algorithm.

The efficacy of utilizing an automated algorithm to identify auditory brainstem responses (ABR) was studied. A microcomputer-based threshold-seeking algorithm utilizing click-evoked ABR was developed to determine evoked-response thresholds for automated hearing screening. The software consists of an evoked-response recognizer unit, which determines the presence or absence of a response, and a threshold-tracking unit, which controls the click intensity in order to track the threshold. The response recognizer is based upon correlation methods. Threshold tracking is accomplished using a Parameter Estimation by Sequential Testing (PEST) procedure, which is commonly used to study psychophysical properties of the auditory system. Sound level is automatically adjusted, based on the results of the recognizer and the threshold tracker. Test results were generally obtained in less than 15 minutes per ear. The results of the automated procedure correlate very highly with expert judgments of ABR threshold and show good test-retest reliability, suggesting that automated procedures are viable alternatives to traditional testing methods.

Computer methods for on-line hearing testing with auditory brain stem responses.

On-line computer methods were developed and evaluated for automated hearing testing with auditory brain stem responses (ABR). The system contained (1) a response recognition unit which recognized the presence or absence of an ABR response and (2) a threshold tracking unit which adjusted stimulus intensity and determined the ABR threshold. Two methods, one nontemplate (variance ratio) and one template (cross-correlation), were evaluated for response recognition while three threshold tracking methods were explored: clinical, Békésy, and PEST (parameter estimation by sequential testing). Both response recognition and threshold tracking units were evaluated by computer simulations for on-line operation using a standard set of experimentally recorded ABRs. The results indicated that while all the response recognition methods were reasonably accurate, their on-line use resulted in significant differences in test efficiency due to differences in tracking methods.