Abstract:

BACKGROUND: Researchers proposed a lot of auditory brainstem response (ABR) automatic detection methods. But these methods were mainly applied to estimate hearing threshold level, and can only detect a finite number of peaks.
OBJECTIVE: By studying the ABR waveform characteristics and related mathematical methods, to propose a new peak identification algorithm---cosine summation algorithm, which can more accurately detect all the sensation level and hearing level peaks of ABR.
METHODS: ABR detection experiments were performed on 3 Individuals without hearing impairment in Xinhua Hospital. With short sound stimulation, by the stimulus frequency of 23 Hz, at various sound stimulus intensity, 1 024 data points were collected within 10 ms, and a priori experimental data was obtained. With the designed new peak identification algorithm, simulation experiments on each data were carried out, and the accuracy and practicality of the algorithm was tested.
RESULTS AND CONCLUSION: The clinical experience values of ABR wavelet peak latency were regarded as the reference values for automatic identification. A large number of experiments showed that the new algorithm can accurately identify each ABR wavelet peak and quickly calculate the values of sub-wave latency and the amplitude.