Implementation and Testing of a Transient Evoked Otoacoustic Emission Measurement System

• Main Authors: Chen, Ya-Han, 陳亞函
• Other Authors: Liu, Yi-Wen
• Format: Others
• Language: zh-TW
• Published: 2016
• Online Access: http://ndltd.ncl.edu.tw/handle/7xda53

碩士 === 國立清華大學 === 電機工程學系 === 105 === Otoacoustic emission (OAE) is commonly used for infants hearing screening. Its main application is to check if the cochlear outer hair cells function normally. Advantages of the OAE include non-invasiveness and correlation to the condition of the outer hair cells. Transient evoked otoacoustic emission (TEOAE) comes from the cochlea when it is stimulated by an acoustic impulse. Within 20ms after the impulse, OAE components occur sequentially from higher to lower frequencies, and the delay time of the 1kHz, 2kHz, 4kHz and 8kHz emission are 11ms, 7.1ms, 4.6ms and 3.0ms respectively. Nowadays, TEOAE is mainly judged by the amplitude instead of the delay time of each frequency for medical care. This study implements a TEOAE system and obtains TEOAE waveforms by averaging across approximately 3000 repetitions. Afterwards, group delay was calculated to view the latency of each frequency, while Hilbert transform is conducted to analyze the instantaneous frequency, and thereby the success was confirmed regarding the acquisition of TEOAE signals. Moreover, by using ConceFT analysis, we can visualize the transient state of each frequency with high precision. It is expected that in the future, TEOAE will be collected effectively within 5 minutes in the clinics. According to current measurement, the average SNR is 13.2dB. This study measures 20 ears repeatedly, acquiring 100 cases of TEOAE waveforms. Classified by the K nearest neighbor, the recognition rate between different ears reaches 94% accuracy, and by principal component analysis, it is found that each data can be sufficiently represented by just 35 dimensions, which is in prospect of applying to biometrics in the future.