Development of Auto-Detection Method for Wave V with a Wireless Automated Auditory Brainstem Response (AABR) Measurement System

• Main Authors: Bo-Yu Chao, 趙柏宇
• Other Authors: Chao-Min Wu
• Format: Others
• Language: zh-TW
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/ubbz3d

碩士 === 國立中央大學 === 電機工程學系 === 105 === With the growing popularity of the newborn hearing screening, R.O.C. Ministry of Health and Welfare subsidize infant hearing screening with Automated Auditory Brainstem Response. The Auditory Brainstem Response typically has seven waves when measured. Wave I、III and V are obvious than all the other waves. Detecting wave V is the common method used to identify hearing loss. Previous studies used Kalman Filter with Exponential Weighted Average in an AABR measurement system, but its detecting method still required manual screenshots. Therefore, the semi-automatic detecting method need to be improved. This study uses the Second Exponential Weighted Average approach to improve the detection of ABR signals under noisy condition. In addition to compare the kalman filter with Exponential Weighted Average, the proposed approach also compares Variance of a Single Point used to calculate in ABR. The compared three methods evaluated in this study are the Fast Fourier Transform method, Image and Pattern Analysis 99 method and the Differential method. The Differential method has been chosen for its fast processing speed and accuracy. In order to assess the feasibility of our approach, 5 male subjects with normal hearing and 1 male subject with hearing loss were participated in the first experiment. Wave V wave latencies were measured between 6 and 7.5ms. Although this method has longer latency (0.5ms), it can eliminate most noises, and the ABR waves are easier to be observed. It can automatically recognize and take screenshots to avoid undesired subjective human judgment. In the second experiment, ABR signals from 2 of previous 5 normal hearing subjects were measured in a hearing exam room without closing the door. With significant background noises in this situation, ABR signals were not measured as the result. Consequently, this study is carried out in hearing exam rooms, to avoid interference of background noises on the measurement of ABR signals. This algorithm and the auto-detection method developed from this study actually improve the previous research's problems about the waveforms which contained too many interference and need to take screenshot manually. Therefore, this study makes our system more complete.