신경성 운동 질환 또는 사지절단 환자를 위한 눈 깜박임 기반의 보청기 빔 형성 제어 프로토콜 개발

Abstract

신경성 운동 장애가 있거나 사고 또는 질병으로 손가락/팔을 절단한 난청인들에게, 기존의 손가락으로 조작하는 보청기 리모콘은 사용이 불가능하므로, 그들 전용으로 손을 쓰지 않고 조작할 수 있는 대안이 필요하다. 본 연구에서는 보청기 사용자를 위한 눈 깜박임 기반의 빔 형성 제어 방식을 제안하고자 한다. 귀 주위에서 측정된 세 개의 뇌전도 (Electroencephalogram, EEG) 신호를 이용해 삼층 인공 신경망을 기반으로 눈 깜박임 패턴을 감지하였다. 제안된 제어 방식의 성능을 평가하기 위해, 방향을 제어하는 실험을 통해 정확도와 지연 시간을 평가하고, 모의 상황에서 이 방식을 사용했을 때 SNR이 향상되는 수치를 객관적으로 비교하였다. 14명의 피험자로부터 얻은 실험 결과는 68.6±18.5%의 정확도, 10.1±0.9초의 지연 시간으로 산출되었다. 지표 비교 실험에서, 제안된 제어 방식은 모든 모의 상황에서 기존의 정면 초점의 빔 형성 알고리즘을 적용했을 때보다 신호 대 잡음비(Signal-to-noise ratio, SNR)와 분절 신호 대 잡음비 모두 평균 1.4±0.9dB 더 향상되었다. 본 연구에서 제안된 제어 방식은 운동 질환이 있거나 사지 절단된 난청 환자들이 보청기를 제어하기 위한 독창적인 핸즈프리(hands-free) 방식으로 사용될 수 있을 것이다.|Nowadays, most hearing aids (HAs) embed beamforming (BF) as one of many noise-reduction algorithms. BF-embedded HA enables the hearing-impared (HI) person to focus on a specific sound in one direction at will. However, currently there is no hands-free method to control the focusing direction for HI individuals who are unable to manipulate their hands freely due to neurological motor disease, limb amputation, etc. Thus, an alternative method of adjusting the BF direction is necessary. In this study, I propose an eye-blinking-based beamforming control protocol for HA users who cannot use their hands freely. To measure eye-blinking signals, three electroencephalogram (EEG) electrodes were attached around the ears. Three-layer artificial neural network (ANN) was utilized to detect eye-blinking patterns from the measured signals. Based on predetermined set of rules, certain eye-blinking patterns would control and adjust the focusing direction of the HA. The viability of the proposed control protocol was validated by subjective experiments and index comparison tests. From the 14 subjects, the success rates and latency times were 68.6±18.5% and 10.1±0.9 seconds, respectively. Index comparison tests showed that, even though both conventional BF algorithm and the proposed protocol improved signal-to-ratio (SNR), the proposed protocol further improves SNR by 1.4±0.9dB in all simulated situations. The proposed HA control protocol will provide a novel, hands-free method for HI persons who are unable to utilize handheld remote controllers.; Nowadays, most hearing aids (HAs) embed beamforming (BF) as one of many noise-reduction algorithms. BF-embedded HA enables the hearing-impared (HI) person to focus on a specific sound in one direction at will. However, currently there is no hands-free method to control the focusing direction for HI individuals who are unable to manipulate their hands freely due to neurological motor disease, limb amputation, etc. Thus, an alternative method of adjusting the BF direction is necessary. In this study, I propose an eye-blinking-based beamforming control protocol for HA users who cannot use their hands freely. To measure eye-blinking signals, three electroencephalogram (EEG) electrodes were attached around the ears. Three-layer artificial neural network (ANN) was utilized to detect eye-blinking patterns from the measured signals. Based on predetermined set of rules, certain eye-blinking patterns would control and adjust the focusing direction of the HA. The viability of the proposed control protocol was validated by subjective experiments and index comparison tests. From the 14 subjects, the success rates and latency times were 68.6±18.5% and 10.1±0.9 seconds, respectively. Index comparison tests showed that, even though both conventional BF algorithm and the proposed protocol improved signal-to-ratio (SNR), the proposed protocol further improves SNR by 1.4±0.9dB in all simulated situations. The proposed HA control protocol will provide a novel, hands-free method for HI persons who are unable to utilize handheld remote controllers.