폐질환 진단을 위한 폐음 분류 알고리즘의 제안 ( Pattern Classification of lung sounds for the diagnosis of pulmonary dysfunctions )

전창익,여송필,김성환 대한전자공학회 1998 대한전자공학회 학술대회 Vol.1998 No.1

CDCTLMS를 이용한 변환평면 적응 필터링

전창익,여송필,전광석,이진,김성환,Jeon, Chang-Ik,Yeo, Song-Phil,Chun, Kwang-Seok,Lee, Jin,Kim, Sung-Hwan 한국음향학회 2000 韓國音響學會誌 Vol.19 No.8

적응 필터링 기법은 시간에 따라 신호의 통계적 특성이 변화한때 많이 응용되는 신호처리의 분야이다. 적응필터의 성능 평가는 보통 구조의 복잡도 (complexity), 수렴 속도 (convergence speed) 및 오조정 (misadjustment)을 가지고 평가한다. 적응 필터의 실시간 구현을 위해서는 구조가 간단해야 피고 수렴속도는 빨라야 한다. 본 연구에서는 수렴속도 향상을 위해 변환 평면에서의 적응 필터링 기법에 대해 제안하였다. 먼저 CZT (Chirp Z-Transform)의 주파수 확대해석하는 특성과 DCT (Discrete Cosine Transform)의 실수연산 특성을 모두 가지는 CDCT (Chirp Discrete Cosine Transform)을 제안하였고, 이를 이용한 CDCTLMS (Chirp Discrete Cosine Transform LMS)을 제안하였다. 제안된 방법은 기존의 NLMS (normalized least mean square)알고리즘과 대표적인 변환평면 적응 필터링 방법인 DCTLMS (discrete cosine transform LMS) 알고리즘과의 학습곡선 (loaming curve)의 비교를 통하여 우수함을 입증하였으며 실제 초음파 신호처리에의 응용을 보였다. Adaptive filtering method is one of signal processing area which is frequently used in the case of statistical characteristic change in time-varing situation. The performance of adaptive filter is usually evaluated with complexity of its structure, convergence speed and misadjustment. The structure of adaptive filter must be simple and its speed of adaptation must be fast for real-time implementation. In this paper, we propose chirp discrete cosine transform (CDCT), which has the characteristics of CZT (chrip z-transform) and DCT (discrete cosine transform), and then CDCTLMS (chirp discrete cosine transform LMS) using the above mentioned algorithm for the improvement of its speed of adaptation. Using loaming curve, we prove that the proposed method is superior to the conventional US (normalized LMS) algorithm and DCTLMS (discrete cosine transform LMS) algorithm. Also, we show the real application for the ultrasonic signal processing.

심질환 진단을 위한 심자도 파라메터의 추출

전창익(Chang-Ik Jeon),허영(Young Huh),김기욱(Ki-Uk Kim),한병희(Byung-Hee Han),진승오(Seung-Oh Jin),장원석(Won-Suk Chang),이현호(Hyun-Ho Lee) 대한전자공학회 2003 대한전자공학회 학술대회 Vol.2003 No.11

열처리 환경에서 웨이브렛 적응 필터를 이용한 초음파 비파괴 검사의 결함 검출

임내묵,전창익,김성환 한국음향학회 1999 韓國音響學會誌 Vol.18 No.7

본 논문에서는 그레인 잡음을 제거하기 위해서 웨이브렛 변환(wavelet transform)에 근간을 둔 웨이브렛 적응 필터(WLMS adaptive filter : Wavelet domain Least Mean Square adaptive filter)를 사용하였다. 보통 그레인 잡음은 고온의 환경에서 금속의 결정구조가 변화함에 따라 발생된다. 웨이브렛 평면에서의 적응 필터링은 필터의 입력신호를 직교 변환하여 입력으로 이용함으로써 수렴 속도를 향상시킬 수 있는 장점을 가지고 있다. 적응 필터의 기준 입력 신호는 원시 입력 신호를 지연시킨 신호를 이용하였으며, 적응 필터의 출력은 다시 CA-CFAR(Cell Average - Constant False Alarm Rate) 임계 추정기(threshold estimator)를 거쳐 자동적으로 원하는 신호부분만 나타내도록 하였다. 우선 신호의 통계적 특성을 알기 위하여 run 테스트를 수행하여 기준 입력 신호가 비정상성(nonstationarity)을 나타냄을 보였고, 웨이브렛 적응필터가 시평면 적응필터보다 수렴속도면에서 우수함을 보였으며, 각 적응 필터의 출력신호에 대해서 신호대 잡음비를 통해 성능평가를 하였다. 시평면 적응 필터링 후에는 신호대 잡음비가 2-3㏈ 향상을 보였고, 반면 웨이브렛 적응 필터링후에는 신호대 잡음비가 4-6㏈ 향상을 보였다. In this paper, we used the WLMS(Wavelet domain Least Mean Square) adaptive filter based on the wavelet transform to cancel grain noise. Usually, grain noise occurs in changes of the crystalline structure of metals in high temperature environment. It makes the detection of flaw difficult. The WLMS adaptive filtering algorithm establishes the faster convergence rate by orthogonalizaing the input vector of adaptive filter as compared with that of LMS adaptive filtering algorithm in time domain. We implemented the WLMS adaptive filter by using the delayed version of the primary input vector as the reference input vector and then implemented the CA-CFAR(Cell Averaging- Constant False Alarm Rate) threshold estimator. CA-CFAR threshold estimator enables to detect the flaw and back echo signals automatically. Here, we used the output signals of adaptive filter as its input signal. To Cow the statistical characteristic of ultrasonic signals corrupted by grain noise, we performed run test. The results showed that ultrasonic signals are nonstationary signal, that is, signals whose statistical properties vary with time. The performance of each filter is appreciated by the signal-to-noise ratio. After LMS adaptive filtering in time domain, SNR improves to about 2-3㏈ but after WLMS adaptive filtering in wavelet domain, SNR improves to about 4-6㏈.

폐질환 진단을 위한 잡음환경에 강건한 폐음 패턴 분류법에 관한 연구

여송필,전창익,유세근,김덕영,김성환 대한전기학회 2002 전기학회논문지 D Vol.51 No.3

- In this paper, a robust pattern classification of breath sounds for the diagnosis of pulmonary dysfunction in noise environment is proposed. The feature parameter extraction method by highpass lifter algorithm and PM(projection measure) algorithm are used. 17 different groups of breath sounds are experimentally classified and investigated. The classification has been performed by 6 different types of combinations with proposed methods to evaluate the performances, such as ARC with EDM and LCC with EDM, WLCC with EDM, ARC with PM, LCC with PM, WLCC with PM. Furthermore, all feature parameters are extracted to 80th orders by 5th orders step, and all experiments are evaluated in increasing noise environments by degrees SNR 24dB to 0dB. As a results, WLCC which is derived from highpass lifter algorithm, is selected for the feature parameter extraction method. PM is more robust than EDM in noisy environments to test and compare experimental results. WLCC with PM method(WLCC/PM) has a better performance in an increasing noise environment for diagnosis of pulmonary dysfunction.

공간필터에 의한 운동단위 활동전위의 분해능 향상에 관한 연구

양덕진,전창익,이영석,이진,김성환 대한전기학회 2004 전기학회논문지 D Vol.53 No.1(D)

Conventional bipolar surface electromyography(EMG) technique detects only the superimposed electromyographic activity of a large number of motor units due to its low spatial resolution. For the diagnosis of neuromuscular disorder, the information of single MU is required. In this paper, 9 channel array surface electrode system was designed and filter was proposed. Also the MCPT(modified convolution processing technique)method was proposed for the improvement of MUAP resolution. For performance evaluation, power spectrum analysis of random data and raw EMG signal comparison of MUAP shape and quantitative estimation of SNR were executed. As a result, the MUAP resolution improvement of 32% was obtained from the standpoint of the signal-to-noise ratio(SNR).

초음파법에 의한 콘크리트 손상의 정량적 해석에 관한 연구

김세동,전창익,노승용,김성환 한국음향학회 2000 韓國音響學會誌 Vol.19 No.1

본 논문에서는 정상 콘크리트 시편과 손상된 콘크리트 시편으로부터 초음파 신호의 특징 추출에 관해서 연구하였다. 정상 시편과 다양한 손상 시편으로부터 수집된 초음파 신호는 콘크리트 자체의 이종특성에 기인하여 특징을 분류하는데 많은 문제점을 가지고 있다. 본 논문에서는 콘크리트 시편의 특징 추출 방법으로 초음파 신호의 분산에 의한 손상 계수 및 영점 교차 횟수를 이용한 복합 해석을 제안하였다. 또한 초음파 속도법과 Suaris가 제안한 최대 진폭에 의한 손상 계수를 살펴보았다. 본 실험에서는 180kg/㎠ 와 240kg/㎠으로 제작된 두 가지 종류의 콘크리트 시편이 사용되었고, 초음파 신호는 직접 투과방법에 의하여 반사신호를 측정하였다. 실험 결과 제안한 복합 해석 방법이 기존의 초음파 속도법이나 최대 진폭을 이용한 손상 계수에 의한 방법보다 우수한 성능을 보였다. In this paper, we have introduced the feature extraction for the ultrasonic signal analysis of the undamaged and damaged concrete specimens. Since the concrete has the heterogeneous nature by itself, it has been difficult to classify the feature with using ultrasonic signal which is acquired from the undamaged and damaged concrete specimens. Therefore, in this paper, we proposed the combinational analysis which is using both the damage coefficient and the number of zerocrossing for the feature extraction. And the pulse velocity method and the damage coefficient, which was proposed by Suaris, were reviewed. In this experiment, two types of concrete specimen have been considered: 180kg/㎠ and 240kg/㎠. The ultrasonic signals were acquired in normal direction. As a result, it has been that combinational analysis method, which is proposed in this paper, shows the better performance than the traditional ultrasonic pulse velocity method and the damage coefficient using maximum amplitude of the ultrasonic signal in the feature extraction.

머리 움직임 인식을 위한 근전도 신호의 패턴 인식 기법에 관한 연구

이태우,전창익,이영석,유세근,김성환 대한전기학회 2004 전기학회논문지 D Vol.53 No.2(D)

This paper proposes a new method on the EMG AR(autoregressive) modeling in pattern recognition for various head motions. The proper electrode placement in applying AR or cepstral coefficients for EMG signature discrimination is investigated. EMG signals are measured for different 10 motions with two electrode arrangements simultaneously. Electrode pairs are located separately on dominant muscles(S-type arrangement), because the bandwidth of signals obtained from S-type placement is wider than that from C-type(closely in the region between muscles). From the result of EMG pattern recognition test, the proposed mIAR(modified integrated mean autoregressive model) technique improves the recognition rate around 17-21% compared with other the AR and cepstral methods.

제주지역의 akabane virus에 대한 역학조사 및 원인체 분리동정

강완철,김은주,현관종,전창익,김희석,이두식 濟州大學校 農科大學 動物科學硏究所 2000 動物科學論叢 Vol.15 No.1

제주마의 혈청단백질 유전적 다형현상

신진아,김상휘,김영훈,전창익,이경갑 濟州大學校 農科大學 動物科學硏究所 1999 動物科學論叢 Vol.14 No.1

To investigate the genetic polymorphism of serum protein of the horses on Cheju, we assigned horses for three populations, they were Cheju-native horses 40herds, which were precious natural monuments, Cheju-racing horses 55herds in Cheju Racing Association and Thoroughbred horses 46herds. We analyzed the phenotypes and allelic frequencies of serum protein in three populations in Alb, Es, AlB, Tf loci by using a polyacrylamide gel electrophoresis. In albumin locus, it was identified that it is controlled by 2 codominant autosommal alleles A and B, The phenotypes and allelic frequencies of serum protein were Alb^(A)=0.325, Alb^(B)=0.675 in Cheju-native horses, Alb^(A)=0.473, Alb^(B)=0.527 in Cheju-racing horses, Alb^(A)=0.139, Alb^(B)=0.860 in Thoroughbreds. In vitamin D bindmg protein locus, The phenotype of FF was observed in 100% in Cheju-native horses, 98.1% in Cheju-racing horses and 86.9% in Thoroughbreds. The phenotype of FS, SS was not observed in Cheju-native horses. Phenotype SS was not observed in all tested population. In Esterase locus, overall, the frequency of phenotype II was high. In Thoroughbreds, phenotype II was the highest in all tested populations. The gene frequencies of the separate populations were Es^(F)=0.337, Es^(1)=0.662, Es^(S)=0 in Cheju-native horses, Es^(F)=0.309, Es^(1)=0.581, Es^(S)=0.109 in Cheju-racing horses and Es^(F)=0.097, Es^(1)=0.815, Es^(S)=0.086 in Thoroughbreds. In glycoprotein locus, Phenotype KK was observed 100% in all tested populations. In transferrin locus, Tf^(F1) was nonexistent in Cheju-native horses, but in Thoroughbreds was the quantitative preponderance among all breeds. The frequencies of the separate alleles D, F1, F2, H2, 0 and R were 0.1, 0, 0.428, 0.014, 0.314 and 0.142 in Cheju-native horses, 0.16, 0.05, 0.41, 0.04, 0.06, 0.28 in Cheju-racing horses, and 0.365, 0.310, 0.229, 0.013, 0.081 and 0 in Thoroughbreds.