http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
특징점 선택방법과 SVM 학습법을 이용한 당뇨병 데이터에서의 당뇨병성 신장합병증의 예측
조백환,이종실,지영준,김광원,김인영,김선일,Cho, Baek-Hwan,Lee, Jong-Shill,Chee, Young-Joan,Kim, Kwang-Won,Kim, In-Young,Kim, Sun-I. 대한의용생체공학회 2007 의공학회지 Vol.28 No.3
Diabetes mellitus can cause devastating complications, which often result in disability and death, and diabetic nephropathy is a leading cause of death in people with diabetes. In this study, we tried to predict the onset of diabetic nephropathy from an irregular and unbalanced diabetic dataset. We collected clinical data from 292 patients with type 2 diabetes and performed preprocessing to extract 184 features to resolve the irregularity of the dataset. We compared several feature selection methods, such as ReliefF and sensitivity analysis, to remove redundant features and improve the classification performance. We also compared learning methods with support vector machine, such as equal cost learning and cost-sensitive learning to tackle the unbalanced problem in the dataset. The best classifier with the 39 selected features gave 0.969 of the area under the curve by receiver operation characteristics analysis, which represents that our method can predict diabetic nephropathy with high generalization performance from an irregular and unbalanced dataset, and physicians can benefit from it for predicting diabetic nephropathy.
계층구조적 분류모델을 이용한 심전도에서의 비정상 비트 검출
이도훈,조백환,박관수,송수화,이종실,지영준,김인영,김선일,Lee, Do-Hoon,Cho, Baek-Hwan,Park, Kwan-Soo,Song, Soo-Hwa,Lee, Jong-Shill,Chee, Young-Joon,Kim, In-Young,Kim, Sun-Il 대한의용생체공학회 2008 의공학회지 Vol.29 No.6
The more people use ambulatory electrocardiogram(ECG) for arrhythmia detection, the more researchers report the automatic classification algorithms. Most of the previous studies don't consider the un-balanced data distribution. Even in patients, there are much more normal beats than abnormal beats among the data from 24 hours. To solve this problem, the hierarchical classification using 21 features was adopted for arrhythmia abnormal beat detection. The features include R-R intervals and data to describe the morphology of the wave. To validate the algorithm, 44 non-pacemaker recordings from physionet were used. The hierarchical classification model with 2 stages on domain knowledge was constructed. Using our suggested method, we could improve the performance in abnormal beat classification from the conventional multi-class classification method. In conclusion, the domain knowledge based hierarchical classification is useful to the ECG beat classification with unbalanced data distribution.
강정훈,조백환,이종실,지영준,김인영,김선일,Kang, Joung-Hoon,Cho, Baek-Hwan,Lee, Jong-Shill,Chee, Young-Joon,Kim, In-Young,Kim, Sun-I. 대한의용생체공학회 2007 의공학회지 Vol.28 No.3
With the convergence of ubiquitous networking and medical technologies, ubiquitous healthcare(U-Healthcare) service has come in our life, which enables a patient to receive medical services at anytime and anywhere. In the u-Healthcare environment, intelligent real-time biosignal aquisition/analysis techniques are inevitable. In this study, we propose a motion artifact cancelation method in portable photoplethysmography(PPG) signal aquisition using an accelerometer and an adaptive filter. A preliminary experiment represented that the component of the pedestrian motion artifact can be found under 5Hz in the spectral analysis. Therefore, we collected PPG signals under both simulated conditions with a motor that generates circular motion with uniform velocity (from 1 to 5Hz) and a real walking condition. We then reduced the motion artifact using a recursive least square adaptive filter which takes the accelerometer output as a noise reference. The results showed that the adaptive filter can remove the motion artifact effectively and recover peak points in PPG signals, which represents our method can be useful to detect heart rate in real walking condition.
디지털 유방영상에서 멀티영상 기반의 컴퓨터 보조 진단에 관한 연구
최형식,조용호,조백환,문우경,임정기,김인영,김선일,Choi, Hyoung-Sik,Cho, Yong-Ho,Cho, Baek-Hwan,Moon, Woo-Kyoung,Im, Jung-Gi,Kim, In-Young,Kim, Sun-I. 대한의용생체공학회 2007 의공학회지 Vol.28 No.1
For the past decade, the full-field digital mammography has been widely used for early diagnosis of breast cancer, and computer aided diagnosis has been developed to assist physicians as a second opinion. In this study, we try to predict the breast cancer using both mediolateral oblique(MLO) view and craniocaudal(CC) view together. A skilled radiologist selected 35 pairs of ROIs from both MLO view and CC view of digital mammogram. We extracted textural features using Spatial Grey Level Dependence matrix from each mammogram and evaluated the generalization performance of the classifier using Support Vector Machine. We compared the multi-view based classifier to single-view based classifier that is built from each mammogram view. The results represent that the multi-view based computer aided diagnosis in digital mammogram could improve the diagnostic performance and have good possibility for clinical use to assist physicians as a second opinion.
Susceptance를 이용한 피부수화도 측정 장비의 개발 및 평가
김홍식,장우영,신건수,조백환,김인영,김선일,Kim, Hong-Sig,Jang, Woo-Young,Shin, Kun-Soo,Cho, Baek-Hwan,Kim, In-Young,Kim, Sun-I. 대한의용생체공학회 2008 의공학회지 Vol.29 No.6
In this paper, a novel system is proposed to measure skin hydration using the susceptance method. This system largely consists of a low-voltage(${\pm}2.6$ V) driven circuit and minimized electrodes of size($5{\times}5mm^2$). To evaluate the accuracy of the novel system in measuring skin hydration, skin hydration values from 105 subjects are measured by the proposed system. The measurements are then compared to those obtained by the golden reference device based on the capacitance method in terms of Intraclass Correlation Coefficient(ICC) and correlation coefficient. All measurements are performed on 7 sites, which are forehead, Crow's foot, cheek, chin, volar forearm, dorsal forearm, and back of the hand, in a room where the temperature and humidity are maintained at an uniform level of $22{\pm}2^{\circ}C$ and $50{\pm}5%$, respectively. ICC values are above 0.9(p=0.001), signifying that the skin hydration values measured by the two methods show a good level of reliability. Correlation coefficient between the two methods is also 0.562(p=0.001). Based on these results, it is expected that the proposed system may be applicable in a variety of clinical or cosmetic areas.
고재영 ( Jae-yeong Ko ),조백환 ( Baek-hwan Cho ),정명진 ( Myung-jin Chung ) 한국정보처리학회 2020 한국정보처리학회 학술대회논문집 Vol.27 No.1
의료 데이터를 이용하여 인공지능 기계학습 연구를 수행할 때 자주 마주하는 문제는 데이터 불균형, 데이터 부족 등이며 특히 정제된 충분한 데이터를 구하기 힘들다는 것이 큰 문제이다. 본 연구에서는 이를 해결하기 위해 GAN(Generative Adversarial Network) 기반 고해상도 의료 영상을 생성하는 프레임워크를 개발하고자 한다. 각 해상도 마다 Scale 의 Gradient 를 동시에 학습하여 빠르게 고해상도 이미지를 생성 해낼 수 있도록 했다. 고해상도 이미지를 생성하는 Neural Network 를 고안하였으며, PGGAN, Style-GAN 과의 성능 비교를 통해 제안된 모델이 양질의 고해상도 의료영상 이미지를 더 빠르게 생성할 수 있음을 확인하였다. 이를 통해 인공지능 기계학습 연구에 있어서 의료 영상의 데이터 부족, 데이터 불균형 문제를 해결할 수 있는 Data augmentation 이나, Anomaly detection 등의 연구에 적용할 수 있다.