http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
김덕환 대한소화기암연구학회 2023 Journal of Digestive Cancer Research Vol.11 No.1
Since the 1990s, serrated polyps have been established to contribute to intermediate cancerdevelopment, and their importance has begun to be recognized. Serrated polyps are morphologically difficult to detect through endoscopy, and an effective resection method has not beenestablished. Among serrated polyps, studies on sessile serrated lesions, with a relatively highrisk of colorectal cancer transformation and detected with difficulty, are in progress. Studiesto date describe the endoscopic features as mucus cap, surface debris or stool, attenuation ofunderlying vasculature, cloud-like surface, dark spots in crypts, and ill-defined irregular border. Additionally, it is expected that relatively large serrated polyps can be safely removed throughcold snare resection. A plan for an effective management of serrated polyps through continuousresearch in the future is warranted.
Support Vector Machine Learning for Region-Based Image Retrieval with Relevance Feedback
김덕환,Jae-Won Song,Ju-Hong Lee,Bum-Ghi Choi 한국전자통신연구원 2007 ETRI Journal Vol.29 No.5
We present a relevance feedback approach based on multi-class support vector machine (SVM) learning and cluster-merging which can significantly improve the retrieval performance in region-based image retrieval. Semantically relevant images may exhibit various visual characteristics and may be scattered in several classes in the feature space due to the semantic gap between low-level features and high-level semantics in the user's mind. To find the semantic classes through relevance feedback, the proposed method reduces the burden of completely re-clustering the classes at iterations and classifies multiple classes. Experimental results show that the proposed method is more effective and efficient than the two-class SVM and multi-class relevance feedback methods.
Real-Time Locomotion Mode Recognition Employing Correlation Feature Analysis Using EMG Pattern
김덕환,조치영,류재환 한국전자통신연구원 2014 ETRI Journal Vol.36 No.1
This paper presents a new locomotion mode recognitionmethod based on a transformed correlation featureanalysis using an electromyography (EMG) pattern. Eachmovement is recognized using six weighted subcorrelationfilters, which are applied to the correlation feature analysisthrough the use of six time-domain features. The proposedmethod has a high recognition rate because it reflects theimportance of the different features according to themovements and thereby enables one to recognize real-timeEMG patterns, owing to the rapid execution of thecorrelation feature analysis. The experiment results showthat the discriminating power of the proposed method is85.89% ( ± 2.5) when walking on a level surface, 96.47%( ± 0.9) when going up stairs, and 96.37% ( ± 1.3) whengoing down stairs for given normal movement data. Thismakes its accuracy and stability better than that found forthe principal component analysis and linear discriminantanalysis methods
김덕환,박재만,강상모,이석민,서창우,이인용,이인중 한국잔디학회 2015 Weed & Turfgrass Science Vol.4 No.4
The present research was carried out in order to investigate the occurrence of problematic weed species in Cnidium officinale Makino. Field in South Korea. Total 53 sites of the 3 different regions in S. Korea were investigated from May to October, 2014. In Cnidium officinale fields, the identified weeds were distributed in 35 families and 99 species. Total 5 communities that consist of Commelina communis, Eleocharis kuroguwai, Persicaria vulgaris, Chenopodium album-Acalypha australis, and Galinsoga ciliata dominated the appearance. The weeds occurred in Cnidium officinale fields were divided into three groups in principal component plot analysis (PCA). It was observed that in control weeds plots; 20 plants of Cnidium officinale fresh weight is 739.9 g while the uncontrolled plots have no Cnidium officinale plants. The current investigation could be useful for estimation of future weeds occurrence, weed flora dynamics and establishment of weed control methods in Cnidium officinale fields in Korea. 천궁밭에 발생하는 잡초와 군락의 유형을 분석하여 천궁밭 잡초방제을 위한 기초자료로 활용하기 위하여 2014년 5 월부터 10월까지 총 6회에 걸쳐 잡초조사를 실시하였다. 천궁밭에는 총 35과 99종이 출현하였고, 일년생잡초는 55종, 다년생잡초는 44종으로 분류되었다. 식물사회학적 군락분석에 의한 결과, 닭의장풀, 털별꽃아재비, 봄여뀌, 올방개, 명아주-깨풀이 우점하여 군락의 형태를 보였다. 또한 PCA plot 분석결과 식물사회학적 군락분석과 유사하게 올방개, 명아주-깨풀, 봄여뀌, 닭의장풀, 털별꽃아재비가 군락으로유형화되었으며, 잡초무방제구의 천궁 생산이 전혀 없을정도로 잡초에 대한 천궁의 피해가 심각하였다. 천궁은 8 월 이전 지상부 생육이 왕성해지기 전까지 집중적인 잡초관리가 이루어져야 하며, 흑색멀칭을 하지 않는 헛골에 발생되는 잡초에 대한 초기방제 체계를 마련해야 할 것으로판단된다.
시험·평가·분석 서비스 정보시스템에 대한 고객만족도 조사 모델 수립 및 적용 사례
김덕환,김예진,권지희,김태경 한국품질경영학회 2021 한국품질경영학회 학술대회 Vol.2021 No.-
본 연구에서는 시험·평가·분석 서비스 과정에서 활용되는 정보시스템에 대해 고객만족도 조사 모델을 수립하고 이를 정보시스템 개선 전/후에 적용하여 분석한 사례를 소개하고자 한다. 시험·평가·분석 서비스는 연구개발 과정에서 성능 확인 등을 위해 시험기관에 의뢰하는 서비스로서, 시료에 대한 특성을 객관화, 정량화하여 성적서의 형태로 그 결과를 제공한다. 시험·평가·분석 서비스는 유상으로 이루어지는 서비스의 개념이므로 일반적인 전자상거래와 유사한 성격을 가지면서도, 성적서라는 지식 기반 산출물을 제공하는 과정에서 위변조 방지 등의 기능을 함께 제공해야 하는 독특한 특성을 가진다. 본 연구에서는 시험·평가·분석 서비스를 지원하는 정보시스템을 개선하는 과정에서 개선기회 식별 및 개선 효과성 검증을 위하여 시험·평가·분석 정보시스템에 대한 고객만족도 조사 모델을 수립하였다. 그리고 고객만족도 조사 모델을 활용하여 정보시스템의 개선 전/후에 대해 각각 조사를 실시하여 그 결과를 비교 분석하였다.
Pathogenesis of Inflammatory Bowel Disease and Recent Advances in Biologic Therapies
김덕환,천재희 대한면역학회 2017 Immune Network Vol.17 No.1
Inflammatory bowel disease (IBD) is a chronic intestinal inflammatory disorder with an unknown etiology. IBD is composed of two different disease entities: Crohn’s disease (CD) and ulcerative colitis (UC). IBD has been thought to be idiopathic but has two main attributable causes that include genetic and environmental factors. The gastrointestinal tract in which this disease occurs is central to the immune system, and the innate and the adaptive immune systems are balanced in complex interactions with intestinal microbes under homeostatic conditions. However, in IBD, this homeostasis is disrupted and uncontrolled intestinal inflammation is perpetuated. Recently, the pathogenesis of IBD has become better understood owing to advances in genetic and immunologic technology. Moreover, new therapeutic strategies are now being implemented that accurately target the pathogenesis of IBD. Beyond conventional immunesuppressive therapy, the development of biological agents that target specific disease mechanisms has resulted in more frequent and deeper remission in IBD patients, with mucosal healing as a treatment goal of therapy. Future novel biologics should overcome the limitations of current therapies and ensure that individual patients can be treated with optimal drugs that are safe and precisely target IBD.