http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
Mammogram 특징 추출을 위한 maskSLIC 기반 CNN 분류 모델
박진혁(Jin Hyeok Park),이병대(Byeong Dae Lee),선우명훈(Myung Hoon Sunwoo) 대한전자공학회 2021 전자공학회논문지 Vol.58 No.10
유방촬영술은 유방암 조기진단을 위한 가장 효과적인 수단이지만, 진단의 정확도가 영상의학과 전문의의 숙련도에 크게 의존하는 경향이 있다. 이를 해결하기 위해 등장한 CNN 기반 유방암 연구는 입력 이미지에 따라 의사 의존적이거나, 효율적인 특징 추출이 어려운 문제점을 내재하고 있었다. 본 논문에서는 유선 조직에 비해 밝은 지점에 병변이 있을 가능성이 높다는 점에서 착안하여, maskSLIC 알고리즘으로 슈퍼픽셀 마스크 이미지를 추출하여 네트워크에 함께 입력하여 전체 이미지의 특징을 효율적으로 추출하였다. 전처리를 거친 각각의 이미지를 특징 추출층과 분류층을 통해 병변의 악성여부를 판별하는 네트워크를 구성하였다. 슈퍼픽셀 이미지 입력과 구성한 네트워크의 유효성을 입증하기 위해 슈퍼픽셀 개수를 N을 조절하며 비교한 결과, N=50일 때 GoogLeNet을 기반으로 한 모델이 정확도 0.8026, AUC 0.8634로 가장 좋은 성능을 보였고, 기존 연구 결과와 비교하여 정확도 4.48~7.6%, AUC 4.63~8.34% 개선된 성능을 보여주었다. Mammography is the most effective means for early diagnosis of breast cancer, but the accuracy of diagnosis tends to depend on the skill level of radiologists. CNN-based breast cancer research, which appeared to solve this problem, has a problems that it is dependent on radiologists and difficult to efficiently extract features depending on the input image. In this paper, focusing on the fact that there is a higher possibility of lesions in bright pixels compared to glandular tissue, the superpixel mask image generated by the maskSLIC algorithm reduces the dependence of the doctor on diagnosis and efficiently extracts the features of the entire image. For each preprocessed image, a network was constructed to determine whether the lesion was malignant through the feature extraction layer and classification layer. As a result of the experiment, the GoogLeNet-based model showed the best performance with an accuracy of 0.8026 and AUC 0.8634, and improved performance by 4.48~7.6% and AUC 4.63~8.34% compared to the previous study results.
시각 자극 기반의 뇌파 분석을 통한 사용자 집중력 분류 모델 설계
박진혁(Jin Hyeok Park),강석환(Seok Hwan Kang),이병문(Byung Mun Lee),강운구(Un Gu Kang),이영호(Young Ho Lee) 한국컴퓨터정보학회 2018 韓國컴퓨터情報學會論文誌 Vol.23 No.11
In this study, we designed a model that can measure the level of user"s concentration by measuring and analyzing EEG data of the subjects who are performing Continuous Performance Test based on visual stimulus. This study focused on alpha and beta waves, which are closely related to concentration in various brain waves. There are a lot of research and services to enhance not only concentration but also brain activity. However, there are formidable barriers to ordinary people for using routinely because of high cost and complex procedures. Therefore, this study designed the model using the portable EEG measurement device with reasonable cost and Visual Continuous Performance Test which we developed as a simplified version of the existing CPT. This study aims to measure the concentration level of the subject objectively through simple and affordable way, EEG analysis. Concentration is also closely related to various brain diseases such as dementia, depression, and ADHD. Therefore, we believe that our proposed model can be useful not only for improving concentration but also brain disease prediction and monitoring research. In addition, the combination of this model and the Brain Computer Interface technology can create greater synergy in various fields.