Recent Advances in Biomedical & Chemical Engineering and Material Science
巻, 号, ページ
pp. 34-39
出版年月
2014年3月15日
出版者
和文:
英文:
会議名称
和文:
英文:
International Conference on Biology and Biomedical Engineering
開催地
和文:
英文:
Venice
アブストラクト
This paper presents a combination of two ideas to segment multi-organs in CT images. First, we modified the K-means clustering through a hierarchical concept. It was aimed to get a correlation between clustered indexes and different types of tissues. Second, we proposed a simple method to segment multiple organs. The proposed method was ased on tissue types, some morphological operations, and basic information of anatomical structures. We applied the proposed method to the CT liver-images acquired by a 4D-CT imaging system. From our experiment, the proposed clustering method described clustered indexes in accordance with five types of regions including background and four different types of tissues. Further, it could reduce the ratio of losing regions, which were randomly occurred when using the original K-means and fuzzy C-means. Further, the multi-organ segmentation method gave attractive outlines on liver, kidneys, and spleens. Moreover, liver-regions were compared with manual-drawing regions performed by a radiologist. From ten sets of 3D-CT images, the proposed method demonstrated the average similarity measures about 87.4 percent.
各種検索
サポート
ヘルプ