Semiautomatic Segmentation of Lung Region from Three Dimensional Color Images of Visible Human.
- Author:
Hideaki KIDO
1
;
Kunihiko KANEKO
;
Akifumi MAKINOUCHI
Author Information
1. Graduate School of Systems Life Sciences Kyushu University, Fukuoka, Japan.
- Publication Type:Original Article
- Keywords:
Semiautomatic Segmentation;
Three Dimensional Image;
Lung Region
- MeSH:
Human Body;
Humans*;
Imaging, Three-Dimensional;
Lung*
- From:Journal of Korean Society of Medical Informatics
2007;13(2):171-176
- CountryRepublic of Korea
- Language:English
-
Abstract:
OBJECTIVE: Watershed algorithm is image segmentation algorithm divides the image into numerous small regions. This paper proposes a new approach to extract the lung region from the three dimensional color image of Frozen Human Body (Visible Human Male) based on watershed algorithm. METHODS: After applying this algorithm to input image and getting the small regions, we merge these small regions into one region with three measures based on color, edge marker, and SURFACE respectively. RESULTS: We can say that the smaller number of FALSE-POSITIVE and TRUE NEGATIVE voxels and the larger number of FALSE POSITIVE voxels are better result. Graph shows change in the number of voxel in above groups of the left lung detection when tau color change with tau em is 0.7. We think that the result at the range of tau color from 110 to 180 are better than the other results in Graph. CONCLUSION: Comparing with our previous work, we newly use Canny edge filter for edge marker and define SURFACE-based dissimilarity to relax the problem of its step. The users must select a point within the lung region and some thresholds (taucolor, tauem, tauhigh, taulow, delta) to detect the target region.
