복부 CT 영상의 장기 검출을 위한 자동 분할 기법
- Author(s)
- 오상언
- Issued Date
- 2014
- Abstract
- In modern society, age of an aging society is come with higher life expectancy by perception of technological development and health care. For this reason, people actively researches on accurate medical information with high availability. One of them is a method of detecting and measuring the body capacity using image processing in multi-phase abdomen Computerized Tomography(CT) images and Magnetic Resonance(MR) images with functional information. Medical images are interpreted on clinic, but it is insufficient to study on computer-assisted abdomen clinic using image processing on medical images.
In this paper, I propose a useful method to devide the organs automatically in the abdomen CT images. The proposed method is using that each organs in the abdomen CT images have the different gray-level. It extract the organ’s seed point by calculating the 1/4 accumulation on the basis of the histogram on the foreground by Otsu method. Based on the extracted seed points, It detect liver area and kidney/spine area by using the automatic region growing method for image segmentation. Finally, it re-extract the seed points through a process of thinning in the detected area and apply it to a adjacent slice by doing so. It detect another liver area and kidney/spine area. For accuracy assessment of the proposed method, I compare it with the manual segmentation result, so it is verified that the result of the proposed method for liver and kedney/spine segmentation is sufficiently accurate based on 5.7% VOE and 2.3% AVME.
Automatic segmentation method on this paper can be used on computer-assisted clinic system for automatic organ segmentation in abdomen CT images. Not only for that, but also it helps doctors to diagnose using automatic computer segmentation CT images in future.
- Authorize & License
-
- AuthorizeOpen
- Embargo2014-08-26
- Files in This Item:
-
Items in Repository are protected by copyright, with all rights reserved, unless otherwise indicated.