Discipline: Computer Sciences and Information Management
Subcategory: Computer Science & Information Systems
Alysia Irwin - Fort Valley State University
Co-Author(s): Chunhua Dong, Fort Valley State University, GA
In the recent years a large amount of research work has been devoted to the development of semi-automatic techniques for the analysis of abdominal Computed Tomography (CT) images. Region growing (RG) is an efficient, semi-automatic approach to segment the organs in an abdominal CT image. However, the RG algorithm requires the user to provide an initial seed pixel and a suitable threshold value for the region membership criterion. Meanwhile, boundaries of segmentation results are not smooth due to the pixel-wise growing strategy. Having a more precise representation of the target organ within the image is imperative for accuracy in computer-aided segmentation.
In this study, we propose an adaptive threshold setting mechanism (AdaptRG) for organ segmentation. In addition, a homogeneous discriminative model is used to confine the growing within a relatively homogenous area and prevent from crossing boundaries. A smaller homogeneity value of a pixel indicates a greater possibility that the pixel is on a boundary. Firstly, a thresholding range is calculated from a small number of user-defined seeds. Then, the AdaptRG algorithm examines the neighborhoods of pixels currently in the region. The neighboring pixels that have intensities within the thresholding range and homogeneity values greater than a predefined parameter are added to the growing region. This process is iterated in the same manner for the whole image until no other pixels satisfy the growing condition. The threshold value is adaptively adjusted according to the pixels in the growing region. We applied our AdaptRG algorithm to segment a liver from an abdominal CT scan. The experimental results show an accurate image of the segmented liver. This method can also be applied to segment other organs such as the kidneys, spleen, or heart. Future research will include the segmentation of individual organs without user-defined seeds.
References: Adams R, Bischof L. (1994) Seeded Region Growing, IEEE Transactions on Pattern Analysis and Machine Intelligence. 16(6):641-647.
Kamdi S, Krishna RK. (2012) Image Segmentation and Region Growing Algorithm, International Journal of Computer Technology and Electronics Engineering (IJCTEE), 2(1): 103-107
Funder Acknowledgement(s): This study was supported by a grant from the Department of Defense awarded to Xiangyan Zeng PhD, Professor of Computer Science at Fort Valley State University.
Faculty Advisor: Xiangyan Zeng, zengx@fvsu.edu
Role: For my part, I proposed the the algorithm as well as implemented the methods used and summarized the results.