الفهرس 
Contribution to complex segmentation of medical images
AbstractOnly 14 pages are availabe for public view
 |  | from | 151 |  |  |
| | | from | 151 | | |
Abstract
This thesis aims to present a reliable methodology for complex segmentation in abdominal images for liver. Liver segmentation is very crucial for surgical operations, and hepatitis patient’s follow up. The study found a new methodology for segmentation instead of the traditional segmentation methods. The traditional methods move around edge-based and region-based methodologies. The tested traditional techniques include region growing, k-means, watershed, local thresholding and mean shift. The traditional methods have been improved and achieved better experimental results. But the new methodology has proved a better efficiency. It moves in a new trend of handling such complex situation, depending on bio-inspired algorithms (also called meta-heuristic optimization algorithms) to improve the accuracy of the segmentation process. The used algorithms include Artificial Bee Colony optimizer, Grey Wolf optimizer, Antlion optimizer, Whale optimizer, Moth-fame optimizer, and Dragonfly optimizer. They reduced the need for filters in the preprocessing phase using these algorithms as a clustering technique. Besides, they improved the accuracy of the segmented liver. The optimization techniques use a fitness function to produce a number of predefined clusters. Then, every pixel in the image will be assigned with the number of the cluster which has the least distance to its intensity value. This will produce an initial segmented liver. A statistical image, representing all possible occurrences of liver in abdominal image, is used to improve the performance of the initial segmented liver, which will be enhanced using either region growing or morphological operations. All methods were tested on a dataset of CT or MRI images under the supervision of a specialist in radiology