الفهرس 
AbstractOnly 14 pages are availabe for public view
 |  | from | 206 |  |  |
| | | from | 206 | | |
Abstract
The proposed liver CAD system was tested and evaluated on difficult data sets. We developed a new approaches based on neutrosophic logic, bio-inspired and evolutionary algorithms and provided both effective and efficient improvements to existing algorithms to segment and diagnosis liver tumors from abdominal CT scan. The overall accuracy obtained from hybrid AT-WRG approach for liver tumors segmentation is 90%. To increase the performance, Neutrosophic Sets (NS) is applied to handle uncertainty, decrease indeterminacy and transfer abdominal CT from image domain to neutrosophic domain based on T,I,F. The accuracy obtained from hybrid NSFCM and WNS-FFCM approaches based on neutrosophic sets are 94% and 95% respectively. The results offered by NS are accurate, less sensitive to noise and effective to extract liver tumors from CT scan. To increase the accuracy of segmentation results, the powerful swarming optimization model was used to optimize fast FCM for the best cluster (NS-PSOFFCM). The FFCM was applied based on histogram of the image intensities and PSO algorithm is used to guide the FFCM to find optimal solution in short CPU processing time. The accuracy obtained is 97% and the false positive regions are largely reduced. Different combinatorial set of feature extraction is obtained from different methods in order to keep and achieve optimal accuracy. A new subset feature selection algorithm based on Bio-inspired Social Spider Optimization algorithm (SSOA) was proposed to select subset of relevant features and eliminate irrelevant ones. In comparison with other feature selection algorithms applied in this thesis. The best optimal features are extracted from SSOA with high accuracy and less time consuming. The over-all accuracy obtained is 99.27%. from the results, SSOA shows a good balance between exploration and exploitation and the results in high local minima avoidance. Different computer vision techniques are used and physician visual judge are considered to evaluate the proposed system.