Skeletal muscle image analysis based on fuzzy logic methods

number: 
835
إنجليزية
Degree: 
Author: 
Shaima Ibraheem Jabbar Al-Dhifery
Supervisor: 
Dr. Nasser N. Khamiss
Dr. Fakher Al-Ani
year: 
2004
Abstract:

Skeletal muscle image analysis represents as the one of the important steps in understanding and classification of muscle fibers. The classification of muscle fiber has clinical significance for diagnosis of muscle diseases. In this thesis, a software system is developed using a fuzzy set techniques to produce effective artificail intellegence system Skeletal muscle Image Analysis System (SIAS). The (SIAS) consists of two phases: the first phase denotes into the preprocessing steps (enhancement and), while the second phase which includes image description to classify the muscle fibers. The first phase consists of two main parts, the first part capable of improving the colored visual appearance of muscle fibers image, while the second part achieved color muscle image segmentation to isolate three types of muscle fibers through using multi-level thresholding method. Image enhancement is perfomed by computer through extraction pixels fuzzy properties. Image enhancement in the fuzzy property domain can be achieved using Contrast Intensification (CI) algorithm . In manner of segmentation process, a novel method of segmentation process is implemented based on the linear conditional probability and maximum value of entropy to reach into desirable threshold values. This method is called Multi-Level Thresholding (MLT)method. At the second phase of (SIAS) employed a successful approach for region description to analyze the skeletal muscle image and classify muscle fibers. This description is based on set of topological features. These features are used togather to get a high quality of discription. Normally muscle picture contains different muscle fibers that have different parameters like shape, size and color. These muscle fibers from medical point of view are classified into three main types A, B, and C corressponding to the variability in the above mentioned parameters. The obtained results indicate the (SIAS) can provide adaptive image segmentation to separate the different types of muscle fibers and extract the essential information for diagnosis.