الفهرس 
Chapter 1: IntroductionOnly 14 pages are availabe for public view
 |  | from | 32 |  |  |
| | | from | 32 | | |
Abstract
Image Processing represents the backbone research area within engineering and computer science specialization. It is promptly growing technologies today, and its
applications founded in various aspects of biomedical fields especially in cancer isease. Breast cancer is considered the fatal one of all cancer types according to recent statistics all over the world. It is the most commonly cancer in women and
the second reason of cancer death between females. About 23% of the total cancer cases in both developing and developed countries. This thesis challenges the breast cancer problem and presents a Computer Aided
Diagnosis (CAD) system to help radiologist in early detect and classify breast tumor. The early detection of breast cancer is an important factor to treat this disease with higher percent of success. Also the early detection can reduce the
mortality rate caused by breast cancer. The proposed CAD system consists of two
main phases; detection phase and classification phase. In the detection phase,
three different schemes are developed to early detect and locate tumors of the
breasts after pre-processing the input mammogram, namely; Modified k-mean
clustering, Template matching, Modified Hough transform. While, in the classification phase, three different strategies are developed to first extract features from the input mammogram image after detecting the tumors. Finally the
classification process was established by using an artificial neural network and support vector machine. The first detection scheme is developed based on k-mean clustering. Where, amodified k-means clustering algorithm is developed for breast image segmentation for the detection of mass lesion. The second detection scheme is based on template-matching procedure. These templates are defined according to
the morphological spectrum (shape) of the tumor masses. Several steps including thresholding, labeling and masking, filtering were suggested to enhance the
tumor’s intensity compared to the surrounding background blood vessels whichappear very similar to tumor in mammogram images. Converting the image into binary one was needed to calculate the properties for all objects in the image.
According to the diameter property of the label image, the 2D Gaussian template will be designed. So a dynamic diameter template according to the specific mammogram image will be created. The third detection scheme is based on modified Hough transform. The main idea in this approach is using the extracted
features from the input image, such as the lines and circles using Hough transform and trying to rebuild the image again using some of these features that represent the contour of the tumor. The first classification scheme is an evolutionary approach based on signaturesdistances from the centroid of the mass to all points located on the boundary of the region of interest (ROI) as a function of a polar angle 4. In this scheme, the signature of a closed boundary is a periodic function, repeating itself on an angular scale of 25. Then encode and describe this closed boundary to arbitrary function through 1-D (radial) Fourier expansion coefficients. The second
classification scheme is performed by calculating angles of the circumference ofthe detected edges and collecting these readings in a vector which describe the
specific mass. The third scheme dependent on the morphologic spectrum of mammographic masses. Malignant tumors had irregular shape percent higher than the benign tumors. In this method the boundary of the tumor will be interpolated by additional pixels to make the boundary smoothen as possible,these needed pixels is proportional with irregularity shape of the tumor, so that the increasing in interpolated pixels meaning the tumor goes toward the
malignant case.The proposed CAD system is implemented using MATLAB programming and tested over several images taken from the Mammogram Image Analysis Society(MIAS) image database. The MIAS offers a regular classification for mammographic studies. The system works faster so that any radiologist can take a clear decision about the appearance of calcifications by visual inspection.