I. LAURENCE AROQUIARAJ AND K. THANGAVEL: MAMMOGRAM IMAGE SEGMENTATION USING AUTO ADAPTIVE FUZZY INDEX MEASURE 274 MAMMOGRAM IMAGE SEGMENTATION USING AUTO ADAPTIVE FUZZY INDEX MEASURE I. Laurence Aroquiaraj 1 and K. Thangavel 2 Department of Computer Science, Periyar University, Tamil Nadu, India E-mail: 1 laurence.raj@gmail.com, 2 drktvelu@yahoo.com Abstract Breast Cancer involves the uncontrolled growth of abnormal cells that have mutated from normal tissues. A radiologist looks for certain signs and characteristics indicative of cancer when evaluating a mammogram. The main task is to obtain the locations of suspicious regions to assist radiologists in diagnosis. Image segmentation has been approached from a wide variety of perspectives: region-based approach, morphological operation, multi-scale analysis, fuzzy approaches and stochastic approaches have been used for mammogram image segmentation but with some limitations. In spite of the several methods available in the literature, image segmentation still a challenging problem in most of image processing applications. The challenge comes from the fuzziness of image objects and the overlapping of the different regions. In this paper we propose fast auto adaptive image segmentation algorithm for finding the optimal thresholds for segmenting gray scale images. The proposed method is based on fuzzy index which decreases the similarity between pixels increases. The system uses initial estimation of the parameters. The fuzzy subsets derived from the image histogram using weighted fuzzy entropywill shows the similar cost measure as in pixels of the same subset. Experimental results demonstrate the effectiveness of the proposed approach. Keywords: X-ray Mammography, Fuzzy Entropy, Ostu multi-level Method, Segmentation 1. INTRODUCTION Breast cancer has been one of the major causes of death among women since the last decades and it has become an emergency for the healthcare systems of industrialized countries. This disease became a commonest cancer among women. If the cancer can be detected early, the options of treatment and the chances of total recovery will increase. Intra-operative diagnosis of the disease has steadily become more important with respect to the recent introduction of sentinel lymph node biopsy. The term benign refers to a condition, tumor or growth that is not cancerous. This means that it does not spread to other parts of the body or invade and destroy nearby tissue. Benign tumors usually grow slowly. In general, benign tumor or condition is not harmful. Breast cancer, also known as carcinoma, is a malignant growth that begins in the tissues of the breast. Image preprocessing and enhancement methods help to improve the visual appearance of mammogram medical images such as removal of film artifacts and labels, filtering the image, normalization and removal of pectoral muscle region. After image acquisition, the first one aims to segment the background and annotations from the whole breast area, while the second one involves separating the pectoral muscle (when present) from the rest of the breast area. For segmenting the breast from the pectoral muscle a new histogram of this biggest region is used. This histogram contains two zones: the pectoral muscle and the breast tissue. A region growing algorithm [14] is used to extract the pectoral muscle region from the breast. The seed of this region growing is placed inside the pectoral with value between the brightness maximum and the minimum between the two zones of the histogram. The last step is the use of morphological operations in order to smooth the boundary of the breast. This biggest region can be extracted using a Connected Component Labeling algorithm. A good survey of both breast and pectoral segmentation types can be founded in [4]. Image segmentation is referred to as the procedure in which the input image is divided into meaningful regions in such a way that the output image will consist of a set of labeled region describing the input image. The output image will contain a set of non-overlapping objects representing pixels of similar gray values [7]. Image segmentation is a crucial step in a wide range of medical image processing systems. It is useful in visualization of the different objects present in the image. Numerous segmentation algorithms have been proposed and surveys of these techniques can be found in [12]. Image segmentation techniques can be categorized into three approaches. The first category uses clustering techniques such as adaptive fuzzy C means and K-means [6, 10, 18]. In the clustering techniques each pixels in the image is assigned as a class according to its features. The second category uses algorithms based on histogram thresholding [5, 18]. Histogram based methods work well for images which are can be clearly separated into two regions but fail there is no significant contrast between the objects and background. The third Category uses iterative approaches to achieve pixels separation [13]. Fuzzy entropy has been used for image segmentation [7, 9]. Most of the image segmentation algorithms produce binary image, or “foreground and background”. While these results are acceptable in some image processing applications such as document processing and Optical Character Recognition systems, they are not satisfactory in medical images where several features, which are present in the image, need to be detected. In modem orthodontic practice, a great reliance is placed on objective and systematic methods of characterizing craniofacial forms, using measurements based on a set of agreed upon points know as craniofacial landmarks. When the X-ray images have been acquired, certain points (anatomical landmarks) on the X-ray mammography image have to be located in order to determine the proper breast treatment or the effect of previous treatment. Distance and angles among these landmarks are compared with normative values to diagnose patient’s deviations from ideal form, evaluate the craniofacial growth and measure the effect of treatment. Without accuracy in land marking, it is impossible to determine craniofacial parameters correctly. This process is carried out manually and consisted of two steps: producing cephalometric tracing then they try to locate the anatomical