International Journal of Engineering Trends and Technology (IJETT) – Volume 14 Number 5 – Aug 2014 ISSN: 2231-5381 http://www.ijettjournal.org Page 213 Detection of Tumor in Mammograms Using Canny Edge Detection Technique Pratishtha Shrivastava and K.G.Kirar Samrat Ashok Technological Institute Vidisha, M.P. 464001, India Samrat Ashok Technological Institute Vidisha, M.P. 464001, India Abstract -Mammography helps to provide some criteria in order to help the physicians to decide whether a certain tumor is malignant or benign. A mammogram is an x-ray of the breast tissue which is designed such that it can approximately identify the abnormalities. In mammograms selection of suspected area is easy as it looks brighter than its surrounding pixels.The method used for detecting tumor in breast is based on edge based segmentation for which we are using Canny Edge Detection Technique. The work proposed is based on the following procedure: (a)Removing the background information (b)Applying the edge detection technique and retrieving the largest ROI (c)After getting the close loops, filling is performed inorder to highlight the tumor (d) Performing the morphological operations which are erosion and dilation. This method was tested over multiple images and implemented using matlab code. Keywords- Mammogram, Canny Edge Detection, Morphological, ROI. I. INTRODUCTION Mortality rate among women is increasing day by day due to cancer .Generally women around 45 years old are most likely suffering from this disease [1]. At present there are various techniques available for the diagnosis of such type of diseases.. Early detection increases the survival rate whereas delayed diagnosis can reach the patient to an unrecoverable stage and hence results in death. However, current methods of treatment are very effective but only for the early stage of breast cancer . At current large number of diagnostic methods are available, among which mammography is the most reliable method for detecting early breast cancer [2]. The analysis of mammograms include following these steps: (a) Enhancement of preselected features (b) Localization of area which is suspected to be a tumor (c) Classification of the extracted areas into Benign or Malignant tumor [3]. The analysis is difficult due to several reasons. It involves the analysis of small features with low contrast which is superimposed onto non-uniform backgrounds. The mammogram images are scanned first then are digitized for further processing that reduces the difference between background and tumor. In addition the presence of noise ,ducts and glands increase the background variations of tumor area. The boundaries of the suspected area are almost fuzzy and in some cases are partially available. Also the early stage tumors are very small in size. Various image processing techniques are available for enhancement of images, localization of objects and pattern classification. Here, our aim is to just extract the tumor and no classification between benign and malignant tumor is performed. Section II involves the details about the recent work in the field of tumor extraction. Section III involves the techniques being used in this paper to extract the tumor in the breast. Section IV contains the experimental results of the techniques that are described in this paper. Section V shows the comparative results by previous method and proposed method .And the final section VI consists of conclusion and the future work information. II. RECENT WORK We have studied about the various techniques which were employed in order to extract the tumor,subtraction based segmentation is one of them. Image subtraction or pixel subtraction is a process whereby the digital numeric value of one pixel or whole image is subtracted from another image which makes the background disappear leaving only the target. It simply compares the previous frame image with the current one. Image subtraction [5]. In previous work implemented subtraction on two images, the segmented image and the converted RGB image and finally obtained the tumor that is present in the screened mammogram. Although this method provides automatic segmentation but works accurately only with the images having low density. Since this method works on the basis of brightness difference between the original and enhanced image, in addition I tumor extraction,it sometimes also extract the irrelevant brighter part of mammogram that leads to false detection. So in order to improve the accuracy of detection we are using canny edge detector for tumor extraction that considers two characteristics for tumor detection. These are