Article Preview
TopIntroduction
Breast cancer is the second leading cancer in worldwide and most common among the women. In 2018, over 2 million new cases were identified by American Institute for Cancer Research (AICR) (Breast cancer statistics). In 2017, American Cancer Society estimated a 2017-2018 years figure as 252,710 new cases were diagnosed among women and approximately 40410 women and 460 men are died because of breast cancer (Breast Cancer Facts & Figures,2017-2018). Some years back in Indian woman, cervical cancer was most occurring cancer but now in the recent years, breast cancer is leading (Malvia and Appalaraju, 2017).
An abnormal cell growth in the breast causes breast cancer. Breast cancer occurs in the breast cells of milk glands which is also known as lobules or in the ducts of lobules. Breast cancer shows no major symptoms in early stage. When the tumor is large it may spread till the under-arm lymph node which causes a swelling of lymph node. Common symptoms of breast cancer are swelling in breast, redness, thickening of breast, pain in breast, nipple abnormalities and nipple discharge (Breast Cancer Facts & Figures, 2018). Early detection of breast cancer is very essential step in breast cancer diagnosis and treatment. Out of all screening techniques, X-ray mammography is gold standard technique. Breast mammography uses low energy X-rays for imaging.
Computer Aided Diagnosis (CAD) is a tool used for better analysis of medical images and helps to reduce the manual error. Mammogram images can be analyzed using computerized CAD system which improves the accuracy of the diagnostic method and reduces the workload and unnecessary breast biopsies (Ali et al., 2015). On the other hand, developing a CAD system for breast mass detection is a challenging task because of varying breast density, size, shape etc. CAD system consists of many stages like Data acquisition, Image Preprocessing, Image segmentation, Feature Extraction and Classification. The image obtained consists of noise and artifacts. The preprocessing technique is used to remove the noise and to enhance the raw images obtained. The Medio Lateral Oblique (MLO) view contains pectoral muscle as shown in Figure 1, which contains the lymph nodes indicates the abnormality in the cells. For CAD system, it acts as an artifact because reduces the performance accuracy. Hence it is important to remove the pectoral muscle using image processing techniques. Some important features are extracted from the pectoral removed breast region to classify the images as normal and abnormal.
Proposed study focuses on the semi-automated pectoral muscle removal technique and analysis on the different feature sets of the GLCM texture features. The proposed scheme is as shown in the Figure 2. To classify the MLO views of mammogram images as normal and abnormal an efficient algorithm is developed. The main stages of the algorithm are as follows:
Figure 2. Block diagram of proposed method