Public health is affected by the health and vitality of women who are the pillars of the families. Among the women between 15 to 54 years old, breast cancer is one of the most important causes of mortality. Every 13 minutes, a woman loses her life due to breast cancer. Statistics show that %12.6 of women suffer from breast cancer. Mammography can be considered as one of the most efficient methods of breast cancer diagnosis; however, this method has still its own limitations. The quality of mammography images can be enhanced by using image processing and the disease can be diagnosed more easily by using image feature extraction. Using Hidden Markov Model and wavelet transform method, a new method for the detection of suspicious areas of breast cancer tumors will be proposed in this study. In addition to the detection of tumors, this method can determine the mass percentage to estimate the progression of the disease. In this study, Markov Model with a tree structure was used to extract statistical properties of the wavelet components. The specific capacity of Markov Model to extract information about edges and the regions of protrusions in image tissues increases the accuracy of cancerous areas detection. The present study sought to estimate the appropriate label (clustering) of the pixels of an image under investigation to segment cancerous areas. Therefore, certain joint distributions were assumed for the pixels of a region or a class and then the maximum similarities between different areas of the examined image and joint distributions were examined through the Maximum-Likelihood Estimation method. To evaluate the proposed method and analyze the results, a combination of the MIAS and PADN databases comprising 150 images was used. The results indicated that the proposed method is more accurate than methods which are solely based on wavelet transform. In the new method, the obtained detection rate was %96 while in the wavelet transform method, it has been reported %71.5 indicating the precision of the proposed method.
