Early stage detection of cancer in mammogram using statistical feature extraction

Breast cancer is one of the leading causes of mortality among women. Mammogram is the most effective method for early detection of breast cancer. In some cases, it is difficult for radiologists to detect the typical diagnostic signs such as masses and micro calcifications on the mammograms because the images are usually translucent and have low contrast. In our proposed method a filtering and multilevel wavelet decomposition method for noise suppression and enhancement in digital mammographic images is done by taking cancer detected image at initial stage as reference. The features extracted by mean, variance, standard deviation, entropy and mean of absolute deviation is calculated for the enhanced mammographic image and these values are taken as reference values. The mammographic image which is under test is taken and the process such as filtering, multilevel wavelet enhancement is done and the statistical values are calculated. Finally the process of detecting cancer is done by comparing the reference values and test values and the detection rate is calculated. Thus this method gives a better accuracy because even in the initial stage the cancer can be detected.