Classification of Mammogram Images: A Survey

Image segmentation is a technique which has opened a wide field of methods to analyze data, which could be in the form of images, videos. This becomes useful in breast cancer detections as images are used to extract features from the region of interest in the provided mammogram. Mammography tests have been identified as a potential source for pre-detection of a cancerous cell and thus, detection done with the technique of image segmentation has been proven effective in early diagnosis of breast cancer. This paper describes how image preprocessing on mammogram images, followed by image classification using neural networks can be utilized to achieve an efficient diagnosis of breast cancer. The analysis shows that there is still sufficient scope of improving efficiency and accuracy of the classification techniques available. Although it is difficult to assess all the available classification methods, this paper will focus on two famous classification methods, i.e. - probabilistic approach and artificial neural networks. The variety of methods that probabilistic approach provides, and the limitless scope for increasing the accuracy that the artificial neural network will be trained to have, may together be combined to obtain the best possible solution to breast image classification.