Date of Publication :22nd December 2016

Abstract: X-ray screening for breast cancer is an important public health initiative in the management of a leading cause of death for women. However, screening is expensive if mammograms are required to be manually assessed by radiologists. Moreover it is subjected to perception and interpretation errors. Automated mammogram classification is promising however relies on the identification of image features that enhance classification accuracies. Features that represent the shape of a tumour have been found to be useful for breast cancer detection using mammographic images. Existing shape feature computation methods are computationally expensive and strongly dependent upon algorithms that can segment an image to localize a region of interest likely to contain the tumour. In this paper, we apply a novel feature extraction technique called Pixel N-grams inspired from character N-gram model in text categorization for classification of shape images. Experiments on a dataset constructed for the purpose, demonstrate that the Pixel N-gram features achieve promising shape classification results irrespective of the size and location of the shape in an image without segmentation. These features also achieve excellent classification accuracy for images of varying resolution. Further, Pixel N-gram features are computationally less complex to generate paving the way for mammogram classification on low resource computers

Reference :