Hybrid Technique Based on N-GRAM and Neural Networks for Classification of Mammographic Images

摘要

Various texture, shape, boundary features have been used previously to classify regions ofinterest in radiological mammograms into normal and abnormal categories. Although, bag-ofphrasesor n-gram model has been effective in text representation for classification or retrievalof text, these approaches have not been widely explored for medical image processing. Ourpurpose is to represent regions of interest using an n-gram model, then deploy the n-gramfeatures into a back-propagation trained neural network for classifying regions of interest intonormal and abnormal categories. Experiments on the benchmark miniMIAS database show thatthe n-gram features can be effectively used for classification of mammograms into normal andabnormal categories in this way. Very promising results were obtained on fatty backgroundtissue with 83.33% classification accuracy.