DCT based features for the detection of microcalcifications in digital mammograms

In this paper, a set of spectral domain features based on the discrete cosine transform DC Of mammograms are extracted from the X-ray image. The extracted features by the proposed methods are exploited to classify regions of interest ROIs into positive ROIs containing clustered microcalcifications and negative ROIs containing normal tissues. A three-layer bac -propagation neural net or is used as a classifier. The results of the neural net or for the extracted features are evaluated by using a receiver operation characteristics ROC analysis. The proposed technique is shown to be superior to the conventional methods with respect to classification accuracy and computational complexity.