Background: Diagnostic performance in breast screening programs may be influenced by the prior probability of\r\ndisease. Since breast cancer incidence is roughly half a percent in the general population there is a large probability\r\nthat the screening exam will be normal. That factor may contribute to false negatives. Screening programs typically\r\nexhibit about 83% sensitivity and 91% specificity. This investigation was undertaken to determine if a system could\r\nbe developed to pre-sort screening-images into normal and suspicious bins based on their likelihood to contain\r\ndisease. Wavelets were investigated as a method to parse the image data, potentially removing confounding\r\ninformation. The development of a classification system based on features extracted from wavelet transformed\r\nmammograms is reported.\r\nMethods: In the multi-step procedure images were processed using 2D discrete wavelet transforms to create a set\r\nof maps at different size scales. Next, statistical features were computed from each map, and a subset of these\r\nfeatures was the input for a concerted-effort set of na�¯ve Bayesian classifiers. The classifier network was constructed\r\nto calculate the probability that the parent mammography image contained an abnormality. The abnormalities\r\nwere not identified, nor were they regionalized.\r\nThe algorithm was tested on two publicly available databases: the Digital Database for Screening Mammography\r\n(DDSM) and the Mammographic Images Analysis Societyâ��s database (MIAS). These databases contain\r\nradiologist-verified images and feature common abnormalities including: spiculations, masses, geometric\r\ndeformations and fibroid tissues.\r\nResults: The classifier-network designs tested achieved sensitivities and specificities sufficient to be potentially\r\nuseful in a clinical setting. This first series of tests identified networks with 100% sensitivity and up to 79%\r\nspecificity for abnormalities. This performance significantly exceeds the mean sensitivity reported in literature\r\nfor the unaided human expert.\r\nConclusions: Classifiers based on wavelet-derived features proved to be highly sensitive to a range of pathologies,\r\nas a result Type II errors were nearly eliminated. Pre-sorting the images changed the prior probability in the\r\nsorted database from 37% to 74%.
Loading....