The use of joint two-view information for improving computerized microcalcification detection accuracy
We investigated the use of joint CC and MLO view information to improve computerized detection of microcalcifications. Microcalcification cluster candidates were located using our previously developed single-view detection algorithm. Object pairs from the CC and MLO views were formed by using their radial distances from the nipple. The pairs were classified using a correspondence classifier, which analyzed feature similarity measures of the members in a pair. The scores from the correspondence classifier were fused with the single-view detection scores. The algorithm was trained using 109 mammogram pairs from the University of Michigan, and tested using an independent public data set from the University of South Florida (USF). The test set included 124 two-view mammograms containing 272 microcalcification clusters. The false-positive (FP) rate was determined by applying the algorithm to 89 normal mammogram pairs from the USF data set. The two-view detection algorithm had a film-based sensitivity of 80% at 0.22 FPs/image. At the same sensitivity, the single-view detection algorithm produced 0.33 FPs/image. Our results indicate that the use of joint two-view information can improve the accuracy of our computerized microcalcification detection algorithm.
KeywordsLinear Discriminant Analysis Digital Mammography Annular Region Object Pair Correspondence Classifier
Unable to display preview. Download preview PDF.
- 6.Chan, H.P., Doi, K., Vyborny, C.J., Schmidt, R.A., Metz, C.E., Lam, K.L., Ogura, T., Wu, Y., and MacMahon, H.: Improvement in radiologists’ detection of clustered microcalcifications on mammograms. The potential of computer-aided diagnosis. Investigative Radiology, 1990, 25: 1102–1110.PubMedCrossRefGoogle Scholar
- 7.Heath, M., Bowyer, K., Kopans, D., Moore, R., and Kegelmeyer, P.: The digital database for screening mammography, in Digital Mammography; IWDM 2000, Yaffe, M.J., Editor. 2001, Medical Physics Publishing: Toronto, Canada, p. 457–460.Google Scholar