Computer-aided detection of breast masses on mammograms: dual system approach with two-view analysis

摘要: The authors previously developed a dual CAD system that merged the decision from twomass detection systems in parallel, one trained with average masses and another subtlemasses, to improve sensitivity without excessively increasing false positives FPs . In this study,they further designed two-view fusion method combine information different mam-mographic views. Mass candidates detected independently by on two-viewmammograms were first identified as potential pairs based regional registration technique. Asimilarity measure was differentiate TP-TP other TP-FP FP-FPpairs using paired morphological features, Hessian feature, texture features.Atwo-view fusionscore for each object generated weighting similarity cross correlationmeasure of pair. Finally, linear discriminant analysis classifier combinethe mass likelihood score single-view classification FPs. A total 2332 mammograms 735 subjects includ-ing 800 normal 200 collected Institutional ReviewBoard IRB approval.