Breast cancer CADx based on BI-RADS™ descriptors from two mammographic views

In this study we compared the performance of computer aided diagnosis (CAD(x)) algorithms based on Breast Imaging Reporting And Data System (BI-RADS(TM)) descriptors from one or two views. To select cases for the study with different mediolateral (MLO) and craniocaudal (CC) view descriptors, we assessed the agreement in BI-RADS(TM) lesion descriptors, BI-RADS(TM) assessment, and subtlety ratings for 1626 cases from the Digital Database for Screening Mammogrpahy (DDSM) using kappa statistics. We used 115 mass cases with different descriptors for the two views to design linear discriminant analysis (LDA) based CAD(x) algorithms. The CAD(x) algorithms used BI-RADS(TM) descriptors and patient age as features. The algorithms based on BI-RADS(TM) descriptors from both the views performed marginally better than algorithms based on BI-RADS(TM) descriptors from a single view. A system that averaged the results of two classifiers trained separately on the MLO and CC views displayed the best performance (A(z)=0.920 +/- 0.027). Thus, some improvement in performance of BI-RADS(TM) based CAD(x) algorithms may be achieved by combining information from two marnmographic views. (C) 2006 American Association of Physicists in Medicine.