Optimizing parameters for computer-aided diagnosis of microcalcifications at mammography

Rationale and Objectives. The purpose of this study was to optimize selecti on of the mammographic features most useful in discriminating benign from m alignant clustered microcalcifications. Materials and Methods. The computer-aided diagnosis (CAD) system automatica lly extracted from digitized mammograms 13 quantitative features characteri zing microcalcification clusters. Archival cases (n = 134; patient age rang e, 31-77 years; mean age, 56.8 years) with known histopathologic results (7 9 malignant, 55 benign) were selected. Three radiologists at three faciliti es independently analyzed the microcalcifications by using the CAD system. Stepwise discriminant analysis selected the features best discriminating be nign from malignant microcalcifications. A classification scheme was constr ucted on the basis of these optimized features, and its performance was eva luated by using receiver operating characteristic (ROC) analysis. Results. Six of the 13 variables extracted by the CAD system were selected by stepwise determinant analysis for generating the classification scheme, which yielded an ROC curve with an area (Az) of 0.98, specificity of 83.64% , positive predictive value of 89.53%, and accuracy of 91.79% for 98% sensi tivity. When patient age was an additional variable, the scheme's performan ce improved, but this was not statistically significant (AZ = 0.98). The RO C curve of the classifier (without age as an additional variable) yielded a high A(z) of 0.96 for patients younger than 50 years and an even higher (P <.02) A(z) of 0.99 for those 50 years or older. Conclusion. Stepwise discriminant analysis optimized performance of a class ification scheme for microcalcifications by selecting six optimized feature s. Scheme performance was significantly (P <.02) higher for women 50 years or older, but the addition of patient age as a variable did not produce a s tatistically significant increase in performance.