Improved mammographic interpretation of masses using computer-aided diagnosis

The aim of this study was to evaluate the effectiveness of computerized ima ge enhancement, to investigate criteria for discriminating benign from mali gnant mammographic findings by computer-aided diagnosis (CAD) and to test t he role of quantitative analysis in improving the accuracy of interpretatio n of mass lesions. Forty sequential mammographically detected mass lesions referred for biopsy were digitized at high resolution for computerized eval uation. A prototype CAD system. which included image enhancement algorithms was used for a better visualization of the lesions, Quantitative features which characterize the spiculation were automatically extracted by the CAD system for a user-defined region of interest (ROI). Reference ranges for ma lignant and benign cases were acquired from data generated by 214 known ret rospective cases. The extracted parameters together with the reference rang es were presented to the radiologist for the analysis of 40 prospective cas es. A pattern recognition scheme based discriminant analysis was trained on the 214 retrospective cases, and applied to the prospective cases. Accurac y of interpretation with and without the CAD system, as well as the perform ance of the pattern recognition scheme, were analyzed using receiver operat ing characteristics (ROC) curves. A significant difference (p < 0.005) was found between features extracted by the CAD system for benign and malignant cases. Specificity of the CAD-assisted diagnosis improved significantly (p < 0.02) from 14% for the conventional assessment to 50 %, and the positive predictive value increased from 0.47 to 0.62 (p < 0.04). The area under th e ROC curve (A(z)) increased significantly (p < 0.001) from 0.66 for the co nventional assessment to 0.81 for the CAD-assisted. analysis. The A(z) for the results of the pattern recognition scheme was higher (0.95). The result s indicate that there is an improved accuracy of diagnosis with use of the mammographic CAD system above that of the unassisted radiologist. Our findi ngs suggest that objective quantitative features extracted from digitized m ammographic findings may help in differentiating between benign and maligna nt masses, and can assist the radiologist in the interpretation of mass les ions.