Computerized analysis of lesions in US images of the breast

Rationale and Objectives. Breast sonography is not routinely used to distin guish benign from malignant solid masses because of considerable overlap in their sonographic appearances. The purpose of this study was to :investiga te the computerized analyses of breast lesions in ultrasonographic (US) ima ges in order to ultimately aid in the task of discriminating between malign ant and benign lesions. Materials and Methods. Features related to lesion margin, shape, homogeneit y (texture), and posterior acoustic attenuation pattern in US images of the breast were extracted and calculated. The study database contained 184 dig itized US images from 58 patients with 78 lesions. Benign lesions were conf irmed at biopsy or cyst aspiration or with image interpretation alone; mali gnant lesions were confirmed at biopsy. Performance of the various individu al features and output from linear discriminant analysis in distinguishing benign from malignant lesions was studied by using receiver operating chara cteristic (ROC) analysis.;, Results. At ROC analysis, the feature characterizing the margin yielded A v alues (area under the ROC curve) of 0.85 and 0.75 in distinguishing between benign and malignant lesions for the entire database and for an "equivocal " database, respectively. The equivocal database contained lesions that had been proved to be benign or malignant at . cyst aspiration or biopsy. Line ar discriminant analysis round-robin runs yielded A values of 0.94 and 0.87 in distinguishing benign from malignant lesions for the entire database an d for the equivocal database, respectively. Conclusion. Computerized analysis of US images has the potential to increas e the specificity of breast sonography.