Evaluation of breast masses by sonoelastography

The purposes of our study were to evalute the diagnostic utilty of scoringmethods in sonoelastography in differentiating benign from malignant breastmasses, compare it with BI-RADS and to depict whethever the combination ofstrain index measurement method with elastography color-coded scoring methodmade this technology more efficient.611 breast lesions (511 solid (404 malign,107 benign) and 100 cystic) in400 patients (range 16-83years, 2 male and 308 female) masses wereprospectively included in this study. After mammography and US, finalBI-RADS categories were assigned. Elasticity images were classified by İtalianscoring method and Ueno-Itoh scoring method. The strain index (fat to lesionstrain ratio) of each solid lesion was calculated. Cystic lesions were examinedwhetever typical three-layered appearance was positive. These findings werecompared with cyto/histopathologic findings and at least 2 years follow-up. Thesensitivity, specificity, positive predictive values and negative predictive valueswere calculated for each modality. Diagnostic performance of strain index, İtalianand Ueno-Itoh scoring methods were compared by ROC curve analysis.Intraobserver and interobserver variabilty were analyzed for elastography colorcodedscoring methods. Diagnostic performance of each method were analyzedaccording to the maximum size and the depth of the lesions and the breastparankimal density. The distribution of elasticity scores and strain index valuesfor malignant breast lesions according to their histologic subclassification asinvasive ductal carsinoma and the other malignant pathologies were analyzed.Sensitivity, specificity, positive and negative predictive value were 100%,71,8%, 48,4%, 100%, respectively for BI-RADS, were 72%, 95,3%, 80,2%,92,8%, for Ueno-Itoh scoring method, were 79,4% 90,8%, 69,7%, 94,3%,respectively, for İtalian scoring method and were 89,7%, 85,9 %, 62,7%, 96,9%for the strain index when a cut off point of 3,61 was used. The areas under theROC curves were 0,898 for Ueno-Itoh scoring method, 0,912 for İtalian scoringmethod and 0,914 for the strain index. Typical three-layered apperance wasdepicted in simple cysts, clustered microcysts. The presence of three- layeredapperance was independent from the size and the depth of the cysts, but dependentto the type of the cysts and the internal structure. Intraobserver agreement was% 70 and interobserver agreements was % 67. The best diagnostic performance ofelasticity scoring methods and strain index were in deep (>10mm), large (>20mm)lesions and lipomatous breast density. There were no significant differencesbetween the mean scores and strain index values for invasive ductal carcinomasand the other malignant pathologies.Our results suggest that sonoelastography has the potential to aid diagnosisof breast lesions by increasing the specificity and positive predictive value.Measuring the strain index may be useful as a complementery technique as itshigh sensitivity and negative predictive value.Key Words: Breast lesions, Sonoelastography, BI-RADS