The positive predictive value of detecting carcinomas by breast ultrasound mainly derives from the sonographic assessment criteria and categories applied. Overall, the range is still low relative to mammography.
Kaplan, 2001 [
49] used a two-armed categorization approach, namely a simple subdivision into negative and positive. All positive results were considered to be potentially suspicious. In contrast to the other studies, Kaplan [
49] also deemed cysts positive if they exceeded a size of 1 cm. The author had a low positive predictive value of 2% [
49]. Three studies (Buchberger et al., 2000 [
53]; Kolb et al., 2002 [
48] and Leconte et al., 2003 [
50]) subdivided their findings into three categories, but applied different definitions of these categories. Categories 1, 2 and 3 were called 'normal', 'benign', and 'suspicious' (Kolb et al. [
48]) or 'benign', 'indeterminate' and 'malignant' (Buchberger et al. [
53]). These authors found positive predictive values of biopsy of 10.3% (Kolb et al. [
48]) and 28% (Buchberger et al. [
53]). The values for Leconte's study could not be ascertained [
50]. Crystal et al., 2003 [
52] used four arms of classification, thereby reaching a positive predictive value for malignancy of 20%. Other than Buchberger et al., Crystal et al. included the 'indeterminate' findings in their calculation of suspicious findings. Only one study used the five categories proscribed by the BI-RADS classification for breast ultrasound scans (Corsetti et al., 2008 [
51]). In this study, however, the positive predictive value could not be determined.