Table summarizes the characteristics of the distribution of microcalcification

Table 3 summarizes the characteristics of the distribution of microcalcification on MMG and US (Figs. 1-4). Microcalcifications were clustered on MMG and US in 56 and 44 lesions, and scattered or linear in 46 and 40 lesions, respectively. Table 3 indicates that acetylcholine inhibitors clusters with the greatest diameter <5 mm are easily seen on MMG. At ultrasonography, scattered or linear distributions detected on MMG were more visible than clusters on MMG (40/46, 87% vs. 44/56, 78.6%). Assessment of BI-RADS category is outlined in Table 4 (Fig. 3). At MMG, 10 of 114 (9%) DCIS lesions were classified as category 3, 62 of 114 (54%) lesions as category 4 and 42 of 114 (37%) lesions as category 5. At ultrasonography, 10 of 96 (10%) and 9 of 33 (27%) DCIS lesions were classified as category 3 in groups 1 and 2, respectively; category 4 in 55 of 96 (58%) and 21 of 33 (64%) lesions as category 4; and 31 of 96 (32%) and 3 of 33 (9%) lesions as category 5, respectively. Even though the BI-RADS classifications on US and MMG differed statistically (p = 0.000), the median BI-RADS classification was category 4c on both US and MMG (p = 0.01). Table 5 outlines the correlations between tumor size and DCIS-Mi on US and MMG (Figs. 4, 5). The proportions of DCIS lesions with maximum dimension ≤2.5, 2.5–5.0 cm and ≥5.0 cm associated with micro-invasion on US were 39% (22/56), 77% (37/48) and 68% (17/25), respectively, and those on MMG were 42% (22/52), 77% (36/47) and 60% (18/30), respectively. Tumor size was correlated with DCIS-Mi on both US and MMG, but the correlation for evaluation using US is higher than that using MMG (p = 0.001 and 0.024, respectively). The correlations among ultrasonic, mammographic and histopathologic findings of 129 DCIS lesions are summarized in Table 6 (Figs. 1–5). At histopathologic analysis, 53 (41%) lesions were pure DCIS and 76 (59%) lesions were DCIS-Mi. Thirty-five of 129 (27%) DCISs were classified as Van Nuys class 1, 44 of 129 (34%) as Van Nuys class 2 and 50 of 129 (39%) as Van Nuys class 3. Sixty-one of 129 (47%) DCIs proved to be comedo carcinoma, with the remainder (68/129, 53%) non-comedo carcinoma (including 39 cribriform, 22 solid and 7 papillary). With US as the method for the detection of DCIS, diagnostic accuracy was significantly associated with higher Van Nuys groups, the presence of micro invasion and comedo carcinoma (p = 0.000, 0.022 and 0.011, respectively). However, diagnostic accuracy on MMG was found not to associate with higher Van Nuys groups, presence of micro-invasion and comedo carcinoma (p = 0.054, 0.093 and 0.256, respectively).
Discussion
Although most cases of DCIS are diagnosed mammographically, 6%–23% of DCIS lesions are not visible at MMG, especially in patients with dense breasts (Barreau et al. 2005; Holland et al. 1994). Dense breasts, which include the ACR BI-RADS heterogeneously dense and extremely dense categories, are frequent among Chinese women. In our series, patients with dense breasts accounted for 38% (48/127) of cases; 11 of 50 (22%) lesions in the 48 cases of DCIS were not visible on MMG. MMG has been proven to have high sensitivity in the detection of DCIS. However, mammographic sensitivity is significantly inversely related to breast density and decreases from 100% in fatty breasts to 45%–48% in extremely dense breasts (Berg et al. 2004; Kolb et al. 2002). With respect to the 15 DCISs not visible on MMG in our study, 7 lesions manifested as mass-forming malignancies, 4 as indistinct hypo-echoic areas and 4 as duct dilation on ultrasonography. These findings indicate that ultrasonography may be more effective than MMG in the detection of DCIS in patients with dense breasts or lesions without microcalcification. Compared with MMG, US has not traditionally been used for the detection of DCIS because crustaceans is less sensitive in the detection of microcalcification than MMG (Cilotti et al. 1997; Jackson 1995). The smaller the microcalcification is, the lower is the sensitivity of US. Although MMG is superior to ultrasonography in the detection of microcalcification, the advances in US equipment and the refinement of breast imaging techniques have made it possible to detect tiny echogenic spots without acoustic shadowing that correspond to the mammographic findings, especially when they are within a mass lesion or clusters larger than 10 mm and lesions that are neither palpable nor mammographically visible (Hashimoto et al. 2001; Moon et al. 2002; Shin et al. 2008; Yang and Tse 2004). For the 84 DCIS lesions with microcalcification detected by US, 73 (87%) presented as mass-forming malignancies with microcalcification or indistinct hypo-echoic areas with microcalcification, which indicates that a hypo-echoic background would enhance US detection of the bright punctate calcification echoes. Although our findings indicate that US is helpful in detecting DCIS with clustered microcalcifications >10 mm in longest diameter, it is difficult to visualize small clusters of microcalcifications, particularly those less than 5 mm (detection rate = 20%).