Keywords:
Breast, Mammography, Experimental, Comparative studies, Experimental investigations, Technical aspects
Authors:
R. Girometti1, C. Zuiani1, A. Taibi2, S. Vecchio3, R. Fazzin1, M. Bazzocchi1; 1Udine/IT, 2Ferrara/IT, 3Pontecchio Marconi/IT
DOI:
10.1594/ecr2012/C-1390
Results
Final diagnosis
- A total of twenty-seven lesions on 21 specimens were confirmed at pathologic examination.
Four specimens showed 2 lesions each,
whereas 1 specimen showed 3 lesions.
The remaining specimen from the patient with CUP syndrome was excluded from analysis,
since it showed no lesions at pathological examination.
- An overview of lesions types and appearance on DM and/or DBT is shown in Table 1.
Diagnostic yield and agreement in measuring lesions size
- DBT detected 27/27 lesions,
corresponding to a DY of 100% (95% C.I.
87.2-100),
whereas DM detected 24/27 lesions,
corresponding to a DY of 88.8% (95% C.I.
70.8-97.6).
Lesions missed by DM were: 3 IDC with intraductal component appearing as a mass (Fig. 2 ),
a mixed lesion and a cluster of microcalcifications (Fig. 3 ).
- Largest lesions diameters as measured by DBT or DM on 24/27 lesions visible with both techniques are reported in Table 2.
According to the Bland and Altman analysis,
average difference between DBT and DM measurements was 0.4 mm,
with 95% limits of agreement of -4.9 to +5.8 mm ().
ICC was 0.92.
Lesions missed by DM ranged 3-9 mm in diameter.
Comparison of subjective scores
- No significant difference (p>0.5) was found in detectability and image quality scores attributed by readers to DBT and DM (Table 3) (Fig. 5 - Fig. 6 ).
Specimen density was assessed as follows: 8 cases BI-RADS 1-2 and 14 cases BI-RADS 3-4.
Lesions missed by DM occurred in specimens with a BI-RADS density of 2 in one case (ICD with i.s.
component appearing as a mass showed in Fig. 2) and 3-4 in two cases (remaining lesions cited above,
including that showed in Fig. 3).
- There was no agreement (k=0.043) in the margins visibility score as attributed by using DBT and DM (24/27 comparable lesions).
This because readers detected more than 50% of lesions margins in 23/24 lesions (95,8%; 95% C.I.
78.8-99.9) with DBT and in 8/24 patients only with DM (33.4%; 95% C.I.
15.6-55.3) (Fig. 7 ).
Of note,
the margins score for DM cases was assumed to be the highest one of two projections.