This poster is published under an
open license. Please read the
disclaimer for further details.
Keywords:
Breast, Mammography, Comparative studies, Diagnostic procedure, Cancer
Authors:
P. Clauser, S. Pullini, A. Linda, V. Londero, L. Tomkova, M. Bazzocchi, C. Zuiani; Udine/IT
DOI:
10.1594/ecr2015/C-1232
Methods and materials
In this retrospective study 127 consecutive patients (mean age: 56 years,
range 50-74 years) that underwent both DM and DBT with 2D-CP,
were included.
DBT parameters were:
• W-target X-ray source |
• Rh/Ag filter |
• a-Se digital detector (ANRAD,
LMAM) |
• Angular range: ±20°; 13 projections |
• Reconstructed voxel: 0.085×0.085x1.0 mm |
• Sensitive area: 24x30 cm2 |
• Iterative reconstruction with Total Variation (TV) regularisation |
AEC dose: 1 DM view x 1.6; central dose responsible for 1/3 of the total dose delivered.
Reasons to perform the examinations were characterization of suspicious findings detected on screening mammography or staging of a known breast cancer.
Two patients were escluded because of significant artifacts in the DBT images.
Two experienced radiologists,
blinded to final diagnosis,
evaluated the images in separate reading sessions.
Readings were divided into two sessions,
separated by more than 4 weeks.
1st session: DM alone (Set-A)
DM + DBT (Set-B)
2nd session: 2D-CP alone (Set-C)
2D-CP + DBT (Set-D)
Breast density was assessed with both modalities and cases were assessed as negative (BIRADS ® 1,2 or 3) or positive (BIRADS ® 4 or 5) [9].
The analysis was conducted on a perbreast basis and only one finding per breast was classified.
When more suspicious findings were present,
the smaller was considered.
Standard of reference was histology or >1 year follow up.
ROC curves were calculated and compared.
Interreader agreement was assessed using weighted-kappa statistic.