Comparison of synthetic mammography, reconstructed from digital breast tomosynthesis, and digital mammography: evaluation of lesion conspicuity and BI-RADS assessment categories

Giovanna Mariscotti, Manuela Durando, Nehmat Houssami, Mirella Fasciano, Alberto Tagliafico, Davide Bosco, Cristina Casella, Camilla Bogetti, Laura Bergamasco, Paolo Fonio, Giovanni Gandini

Research output: Contribution to journalArticlepeer-review

Abstract

Purpose: To compare the interpretive performance of synthetic mammography (SM), reconstructed from digital breast tomosynthesis (DBT), and full-field digital mammography (FFDM) in a diagnostic setting, covering different conditions of breast density and mammographic signs. Methods: A retrospective analysis was conducted on 231 patients, who underwent FFDM and DBT (from which SM images were reconstructed) between September 2014–September 2015. The study included 250 suspicious breast lesions, all biopsy proven: 148 (59.2%) malignant and 13 (5.2%) high-risk lesions were confirmed by surgery, 89 (35.6%) benign lesions had radiological follow-up. Two breast radiologists, blinded to histology, independently reviewed all cases. Readings were performed with SM alone, then with FFDM, collecting data on: probability of malignancy for each finding, lesion conspicuity, mammographic features and dimensions of detected lesions. Results: Agreement between readers was good for BI-RADS classification (Cohen’s k-coefficient = 0.93 ± 0.02) and for lesion dimension (Wilcoxon’s p = 0.76). Visibility scores assigned to SM and FFDM for each lesion were similar for non-dense and dense breasts, however, there were significant differences (p = 0.0009) in distribution of mammographic features subgroups. SM and FFDM had similar sensitivities in non-dense (respectively 94 vs. 91%) and dense breasts (88 vs. 80%) and for all mammographic signs (93 vs. 87% for asymmetric densities, 96 vs. 75% for distortion, 92 vs. 85% for microcalcifications, and both 94% for masses). Based on all data, there was a significant difference in sensitivity for SM (92%) vs. FFDM (87%), p = 0.02, whereas the two modalities yielded similar results for specificity (SM: 60%, FFDM: 62%, p = 0.21). Conclusions: SM alone showed similar interpretive performance to FFDM, confirming its potential role as an alternative to FFDM in women having tomosynthesis, with the added advantage of halving the patient’s dose exposure.

Original languageEnglish
Pages (from-to)765-773
Number of pages9
JournalBreast Cancer Research and Treatment
Volume166
Issue number3
DOIs
Publication statusPublished - 2017

Keywords

  • BI-RADS category assessment
  • Digital breast tomosynthesis
  • Full-field digital mammography
  • Synthetic mammography

ASJC Scopus subject areas

  • Oncology
  • Cancer Research

Fingerprint Dive into the research topics of 'Comparison of synthetic mammography, reconstructed from digital breast tomosynthesis, and digital mammography: evaluation of lesion conspicuity and BI-RADS assessment categories'. Together they form a unique fingerprint.

Cite this