Impact of slab thickness in digital breast tomosynthesis on radiologists' diagnostic accuracy and interpretation time in women with breast cancer: a single-center retrospective study.

Background: Digital breast tomosynthesis (DBT) improves breast cancer detection compared to digital mammography but requires longer reading times. DBT slices are reconstructed and combined into thicker slabs. This study aimed to compare the diagnostic performance and interpretation time between a 6-mm thick slab and a 1-mm thin slice procedure in DBT, both using synthetic mammograms. Additionally, this study aimed to evaluate intra- and inter-observer agreement on image visibility. Methods: This retrospective study evaluated DBT performed on symptomatic women between September 2020 and June 2021. Two breast-certified radiologists independently interpreted DBT images from both procedures, with a 4-week washout period between interpretations. Of the 184 eligible women included, 97 had biopsy-proven malignancies, which consisted of 53 masses, 23 calcifications, 20 distortions, and 1 asymmetry. A Wilcoxon signed-rank test was used to compare reading times between each procedure. Sensitivity and specificity were evaluated using McNemar's test, while area under the receiver operating characteristic curve (AUC) was compared using the DeLong test. The non-inferiority of thick slab compared to thin slice of DBT was assessed using a one-sided confidence interval (CI). Observer agreement was measured using Cohen's kappa. Results: Sensitivity for the thick slab procedure of DBT was 91.8% (95% CI: 87.4-96.4%), while for thin slice procedure it was 95.9% (95% CI: 89.8-98.9%). The difference between two procedures was not statistically significant (P=0.39). Specificity for thick slab was 65.5% (95% CI: 58.6-75.4%), while for thin slice procedure it was 67.8% (95% CI: 56.9-77.4%). The difference between two procedures was not statistically significant (P=0.73). The AUC for thick slab was 0.79 (95% CI: 0.73-0.84), while for thin slice procedure it was 0.82 (95% CI: 0.77-0.87). No statistically significant difference between the two procedures was found (P=0.18). The mean interpretation time for thick slab was 11.2% (0.42 minutes) shorter compared to the thin slice procedure. The intra-rater agreement on image visibility between two procedures was 0.75 (95% CI: 0.65-0.81). Inter-rater agreement was 0.86 (95% CI: 0.84-0.87) for thick slab and 0.60 (95% CI: 0.59-0.61) for thin slice procedure. Conclusions: In this symptomatic patient cohort, the use of 6-mm thick slabs of DBT effectively reduced interpretation time without compromising diagnostic performance, demonstrating non-inferiority in sensitivity, specificity, and AUC for breast cancer detection compared to 1-mm slice standard DBT. Given the limited cohort size, further validation in a broader screening or asymptomatic population would be beneficial.