Learning curve of digital breast tomosynthesis (DBT) reading time: evidence from the UK National PROSPECTS DBT Screening Trial

Purpose: A barrier to the uptake of Digital Breast Tomosynthesis (DBT) in screening is its prolonged reading time compared to 2D digital mammography (2DDM). We investigate how DBT reading times speed up after an extended period of implementation in routine breast screening and identify a possible learning curve.

Methods: The PROSPECTS Trial is comparing DBT to 2DDM in the NHS Breast Screening Programme (NHSBSP), with recruitment taking place between 2019 and 2025. All trial screening cases were double-read, and readers recorded their reading times. For the present analysis, reading time data from one trial site, with readers reading a total of 726 screening batches, equating to 6873 2DDM and 8535 DBT exams over a four-year period were included. Reading times were grouped into 6-month intervals and trends were analysed using Kruskal–Wallis tests with post-hoc Dunn’s tests to identify significant changes over time.

Results: Significant reductions in reading times were observed for both DBT and 2DDM over the four-year period (ps < 0.001). Post-hoc analyses revealed substantial reductions in reading time after two years of reading for both modalities (ps < 0.001). The median DBT reading time per case decreased from 2.14 min (IQR: 1.75–2.47) in the first two years to 1.17 min (IQR: 0.96–1.51) in years 3–4 (p < 0.001). Similarly, 2DDM reading times declined from a median of 0.98 min (IQR: 0.80–1.21) in years 1–2 to 0.68 min (IQR: 0.50–0.84) in years 3–4 (p < 0.001).

Conclusion: A learning curve effect was observed, with significant reductions in DBT reading times over a four-year period.

(British society of breast radiology annual scientific meeting abstracts - abstract O1. BSBR Annual Scientific Meeting 2025 Brighton, UK 9–11 November 2025 https://www.delegate-reg.co.uk/bsbr2025.)