Optimization of microcalcification cluster detection in wide-angle flying focal spot digital breast tomosynthesis and...

To investigate the impact of angular range and dose distribution on microcalcification cluster detection in digital breast tomosynthesis (DBT) and synthetic mammography (SM) of a clinical wide-angle DBT system with flying focal spot (FFS), through virtual imaging techniques.

DBT projection sets were acquired from ten patients at twice the automatic exposure controlled (AEC) dose. Noise was added to each projection to create four projection sets: (1) 25 projections at AEC dose ("AEC"), (2) 25 projections with a convex dose distribution ("convex"), (3) 25 projections with increased dose in the central three projections ("peak 3"), and (4) 19 projections covering a 40° angular range instead of the standard 50° ("40 degrees"). Total scan dose and angular spacing were maintained constant across all setups. Microcalcification clusters were simulated within these projection sets, followed by a human observer detection study with DBT and SM patches. Performance was analysed using jackknife-alternative free-response receiver operating characteristic (JAFROC) analysis.

For DBT, the area under the curve (AUC) was 0.89 ± 0.03 (AEC), 0.91 ± 0.03 (convex), 0.88 ± 0.03 (peak 3), and 0.91 ± 0.04 (40 degrees). For SM, AUC values were lower: 0.73 ± 0.04, 0.74 ± 0.02, 0.73 ± 0.04, and 0.74 ± 0.02, respectively. No significant improvements were observed compared to the AEC setup for either modality.

Increasing the dose to the central projections or reducing the angular range from 50° to 40° did not significantly affect calcification detection in DBT or SM compared to the standard AEC setup for an FFS system.