Variations in signal-to-noise characteristics of tissue-equivalent attenuators for mammographic automatic exposure control system performance evaluation.

Abstract

This work investigates the impact of tissue-equivalent attenuator choice on measured signal-to-noise ratio (SNR) for automatic exposure control (AEC) performance evaluation in digital mammography. It also investigates how the SNR changes for each material when used to evaluate AEC performance across different mammography systems. AEC performance was evaluated for four mammography systems using seven attenuator sets at two thicknesses (4 and 8 cm). All systems were evaluated in 2D imaging mode, and one system was evaluated in digital breast tomosynthesis (DBT) mode. The methodology followed the 2018 ACR digital mammography quality control (DMQC) manual. Each system-attenuator-thickness combination was evaluated using For Processing images in ImageJ with standard ROI size and location. The closest annual physicist testing results were used to explore the impact of varying measured AEC performance on image quality. The measured SNR varied by 44%-54% within each system across all attenuators at 4 cm thickness in 2D mode. The variation appeared to be largely due to changes in measured noise, with variations of 46%-67% within each system across all attenuators at 4 cm thickness in 2D mode. Two systems had failing SNR levels for two of the materials using the minimum SNR criterion specified in the ACR DMQC manual. Similar trends were seen in DBT mode and at 8 cm thickness. Within each material, there was 115%-131% variation at 4 cm and 82%-114% variation at 8 cm in the measured SNR across the four imaging systems. Variation in SNR did not correlate with system operating level based on visual image quality and average glandular dose (AGD). Choice of tissue-equivalent attenuator for AEC performance evaluation affects measured SNR values. Depending on the material, the difference may be enough to result in failure following the longitudinal and absolute thresholds specified in the ACR DMQC manual.