Increasing efficiency in the ACR Digital Mammography QC Manual's DBT Z-resolution evaluation

The Z-resolution evaluation in the ACR Digital Mammography (DM) QC manual measures a pseudo full width half max (pFWHM) from speck group signals through slices of the phantom for comparison to a baseline value.

The evaluation requires measurement of seven ROIs on up to nine slices, for a potential of up to 63 ROIs. This work evaluates whether measuring fewer ROIs would produce similar results.

Z-resolution data from 468 surveys from 147 mammography units were retrospectively analyzed. The pFWHM was calculated using the 6-speck mean (pFWHM₆) as described in the ACR manual, as well as using each single speck (pFWHM₁), and all combinations of a 2-speck mean (pFWHM₂) using the center and each peripheral speck. Deviation from the respective initial measured pFWHM_initial was determined for all units with at least two surveys.

The pFWHM₁ had an average absolute error of 6.0% relative to pFWHM₆ across all surveys, compared to 3.9% for pFWHM₂. For the 111 units with longitudinal data, the mean absolute deviation from pFWHM_initial was 7.8% for pFWHM₁, compared to 5.6% for pFWHM₂, and 5.1% for pFWHM₆. For longitudinal testing, pFWHM₆ produced two failures that were related to major component replacement. Both pFWHM₁ and pFWHM₂ methods caught those failures for a false negative rate of 0%. However, pFWHM₁ had a false positive rate of 12.9%, while pFWHM₂'s was only 2.8%. No positive test from any method was attributable to actual equipment failure.

A single-speck pFWHM for ACR DM QC resulted in high variability and frequent unit failures for the Z-resolution evaluation. However, a 2-speck pFWHM produced results clinically equivalent to the 6-speck method. Using the 2-speck method as an alternative methodology for Z-resolution evaluation achieves similar results with 57% fewer measurements, increasing the efficiency of testing.