Development of a Comprehensive QC Protocol for Digital Breast Tomosynthesis (DBT)
S Vecchio1*, C Brunner2, M Taylor3, M Williams4, R Kaczmarek5, K Chakrabarti6, (1) Food and Drug Administration, Silver Spring, MD, (2) Food and Drug Administration, Silver Spring, MD, (3) Inova Fairfax Hospital, Falls Church, VA, (4) Univ Virginia, Charlottesville, VA, (5) Food and Drug Administration, Silver Spring, MD, (6) CDRH/FDA, Silver Spring, MDSU-C-116-4 Sunday 1:00PM - 1:55PM Room: 116
Purpose: We define a general method for assessing imaging and dosimetric characteristics of tomosynthesis units that can be used under MQSA for acceptance and maintenance tests under 900.12 (e).
Methods: Several QC tests have been selected as criteria for the evaluation, which are: geometric accuracy, flat-field uniformity and noise, in-plane resolution and slice sensitivity, consistency of AEC performance with varied thickness and glandularity and associated radiation dose. Both projection and reconstructed images were analyzed whenever possible. Several commercially available phantoms were used to evaluate these parameters. The acquisitions were performed on Hologic Selenia Dimension units in Combo mode and in a QC mode where all the tomo projections are acquired at zero-degree position.
Results: To our liking, the systems evaluated did not show any geometric distortion, and distance accuracy was found to be independent of object height above the breast support. Flood-field illumination voxel value was less than 3% within the entire reconstructed volume. The FWHM of the Slice Sensitivity Profile of 1 mm Al beads is within (10 +/- 1) mm for any bead position and PMMA phantom thickness. The FWHM of the Line Spread Function of a 20 um tungsten wire is (180+/-10) um and (230 +/-10) um in the two in-plane directions, i.e. a degradation of the resolution along the direction of tube movement has been detected. Dose delivered in our AEC assessment, ranged from 1.2 mGy to 3.1 mGy.
Automatic analysis of the contrast-detail phantom images showed similar trends in both projection and reconstructed images.
Conclusion: We have identified particular metrics that can serve as part of a quality assurance protocol under MQSA, and that can be evaluated with commercial phantoms. In order to fully characterize a DBT system, further measurements should still be required, such as ghost and lag measurements, or artifact evaluation.