Encrypted login | home

Program Information

Measuring Spatial Resolution in Digital Breast Tomosynthesis: Update of AAPM Task Group 245

D Scaduto

DA Scaduto1*, M Goodsitt2, H-P Chan2, H Olafsdottir3, M Das4, E Fredenberg5, W Geiser6, D Goodenough7, P Heid8, Y-H Hu1, B Liu9, J Mainprize10, I Reiser11, R Van Engen12, V Varchena13, S Vecchio14, S Glick15, W Zhao1, (1) Stony Brook Medicine, Stony Brook, NY, (2) University Michigan, Ann Arbor, MI, (3) Image Owl, 105 Reykjavik, Iceland, (4) University Houston, Houston, TX, (5) Philips Healthcare, Solna, Sweden, (6) UT MD Anderson Cancer Center, Houston, TX, (7) The George Washington University, Washington, DC, (8) ARCADES, Marseille, France, (9) Massachusetts General Hospital, Boston, MA, (10) Sunnybrook Health Sciences Centre, North York, ON, (11) The University of Chicago, Chicago, IL, (12) LRCB, Nijmegen, Netherlands,(13) CIRS Inc., Norfolk, VA, (14) I.M.S., Pontecchio Marconi (BO), Italy, (15) Food and Drug Administration, Silver Spring, MD


WE-DE-207B-5 (Wednesday, August 3, 2016) 10:15 AM - 12:15 PM Room: 207B

Purpose: Spatial resolution in digital breast tomosynthesis (DBT) is affected by inherent/binned detector resolution, oblique entry of x-rays, and focal spot size/motion; the limited angular range further limits spatial resolution in the depth-direction. While DBT is being widely adopted clinically, imaging performance metrics and quality control protocols have not been standardized. AAPM Task Group 245 on Tomosynthesis Quality Control has been formed to address this deficiency.

Methods: Methods of measuring spatial resolution are evaluated using two prototype quality control phantoms for DBT. Spatial resolution in the detector plane is measured in projection and reconstruction domains using edge-spread function (ESF), point-spread function (PSF) and modulation transfer function (MTF). Spatial resolution in the depth-direction and effective slice thickness are measured in the reconstruction domain using slice sensitivity profile (SSP) and artifact spread function (ASF). An oversampled PSF in the depth-direction is measured using a 50 μm angulated tungsten wire, from which the MTF is computed. Object-dependent PSF is derived and compared with ASF. Sensitivity of these measurements to phantom positioning, imaging conditions and reconstruction algorithms is evaluated. Results are compared from systems of varying acquisition geometry (9-25 projections over 15-60°). Dependence of measurements on feature size is investigated.

Results: Measurements of spatial resolution using PSF and LSF are shown to depend on feature size; depth-direction spatial resolution measurements are shown to similarly depend on feature size for ASF, though deconvolution with an object function removes feature size-dependence. A slanted wire may be used to measure oversampled PSFs, from which MTFs may be computed for both in-plane and depth-direction resolution.

Conclusion: Spatial resolution measured using PSF is object-independent with sufficiently small object; MTF is object-independent. Depth-direction spatial resolution may be measured directly using MTF or indirectly using ASF or SSP as surrogate measurements. While MTF is object-independent, it is invalid for nonlinear reconstructions.

Contact Email: