Assessment of multi-directional MTF for breast tomosynthesis

A method was developed to assess the multi-directional modulation transfer function (MTF) of breast tomosynthesis imaging systems using a sphere phantom. The method was initially developed based on a simulation dataset. Projections were simulated for a uniform voxelized breast phantom with sphere inserts using a fluence modeled from a 28 kVp beam incident upon an indirect flat-panel detector. Based on cascaded systems modeling, characteristic noise and blurring were added to each projection. The projections were reconstructed using a standard filtered backprojection technique, producing a 3D volume with an isotropic voxel size of 200 µm. ROIs that completely encompassed single spheres were extracted and conical regions were prescribed along the three major axes extending from the centroids. Pixels within the cones were used to form edge spread functions (ESFs), from which the directional MTFs were calculated. Binning size and conical range were adjusted to maximize the accuracy and to minimize the noise of the MTF. A method was further devised to remove out-of-plane artifacts from the ESF in the x-y plane. Finally, the method was applied to experimentally assess the directional MTF of a prototype tomosynthesis system. Comparisons of the sphere-based MTF along the different axes and the theoretical MTF yielded good agreement. A 30° angular cone and a 20 µm sampling were found to provide an ideal trade-off between the noise and accuracy of the measurement. The removal of artifacts in ESF yielded 'modified' MTFs that enabled a resolution-only characterization of the in-slice resolution of tomosynthesis. Drop-off frequencies in the x- and y-directional MTFs were 1.6 cycles mm(-1) and 1.5 cycles mm(-1), respectively. The presented method of separating the effective resolution and artifacts from the measured ESF was found experimentally implementable and is expected to facilitate the interpretation of MTF measurements in tomosynthesis.