Digital tomosynthesis mammography: Improvement of artifact reduction method for high-attenuation objects on reconstructed slices

One major image quality problem in digital tomosynthesis mammography (DTM) is the poor depth-resolution caused by the inherent incomplete sampling. This problem is more pronounced if high-attenuation objects, such as metallic markers and dense calcifications, are present in the breast. Strong ghosting artifacts will be generated in the depth direction in the reconstructed volume. Incomplete sampling of DTM can also cause visible ghosting artifacts in the x-ray source motion direction on the off-focus planes of the objects. These artifacts may interfere with radiologists' visual assessment and computerized analysis of subtle mammographic features. We previously developed an artifact reduction method by using 3D geometrical information of the objects estimated from the reconstructed slices. In this study, we examined the effect of imaging system blur in DTM caused by the focal spot and the detector modulation transformation function (MTF). The focal spot was simulated as a 0.3 mm square array of x-ray point sources. The detector MTF was simulated using the Burgess model with parameters derived from published data of a GE FFDM detector. The spatial-variant impulse responses for the DTM imaging system, which are required in our artifact reduction method, were then computed from the DTM imaging model and a given reconstruction technique. Our results demonstrated that inclusion of imaging system blur improved the performance of our artifact reduction method in terms of the visual quality of the corrected objects. The detector MTF had stronger effects than focal spot blur on artifact reduction under the imaging geometry used. Further work is underway to investigate the effects from other DTM imaging parameters, such as x-ray scattering, different polyenergetic x-ray spectra, and different configurations of angular range and angular sampling interval.