Optimization and Comparison of Different Digital Mammographic Tomosynthesis Reconstruction Methods

摘要

Digital breast tomosynthesis is a three-dimensional imaging technique with limited-angle series of projection images that allows the reconstruction of depth information along the breast. This study investigated optimized and compared different tomosynthesis methods and chose the optimal one for breast tomosynthesis imaging. We investigated several 3-D tomosynthesis reconstruction algorithms and studied the effect of acquisition parameters for different candidate reconstruction algorithms according to computer simulation and physical measurements of impulse response analysis modulation transfer function and noise power spectrum. We applied a relative noise-equivalent quanta (NEQ) analysis to compare candidate algorithms and acquisition parameters. Additionally the importance of point-by-point back projection (BP) for isocentric motion in breast tomosynthesis was investigated. It improves the in-plane sharpness of structures such as microcalcifications which bear important meaning in clinical tasks for breast cancer detection.