Our work on iterative image-reconstruction applied to Digital Breast Tomosynthesis is split in two efforts: identifying optimization-based algorithms that provide useful images with few iterations, and developing image quality metrics that guide the parameter settings of the image-reconstruction algorithms. We prototype many optimization-based approaches, involving total-variation, using the Chambolle-Pock primal-dual algorithm. We also present image quality metrics tailored to the task of tumor/mass classification, and discuss the difference in image properties with respect to scan-angle.

This presentation is part of Minisymposium “MS7 - Limited data problems in imaging (2 parts)”
organized by: Bernadette Hahn (University of Würzburg) , Gaël Rigaud (Saarland University) , Jürgen Frikel (OTH Regensburg) .

Authors:

Emil Sidky (University of Chicago)

Sean Rose (University of Chicago)

Ingrid Reiser (University of Chicago)

Xiaochuan Pan (University of Chicago)

Keywords:

computed tomography, digital breast tomosynthesis, image reconstruction, inverse problems, limited-angular range tomography, nonlinear optimization