Zipline: A fast update scheme for reconstruction with separable system models

While iterative reconstruction techniques for X-ray computed tomography (CT) data offer important opportunities for improving image quality and reducing noise or radiation dose, the computational burden and subsequent run-time associated with such methods remains a challenge. In this paper, the authors introduce a computational approach, herein named Zipline, that improves performance for some update schemes when employing system models exhibiting a certain separability property. Specifically, we focus on iterative coordinate descent (ICD) updates.