ECG-gated C-arm computed tomography using L1 regularization

Cardiac C-Arm computed tomography leads to a view-starved reconstruction problem because of electrocardiogram gating. The lack of data has to be compensated by a-priori information on the solution. While standard regularization is performed by minimizing a quadratic penalty term, recently proposed limited-view reconstruction techniques perform it by minimizing the L1-norm of the signal in a basis where it is supposed to be sparse. Although some algorithms are formulated with this generic L1-norm term, their practical implementations replace it by total variation, which leads to piecewise constant images. In this article, we investigate the benefits of using a different sparsifying basis, which can be chosen depending on the expected properties of the solution. It results in more flexibility in the reconstruction. We provide an in-depth description of the algorithm used for the minimization.