A spline framework for sparse tomographic reconstruction

We present a spline-based sparse tomographic reconstruction framework. The proposed method utilizes the closed-form analytical Radon transform of B-splines and box splines of any order and integrates the (transform-domain) sparsity of the image into the reconstruction algorithm. Our experiments show that the synergy of sparse reconstruction together with higher order basis functions (e.g., cubic B-splines) improves the accuracy of the reconstruction. This gain can also be exploited for reducing the number of projection angles in the data acquisition.