Practical hybrid convolution algorithm for helical CT reconstruction

Great strides have been taken in the last few years in the development of both approximate and exact reconstruction algorithms for helical cone-beam CT. The proposed algorithm uses so-called hybrid convolution which is a sum of ramp filtering and Hilbert transform of projection data. We show that it satisfies the requirements of a modern practical CT reconstruction algorithm. In this work we evaluate the new method and compare it to several previously published algorithms. We evaluate spatial resolution and noise properties, and also the cone beam artifact. Evaluation demonstrated that the proposed algorithm outperforms the helical Feldkamp algorithm and another candidate in terms of image noise-to-resolution balance, as well as the cone beam artifact. We also discuss implementation issues and propose a simplified version for over-scan reconstruction