New approximate filtered backprojection algorithm for helical cone-beam CT with redundant data

This paper deals with image reconstruction for helical cone beam CT with multi-row detectors. This problem has been investigated actively in recent years. An exact filtered back projection (FBP) algorithm is known for this problem but it is commonly recognized that this algorithm suffers two drawbacks: (1) the exact algorithm is computationally inefficient and difficult to implement due to the complexity of the filtering step and (2) the exact algorithm essentially uses detector data corresponding to the minimum detector region compatible with exact reconstruction called the PI detector window. This paper proposes a new approximate FBP algorithm which overcomes the said drawbacks. The main feature of the proposed algorithm is the ability to use an arbitrary length of helix segment to reconstruct each transaxial slice. This feature allows the improvement of noise performances when scanners are operated with a smaller pitch than its allowed maximum value. Furthermore, the hybrid algorithm was proposed as a framework which can achieve a better tradeoff between the noise performances and the artifact level.