A New Voxelization Strategy in Compressed-Sensing (CS)-Based Iterative CT Reconstruction for Reducing Computational Cost: Simulation and Experimental Studies

Abstract

In recent computed tomography (CT) reconstruction, iterative methods have been often used owing to the potential to provide three-dimensional (3D) images of superior reconstruction quality to common filtered-backprojection (FBP)-based methods. However, these methods require enormous computational cost in the iterative process, which has still been an obstacle to put them to practical use. In this work, to overcome these difficulties, we propose a new cone-beam CT reconstruction with a dual-resolution voxelization strategy for a small region-of-interest (ROI) reconstruction in which the voxels outside the target ROI are binned by, for example, 2 × 2 × 2, 4 × 4 × 4, 8 × 8 × 8, etc., while the other voxels remain unbinned. We considered a compressed-sensing (CS)-based algorithm with a dual regularization strategy, rather than common FBP-based methods, for more accurate CT reconstruction. We implemented the proposed CS-based algorithm and performed a systematic simulation and experiment. Our results indicate that the proposed method seems to be effective for reducing computational cost considerably in iterative CT reconstruction, keeping the reconstruction quality inside the target ROI not much degraded.