Development of scatter correction methods using a beam-blocker in cone-beam CT

Abstract

X-ray scatter constitutes a dominant source of degradation of image quality such as cupping and dark steak artifacts in cone-beam CT. Although many methods were introduced for scatter correction, a beam-blocker composed of multiple strips is a useful gadget for scatter correction and/or for dose reduction in cone-beam CT (CBCT). However, the use of such a beam-blocker would yield cone-beam data that can be challenging for accurate image reconstruction from a single scan in the filtered-backprojection (FBP) framework. The focus of the work was to develop an analytic image reconstruction method for CBCT that can be directly applied to partially blocked cone-beam data in conjunction with the scatter correction. We developed two types of rebinned backprojection-filteration (BPF) algorithm for reconstructing images from the partially blocked cone-beam data in a circular scan. We also proposed a beam-blocking geometry considering data redundancy such that an efficient scatter estimate can be acquired and sufficient data for BPF image reconstruction can be secured at the same time from a single scan without using any blocker motion. Additionally, a hybrid type scatter correction method and noise reduction scheme have been developed. Hybrid type scatter correction is a combined method of measurement and convolution-based scatter correction. We have performed both simulation and experimental studies to validate the rebinned BPF algorithm for image reconstruction from partially blocked cone-beam data. Quantitative evaluations of the reconstructed image quality were performed in the experimental studies. The simulation study revealed that the developed reconstruction algorithm successfully reconstructs the images from the partial cone-beam data. In the experimental study, the proposed method effectively corrected for the scatter in each projection and reconstructed scatter-corrected images from a single scan. Reduction of cupping artifacts and an enhancement of the image contrast have been demonstrated. We have successfully demonstrated that the proposed scanning method and image reconstruction algorithm can effectively estimate the scatter in cone-beam projections and produce tomographic images of nearly scatter-free quality.