Abstract
One-dimensional linear X-ray grids are typically used in clinical practice to reduce the number of scattered X-rays reaching the detector surface, thus improving image contrast in radiography. However, such grids can only remove scattered X-rays in one direction. Additionally, the primary challenge in using X-ray grids in digital radiography is the presence of grid artifacts, such as grid strip shadows and moiré, which can reduce image quality. To overcome these difficulties, in this study, we developed a prototype two-dimensional (2D) crisscrossed grid with a strip density of 1.724 lines/mm by adopting a high-precision sawing process to further improve its scatter radiation removal ability and implemented it in a cone-beam computed tomography (CBCT) system to enhance image quality. We propose a new software-based method called the grid artifact reduction (GAR) algorithm to efficiently eliminate grid artifacts of a 2D grid in CBCT. The GAR algorithm involves three main steps: 1) preprocessing for flat-field compensation, 2) postprocessing for GAR, and 3) CBCT reconstruction. To verify the efficacy of the proposed approach, we conducted an experiment on a CT quality control phantom (Pro-CT MK II™) and quantitatively evaluated the CBCT image quality using the contrast-to-noise ratio (CNR) and Hounsfield unit (HU) accuracy error. The CNR value measured in the CBCT image obtained using the 2D grid and proposed GAR algorithm was 3.22, which is approximately 4.4 times of the value obtained without GAR. Furthermore, the corresponding HU accuracy error was 30.47, an improvement of approximately 1.3 times. These results indicate that the proposed approach is highly efficient in eliminating scattered X-rays, thereby improving the quality of CBCT images. Consequently, high-quality CBCT images were obtained using the proposed approach (i.e., using a 2D crisscrossed grid in CBCT and the proposed GAR algorithm).