Binary moving-blocker-based scatter correction in cone-beam computed tomography with width-truncated projections: proof of concept. (21st February 2017)

• Record Type: Journal Article
• Authors: Lee, Ho; Fahimian, Benjamin P; Xing, Lei
• Is Part Of: Physics in medicine & biology. Volume 62:Number 6(2017:Mar.)
• Page Start: 2176
• Page End: 2193
• Publication Date: 2017-02-21
• Subjects: CBCT -- binary moving blocker -- scatter correction -- compressed sensing; Biophysics -- Periodicals; Medical physics -- Periodicals; 610.153
• DOI: 10.1088/1361-6560/aa5913
• Languages: English
• ISSNs: 0031-9155

Abstract

Abstract: This paper proposes a binary moving-blocker (BMB)-based technique for scatter correction in cone-beam computed tomography (CBCT). In concept, a beam blocker consisting of lead strips, mounted in front of the x-ray tube, moves rapidly in and out of the beam during a single gantry rotation. The projections are acquired in alternating phases of blocked and unblocked cone beams, where the blocked phase results in a stripe pattern in the width direction. To derive the scatter map from the blocked projections, 1D B-Spline interpolation/extrapolation is applied by using the detected information in the shaded regions. The scatter map of the unblocked projections is corrected by averaging two scatter maps that correspond to their adjacent blocked projections. The scatter-corrected projections are obtained by subtracting the corresponding scatter maps from the projection data and are utilized to generate the CBCT image by a compressed-sensing (CS)-based iterative reconstruction algorithm. Catphan504 and pelvis phantoms were used to evaluate the method's performance. The proposed BMB-based technique provided an effective method to enhance the image quality by suppressing scatter-induced artifacts, such as ring artifacts around the bowtie area. Compared to CBCT without a blocker, the spatial nonuniformity was reduced from 9.1% to 3.1%. The root-mean-square error of the CT numbers in the regions of interest (ROIs) was reduced from 30.2 HU to 3.8 HU. In addition to high resolution, comparable to that of the benchmark image, the CS-based reconstruction also led to a better contrast-to-noise ratio in seven ROIs. The proposed technique enables complete scatter-corrected CBCT imaging with width-truncated projections and allows reducing the acquisition time to approximately half. This work may have significant implications for image-guided or adaptive radiation therapy, where CBCT is often used.