Can image quality of daily CBCT be improved by a new reconstructor?
Mao W, Liu C, Snyder K, Zhao B, Gardner SJ, Kumarasiri AD, Kim J, Wen N, Chetty IJ, and Siddiqui F. Can image quality of daily CBCT be improved by a new reconstructor? Int J Radiat Oncol Biol Phys 2017; 99(2):S93.
Oncol Biol Phys
Purpose/Objective(s): Image quality of on-board cone-beam CT (CBCT) imaging falls short of fan-beam CT particularly in terms of low contrast visibility. This limits the application of CBCT mainly to patient setup with high contrast structures. Nevertheless, it might be possible to extract much more information from CBCT imaging using next generation preprocessing and reconstruction algorithms. We have assessed the image quality of a prototype CBCT reconstructor to determine if software-only-based changes can improve image-guidance and possibly facilitate advanced applications. Purpose/Objective(s): The current TrueBeam CBCT reconstruction removes scatter using a kernel-based correction followed by filtered back-projection-based reconstruction (FDK). In the prototype CBCT reconstruction pipeline these steps have been replaced by a finite element solver (AcurosCTS) based scatter correction and a statistical (iterative) reconstruction. Image quality improvements due to the prototype reconstruction pipeline have been quantitatively analyzed on scans of a standard phantom, as well as on daily CBCT scans of head/neck patients acquired for image-guidance. Both a standard full-fan Head CBCT and an institutional half-fan full-rotation Head CBCT protocols have been quantitatively investigated including uniformity, consistency, spatial resolution, and noise level by a commercially available software package to keep analysis consistent. In addition, clinical image datasets for head/neck patients have been qualitatively evaluated by certified physicians. Results: Phantom studies show that noise level is reduced to around 64% compared to that resulting from the current reconstructions while modulation transfer function (MTF) measurements indicate that spatial resolution is maintained. HU uniformity improved from 7.4 ± 1.8 to 2.4 ± 1.4 for full-fan Head CBCT and from 27.5 ± 8.1 to 22.5 ± 8.4 for half-fan Head CBCT. Contrast to noise ratio (CNR) was analyzed based on a 1% contrast insertion with a diameter of 15 mm. CNR was improved from 0.7 to 1.5 and from 0.8 to 3.7, for the full-fan and half-fan Head CBCT, respectively. This is mainly due to the significant reduction in image noise. Artifacts in head/neck patient images due to photon starvation, cone-beam artifacts, and streak artifacts are all reduced by the prototype reconstruction. Qualitatively, prototype reconstructions show enhanced soft-tissue definition and improve detectability of tissue boundaries, significantly improving soft-tissue visualization. Conclusion: This prototype reconstructor improves the soft-tissue definition that is necessary for accurate visualization, contouring, dose calculation, and deformable image registration in clinical applications. These are expected to facilitate soft tissue-based patient positioning and may enable new applications, such as CBCT-based online adaptive radiotherapy.