Comprehensive Image Quality Evaluation and Motion Phantom Studies of an Ultra-Fast (6-Second) Cone-Beam Computed Tomography Imaging System on a Ring Gantry Linear Accelerator

Abstract

Purpose: To evaluate the image quality of an ultrafast cone-beam computed tomography (CBCT) system—Varian HyperSight. Methods and Materials: In this evaluation, 5 studies were performed to assess the image quality of HyperSight CBCT. First, a HyperSight CBCT image quality evaluation was performed and compared with Siemens simulation-CT and Varian TrueBeam CBCT. Second, a visual comparison of image quality among simulation-CTs, HyperSight CBCT, and TrueBeam CBCT was performed for a patient with head and neck cancer and patients with metal dental fillings and prostheses. Third, the Hounsfield unit (HU) versus electron density curve of HyperSight CBCT was compared with GE and Siemens simulation CTs. Fourth, Siemens simulation-CT and HyperSight CBCT scans were acquired on the Catphan set-up at different locations inside the bore (±10 cm in all 3 principal directions from the center), and the HU variations for different materials were evaluated. Fifth, a 4-dimensional lung tumor phantom study was performed to assess moving tumor alignment during image registration. Results: Significant improvement of image contrast, HU constancy, and noise level on HyperSight CBCT was observed compared with TrueBeam CBCT. Significant image quality improvement was observed on HyperSight CBCT for patients with dental fillings and prostheses compared with simulation-CT without metal artifact reduction. The linear fit trendline of HU versus electron density curves for GE simulation-CT, Siemens simulation-CT, and HyperSight CBCT showed a 0.6% difference for HU values below 2000. The maximum HU difference for HyperSight CBCT when Catphan was positioned within ±10 cm in all 3 principal directions was ≤ 98 on bone 50%, ≤ 29 other than bone, and was ≤ 31 on bone 50%, and ≤ 17 other than bone for Siemens simulation-CT. Both tumor shape and tumor alignment discrepancies on CBCT scans were observed in a 4-dimensional phantom study. Conclusions: This evaluation shows significant image improvement of HyperSight CBCT over conventional CBCT on image contrast, HU constancy, and noise level with scatter correction and metal artifact reduction reconstruction methods. HyperSight CBCT has similar image quality to simulation-CTs and shows the potential application for treatment planning. The rapid acquisition of HyperSight CBCT showed both tumor shape and tumor alignment discrepancies of moving targets. Careful considerations of patient respiratory motion monitoring and target matching are highly recommended.