Room: Exhibit Hall | Forum 6
Purpose: To assess cone-beam computed tomography (CBCT) image quality performance across multiple institutions, using prescribed and center-specific imaging protocols to identify variability.
Methods: A CatphanÂ® 504 was sent to 16 different cancer centers and CBCT images were acquired for a vendor-specific standard imaging protocol (Ps - 120-125kVp, 800mAs) and a local pelvis imaging protocol (Pp), on at least one imaging system per center. In total, 84 CBCTâ€™s acquired on both Varian OBI (n=39) and Elekta XVI (n=45) CBCT platforms have been analyzed to date. These images were analyzed with a commercially available image-analysis application, and metrics were extracted including uniformity, contrast to noise ratio (CNR), CT number accuracy, and presence of artifact.
Results: Artifacts included rings, truncation, dinner plates, and uniformity (cupping/capping), and were present on more than half of the datasets. The mean (standard deviation) CT number of air for Ps was -833 (17) HU and -1018 (1) HU for XVI and OBI, respectively, while for Teflon the values were 690 (46) HU and 932 (41) HU. The average CT numbers for air and Teflon were comparable for Pp, with a noticeable increase in variability for XVI systems. The average Ps CNR was lower for XVI (7.8 (1.3)) than for OBI (17.0 (5.8)) due to the smaller voxel size of the XVI reconstruction. CNR improved for Pp as voxel sizes increased for these protocols. On average, uniformity performance was higher on the XVI platform, as the OBI uniformity was susceptible to common artifacts present in many of these datasets.
Conclusion: This work demonstrates the large variability in CBCT image quality performance across both venders and institutions. This work will help inform new performance standards and guidelines which will be necessary as we enter the new era of CBCT-based adaptive workflows.