Abstract
AIM: This work presents a multi-institutional image comparison between novel O-ring linac cone-beam computed tomography (CBCT) and fan-beam computed tomography simulator (FBCT), among five institutions. MATERIALS AND METHODS: A phantom was sent to five institutions with Ethos and Halcyon units equipped with HyperSight CBCT (HS-CBCT). HyperSight used a 125 kVp/176 mAs protocol and an iterative-CBCT reconstruction algorithm with scatter correction. FBCT imaging protocols used 125 kVp and exposure from 176 to 379 mAs. RESULTS: Linear fitting of relative electron density versus Hounsfield unit (RED-vs-HU) curves was determined for RED ≤ 1 and RED ≥ 1. The contrast was evaluated with regard to solid water. Noise and contrast-to-noise ratio (CNR) were evaluated with and without exposure normalization. The RED-vs-HU curve for RED ≤ 1 of FBCT shows a 4% greater slope than HS-CBCT (P = 0.024), while for the RED ≥ 1 case, the FBCT slope is 15% larger compared to (P = 0.00004). HS-CBCT has a larger contrast as RED differs from 1, although at RED > 1, the difference is significant, up to 14%. However, noise is larger for HS-CBCT, especially for RED > 1, up to 10 times. Even with exposure correction, noise is more significant for HS-CBCT, which leads to a CNR ten times smaller than FBCT (RED = 1.78). CONCLUSION: This multi-institutional analysis of HS-CBCT showed reduced slopes of the RED-vs-HU curves compared to FBCT, which results in a greater sensitivity of the HS-CBCT to RED changes. HS-CBCT shows better contrast performance, although the broader beam leads to noise up to 10 times larger noise, even with the scatter correction. The CNR of the FBCT is up to one order of magnitude greater than that for HS-CBCT.