Evaluation of CBCT Image Quality of the KOALA Dual-Robot Radiotherapy System
Abstract
Purpose
To evaluate image quality for the Kilovoltage Optimized AcceLerator Adaptive radiotherapy (KOALA) system, a dual-robot cone-beam CT (CBCT) imaging and radiotherapy device.
Methods
The KOALA system, comprising a dual-robot system with a 225kVp industrial x-ray tube and a flat panel detector, was utilized to obtain CBCT images of a CATPHAN504 phantom and a lamb head. Half-rotation CBCT scans of the CATPHAN504 were acquired on KOALA with 2mm Al filtration, 100kVp, and 152.7mAs, and on a TrueBeam system at 100kVp with 580mAs. Scans were reconstructed using proprietary Orimtech and Varian reconstruction algorithms for the KOALA and TrueBeam scans, respectively. Geometric accuracy, modulation transfer function (MTF), contrast-to-noise ratio (CNR), and HU accuracy were evaluated. Finally, a lamb head was imaged on KOALA (2mm Al, 100kVp, 152.7mAs), TrueBeam (100kVp, 250mAs), and a GE Optima CT580 clinical CT scanner (100kVp, 428.3mAs) and compared qualitatively.
Results
The CATPHAN504 geometric accuracy was within tolerance, with distance errors of -0.5mm and 0.1mm for KOALA and TrueBeam, respectively. Spatial resolution patterns of 13lp/cm for KOALA and 7lp/cm for TrueBeam were visible. The 10%MTF evaluated using the phantom edge was 8.9lp/cm for KOALA and 6.0lp/cm for TrueBeam. KOALA had worse contrast with a Teflon CNR of 11 compared to 53 on the TrueBeam. Finally, HU accuracy was similar with mean absolute errors of 11HU and 8HU for KOALA and TrueBeam, respectively, and higher noise in the KOALA scan (78HU) compared to the TrueBeam scan (15HU) was observed. The lamb head scan showcased KOALA’s ability to resolve fine structures such as porous trabecular bone, which were not resolvable in either TrueBeam or clinical CT.
Conclusion
While KOALA CBCT offers higher spatial resolution, CNR is low compared to TrueBeam CBCT. Future work will explore optimal filtration, a bowtie filter, and post-processing techniques to decrease image noise.