Comparison of Three Dual Energy Cone Beam CT Configurations for the Hypersight Imaging System, Including a Novel and Cost Effective Source Splitting Technique.
Abstract
Purpose
To compare different dual-energy cone-beam CT (DE-CBCT) configurations for the Varian HyperSight system via relative electron density (ED) and effective atomic number (EAN) quantification, as well as contrast improvement.
Methods
The HyperSight imaging system was simulated in a Monte Carlo environment to test various DE configurations: two sequential acquisitions at 80 kVp and 120 kVp, a dual-layer detector system at 120 kVp, and a split filter placed near the source at 120 kVp. Scatter-corrected images were generated using the EGSnrc application egs_cbct and used to compare these three techniques. A reference phantom with different tissue-equivalent inserts ranging in HU from lung to bone was used to quantify EAN and ED using stoichiometric calibration. Image-based virtual monochromatic images (VMIs) were created, and contrast-to-noise ratio (CNR) was used to quantify soft-tissue visibility in head-and-neck patients.
Results
Stoichiometric calibration yielded comparable results for sequential, dual-layer, and split-filter images. The mean absolute errors for ED and EAN were below 2% and 6%, respectively, for all techniques, with the sequential and split-filter calibration showing better accuracy than with the dual-layer. The average CNR for the phantom’s inserts increased by around 20% for all configurations in soft tissues, while increases of up to 10x were observed for high and low HU materials in low-energy VMIs. A similar increase was observed with patient data around soft tissues, including lymphatic nodes in the head-and-neck regions.
Conclusion
The split filter provides estimates of ED and EAN that are similar or better than those from sequential acquisitions and dual-layer detectors while requiring only a single scan and simple hardware, while also being able to increase contrast via low-energy VMIs. A deep learning scatter correction model will be used alongside CBCT data acquired from the Ethos imaging system to further evaluate its clinical application in adaptative radiotherapy (ART).