A 3D verification system using dual-energy cone-beam CT with background subtraction and correction for x-ray CT-based polymer gel dosimetry.

BACKGROUND

Magnetic resonance imaging of polymer gel dosimeters remains the de facto standard to obtain high-quality dose information. However, magnetic resonance imaging scanner access is limited and scan times are long. x-Ray computed tomography-based polymer gel dosimeters (XCT-PGDs) offer convenience owing to easier access to CT scanners, especially cone-beam CT (CBCT) scanners integrated with linear accelerators, although they suffer from low dose resolution and high noise sensitivity. Conventional methods use multiple scans to enhance the contrast-to-noise ratio (CNR), which is time-intensive. Dual-energy CT provides a higher CNR, while dual-energy CBCT (DECBCT) reduces the required scan sets and improves CNR more efficiently, making it a valuable option for accurate XCT-PGD dose verification.

PURPOSE

To investigate the effectiveness of DECBCT and a background correction and subtraction (BCS) method, which was newly developed for minimizing background variations across CBCT scans, in improving the dosimetric accuracy and efficiency of three-dimensional dose distribution verification systems using XCT-PGDs.

METHODS

Three types of plans were created to compare DECBCT combined with BCS against conventional image averaging: mock C-shape volumetric modulated arc therapy, mock multi-target stereotactic radiosurgery, and clinically delivered single-target stereotactic radiosurgery plans. XCT-PGD readout involved pre- and post-irradiation CBCT scans on a linear accelerator at 80 and 140 kV. First, the acquired image sets were processed using the BCS method to subtract background noise, which involved aligning background levels across pre- and post-irradiation scans to minimize variations. Second, the processed image sets were used to generate virtual monochromatic images (VMIs) with DECBCT by applying variable weighting factors. Simple background subtraction (BS) and averaging were also used for the comparative analysis. Dose difference (DD) and gamma analyses were performed with criteria of 3% and 3%/2 mm, respectively.

RESULTS

Compared with conventional BS, the BCS method improved the mean gamma pass rate across image types, with increases of 2.8%, 7.5%, and 10.4% for mock C-shape volumetric modulated arc therapy, mock multi-target stereotactic radiosurgery, and clinically delivered single-target stereotactic radiosurgery plans, respectively. Among the variable weighting factors ranging from -0.5 to 1.5, VMI with the optimal weighting factor of 0.5 provided the highest DD pass rate, although the gamma pass rate remained unchanged. Compared with image averaging, VMI showed comparable gamma pass rates and consistently higher DD pass rates. It improved the DD pass rate by at least 4.7% compared with averaging two images across all plans, and it also achieved comparable or better results than averaging three images. With DECBCT and the BCS method, the gamma pass rate was over 95% across all plans.

CONCLUSION

DECBCT achieved comparable dosimetric accuracy with fewer scans than image averaging. Our approach required only 10 min of total scan time and thus was more efficient than image averaging, which requires over 1 h for image acquisition. Moreover, the BCS method effectively suppressed systematic background variations across CBCT scans. Therefore, DECBCT and the BCS method provide an effective solution for improving the dosimetric accuracy and efficiency of XCT-PGD verification systems.