198 - Photon-Counting Cone-Beam CT on a Medical Linear Accelerator for Image Guided Radiotherapy: First Demonstration in Phantom Studies
Presenter(s)
Y. Zhong1, X. Hu1, W. Mao1, K. Yang2, and X. Jia1; 1Department of Radiation Oncology and Molecular Radiation Sciences, Johns Hopkins University, Baltimore, MD, 2Department of Radiology, Mass General Brigham, Boston, MA
Purpose/Objective(s): Photon-counting detector (PCD) technology enables energy-resolved x-ray imaging with improved dose efficiency and intrinsic spectral capability, offering unique opportunities for image guided radiotherapy (IGRT). Integration of photon-counting cone-beam CT (PC-CBCT) with a clinical medical linear accelerator (LINAC) has not previously been demonstrated. The purpose of this work is to report the first implementation of PC-CBCT on a LINAC and to evaluate its feasibility for IGRT through phantom studies.
Materials/Methods: The cover of the CBCT x-ray detector on a LINAC was removed to expose mounting brackets, allowing installation of a CdTe-based PCD (5×10 cm² active area, 0.75 mm thickness, 0.1 mm pixel size) in front of the flat-panel detector with an ~ 4.0 cm lateral offset to extend the field of view. Data were acquired in 2×2 binning mode using two energy thresholds (26 and 64 keV) at 120 kVp with tube settings of 20 mA and 40 ms per frame. Due to the lack of synchronization between detector acquisition and x-ray pulses, ~700 x-ray pulses were generated during a full 2-min gantry rotation, while the detector continuously acquired data at 75 frames/s. Frames corresponding to each pulse were identified and grouped, forming ~700 projections. Calibration datasets were collected across multiple tube currents, and inhomogeneous pixel responses were corrected using a pixel-wise polynomial model to convert raw signals to photon counts. Geometry calibration was performed using a ball-bearing phantom positioned at multiple locations to determine key CBCT geometric parameters. A phantom containing iodine solutions with concentrations ranging from 1% to 10% (w/w) in water was scanned.
Results: The modified imaging geometry resulted in a source-to-isocenter distance of 1.0 m and a source-to-detector distance of 1.37 m, reduced from 1.50 m in the original CBCT configuration due to detector mounting. Reconstructed images showed higher iodine-to-water contrast with energy threshold of 26 keV than 64 keV as expected. Dual-energy decomposition demonstrated linear agreement between estimated and nominal iodine concentrations, with a root-mean-square error of 0.41% (w/w). Contrast-to-noise ratio (CNR) of iodine inserts ranged from ~2 to 18 depending on concentration, and the CNR in decomposed iodine images was reduced by ~2x due to noise amplification inherent in the decomposition process.
Conclusion: This work presents the first demonstration of PC-CBCT implemented on a clinical LINAC for IGRT, validated through phantom studies. The results establish the feasibility of acquiring PC-CBCT images in the modified LINAC platform, laying the foundation for incorporating quantitative multi-energy imaging to enhance treatment guidance, verification, and adaptive radiotherapy.