Purpose: Dual energy imaging techniques such as kV switching increase system complexity and are prone to motion or registration issues. Conversely, a dual-layer flat panel detector (FPD) has built-in dual energy capabilities from a single exposure. This work evaluates a prototype dual-layer FPD for dual-energy cone-beam CT (DE-CBCT).
Methods: The prototype detector active area is 43x43 cmÂ², with a thinner top scintillator coupled to an a-Si panel, a middle filter to increase spectral separation, and a thicker bottom scintillator. CBCT scans were acquired on a benchtop system, using a simple phantom consisting of several tissue-equivalent inserts. The system geometry (SAD 100 cm, SID 150 cm), x-ray energy (120 kV), and detector frame rate (15 fps) were typical of an on-board imager for radiotherapy. The dual-layer projection images were decomposed into water/bone projections using a previously determined empirical calibration, and then reconstructed to form water/bone images and virtual monoenergetic (VM) images. Contrast-to-noise ratio was assessed for the VM image that maximized CNR between water and polyethylene.
Results: The material-specific images recovered water and bone in their respective images. Water was quantified to within 1.8% accuracy and bone to 0.6%. The VM images showed reduced beam hardening compared to images from a single-layer detector. At the optimal VM energy of 71 keV, CNR increased to 8.49 compared to that of the top layer (5.64) or bottom layer (4.80) alone.
Conclusion: This work demonstrates the feasibility of a dual-layer detector for quantitative DE-CBCT. Future work will assess the quantitative accuracy of bone/iodine material quantification for interventional and diagnostic applications and electron density for radiotherapy applications. We will also compare the performance against a kV switching acquisition at equal dose.
Funding Support, Disclosures, and Conflict of Interest: ES, AS, JZ, RC, JSL, and ML are employees of Varex Imaging. AW has a Collaboration Agreement with Varex Imaging.