TH-AB-204-10: Drastic Performance Improvement of a Polychromatic Cone- Beam X-Ray Fluorescence Computed Tomography (XFCT) System Using a Kilowatt-Range X-Ray Source




To demonstrate drastic performance improvement of a polychromatic cone-beam x-ray fluorescence computed tomography (XFCT) system using a kilowatt-range x-ray source.


An experimental polychromatic cone-beam XFCT system capable of imaging gold nanoparticles (GNPs) within small-animal-sized objects by stimulating K-shell XRF from gold and detection using a cadmium-telluride detector has been under development. Originally configured with a low-power (∼50 W) x-ray source (105 kVp, 0.45 mA, 0.9-mm tin filter), the system has now been retrofitted with a high-power (∼3 kW) source, operated at 125 kVp, 25 mA. The higher flux allowed the use of a 2-mm tin filter to harden the incident x-ray spectrum, thereby optimizing XRF production. The material detection limit was determined at these settings using GNP/water-containing calibration phantoms (0, 0.1, 0.5, 1, 2 wt. % of GNPs). Subsequently, a 3-cm-diameter PMMA phantom with 0.2, 0.5, and 1 wt. % GNP/water inserts was imaged. The image reconstruction algorithm was modified with Compton-scatter-based background fitting for accurate XRF signal extraction and empirical attenuation correction.


Under the optimized conditions, the detection limit was remarkably lower (factor of 2.5), compared to the original configuration, allowing imaging of GNP concentrations as low as 0.2 wt. %. The reconstructed image demonstrated accurate localization of all inserts. Scan time was drastically decreased (factor of 4), enabling imaging of the phantom in 1.5 hours with just one detector translated 11 times. By reverting to the 0.9-mm filter, as used previously, scan time could be lowered to 30 minutes, at the expense of sensitivity. Implementation of additional/array detectors for parallel data acquisition will lower scan time further.


By adopting a high-power source for the XFCT system, dramatic improvements were made, while meeting realistic constraints. This work shows the practicality of performing XFCT for routine preclinical ex-/in-vivo imaging using GNPs.