TH‐A‐141‐05: Ultra‐Sensitive X‐Ray Fluorescence Computed Tomography

Purpose: X‐ray fluorescence computed tomography (XFCT) has previously demonstrated element‐specific imaging in phantoms, but has not yet shown adequate sensitivity to trace amounts of imaging contrast agents required for in vivo molecular imaging. This work addresses this important limitation by reducing the background signal from scattered x‐rays, greatly increasing the sensitivity of XFCT. This improvement will potentially enable clinical x‐ray based molecular imaging. Methods: X‐ray fluorescence imaging (also known as characteristic x‐ray imaging) uses the absorption of x‐rays by heavy elements and the subsequent emission of fluorescent (characteristic) x‐rays to produce images reflecting the concentration of molecular probes. Simulations of x‐ray fluorescence and imaging acquisition were performed using Monte Carlo modeling of radiation transport through a water phantom containing vials of various concentrations of gold (0.01% – 0.5% by weight) in aqueous solution. The lowest detectable limit extrapolated from previous studies was 0.2%. We simulated a scan of the phantom with an x‐ray pencil beam in a first generation CT geometry. Two methods were used to improve the signal‐to‐background ratio: 1) mono‐energetic 82‐keV x‐rays right above the K‐edge of gold; and 2) a geometry avoiding signal detection at 80°–120° scattering angles in order to minimize the detection of scattered x‐rays. Results: The reconstructions show that 0.01% concentration of gold is easily detected with a 40 mGy imaging dose. The optimization of detection geometry to avoid scattered x‐rays improved the signal to background of the fluorescence in the measured spectrum by a factor of 6. Conclusion: We demonstrated an order of magnitude improvement of sensitivity over our previously simulated results. Our results will allow imaging of smaller amounts of contrast agent in molecular imaging and guide the design of an XFCT imaging system. This research was supported by a fellowship from the NIH ‐ Stanford Molecular Imaging Program.