Development of an attenuation correction method for direct x-ray fluorescence (XRF) imaging utilizing gold L-shell...

This work proposes a semiempirical correction method for attenuation of x-ray fluorescence (XRF) photons and/or an excitation beam during direct XRF imaging (i.e., mapping) of gold nanoparticle (GNP) distribution utilizing gold L-shell XRF photons.

The current method was first devised by finding the two following relationships: (a) ratio of gold XRF peak intensity (L_α at ~9.7 keV and L_β at ~11.4 keV) vs pathlength of XRF photons; (b) XRF photon counts produced (N_xrf ) vs scattered photon counts produced (N_scat ). Monte Carlo simulations were performed using the Geant4 tool kit to characterize the aforementioned relationships for different tissue-like media. The applicability of the method was tested experimentally by acquiring 2D L-shell XRF images of custom-made phantoms using an experimental benchtop x-ray fluorescence computed tomography setup.

The results show that the ratio of gold L-shell XRF peak intensities allowed an estimation of the pathlength of XRF photons, thus can be utilized to correct for attenuation of XRF photons after emission. The results also demonstrate that N_scat , through a proportionality $N_{\mathrm{xrf}}\propto N_{\mathrm{scat}}^T$ where the exponent T depends on the energy of scattered photons, could be used to correct for attenuation of an excitation beam prior to producing XRF photons. The corrected XRF signal was found independent of the densities of tissue-like media present along the passage of an excitation beam or emitted XRF photons.

The current results suggest that the developed attenuation correction method plays an essential role for the detection of GNPs on the order of parts-per-million, and also for the determination of GNP concentration/location within the imaging object made of tissue-like media, without any prior knowledge of the imaging object shape, under the conditions deemed relevant to biomedical applications of gold L-shell XRF-based imaging.