Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2023 Feb 11.
Published in final edited form as: Nucl Instrum Methods Phys Res A. 2021 Dec 13;1025:166198. doi: 10.1016/j.nima.2021.166198

Sensitivity enhancement of an experimental benchtop x-ray fluorescence imaging system by deploying a single crystal cadmium telluride detector system optimized for high flux x-ray operations

Hem Moktan 1, Sandun Jayarathna 1, Sang Hyun Cho 1,2,*
PMCID: PMC8942383  NIHMSID: NIHMS1764226  PMID: 35340930

Abstract

In this work, an energy-resolving thermoelectrically cooled single crystal cadmium telluride (CdTe) detector system upgraded with the latest firmware was optimized for high x-ray flux operations using high bias voltage and fast peaking time. This detector system was deployed into an experimental benchtop x-ray fluorescence (XRF) imaging/computed tomography (XFCT) system developed for quantitative imaging of metal nanoprobes such as gold nanoparticles (GNPs). Using the firmware-upgraded and existing/old CdTe detector systems, the Compton/XRF spectra from small (8 mm diameter) GNP-containing phantoms were acquired. The phantoms were irradiated with 1.8 mm Sn-filtered 125 kVp cone beam x-rays at 24 mA. The firmware-upgraded detector system produced relatively lower dead time under high x-ray flux, compared with the old detector system, and performed well with the spectral resolution of ~0.7 keV (in full width at half maximum) at 69 keV photon energy. Given the same 2 mm aperture detector collimator and irradiation time of 10 s, this detector system managed to score nearly 50% more gold XRF signals than the existing one at all GNP concentrations tested. This improvement resulted in the GNP detection limit of 0.02 wt. % which was lower than that (0.03 wt. %) achievable with the existing detector system. When combined with the detector collimator containing a larger (3 mm) aperture, the firmware-upgraded detector system produced drastically more gold XRF signal at a given GNP concentration (e.g., 9 times more for 1 wt. % GNP solution and irradiation time of 10 s), leading to further reduction in the GNP detection limit (i.e., 0.01 wt. %). The present investigation showed that the firmware upgraded CdTe detector system optimized for high x-ray flux operations allowed for better photon counting efficiency, thus leading to sensitivity enhancement of an experimental benchtop XRF/XFCT imaging system.

Keywords: x-ray fluorescence, benchtop x-ray fluorescence imaging, single crystal cadmium telluride detector, gold nanoparticles

1. Introduction

X-ray fluorescence (XRF), the emission of characteristic x-rays from materials when irradiated with external x-ray, has been extensively utilized for elemental and chemical analysis as well as imaging applications [1–3]. The success of XRF-based applications depends critically on the efficient production and detection of XRF photons. Therefore, the choice of excitation x-ray sources and x-ray detectors heavily influence the performance of XRF analysis or imaging systems/devices. In particular, the role of x-ray detectors becomes even more critical for XRF-based applications that utilize polychromatic x-ray sources on a benchtop setting, instead of monochromatic x-ray sources at synchrotron facilities, because of added complexity in extracting XRF signals from the scattered photon background [4–6].

Over the last decade, benchtop XRF imaging/computed tomography (XFCT) has emerged as one of the important XRF-based applications. The vast majority of benchtop XRF/XFCT imaging systems developed so far have adopted single crystal x-ray detectors [4, 5, 7–13]. Specifically, energy-resolving thermoelectrically cooled single crystal cadmium telluride (CdTe) detectors have been preferred for benchtop XRF/XFCT imaging because of their commercial availability, excellent energy resolution, compact size, and easy operation without requiring relatively cumbersome liquid nitrogen cooling [14]. For example, such detectors have been successfully incorporated into benchtop XRF/XFCT imaging systems, allowing detection and imaging of low concentration (< 1 wt.%) of gold nanoparticles (GNPs) within small animal and comparably sized phantoms [4, 5, 10–13, 15]. As reported in a previous study [13], however, the detector system (typically consisted of detector, preamp, and multichannel analyzer) may experience high dead time fraction (greater than 50%) when dealing with high photon flux, e.g., during the acquisition of XRF/scattered photon spectra produced by high power (e.g., kilowatt) x-ray sources. While it would be less of a concern for traditional XRF analysis of samples, the photon counting efficiency of the detector directly affects the performance of benchtop XRF/XFCT imaging systems for biomedical applications, in terms of the system sensitivity (or detection limit), x-ray dose, and imaging/scan time [13].

In this study, we investigated the performance of an experimental benchtop XRF/XFCT imaging system adopting an “OEM” version of an energy-resolving thermoelectrically cooled single crystal CdTe detector system (AXR-CdTe, Amptek, Inc., Bedford, MA, USA) upgraded with the latest firmware (FW6). The latest version of firmware offered multitude of features that allowed the detector system to operate at a higher bias voltage and faster peaking time, compared with the detector system (adopting the previous version of firmware (FW5)) used in our existing experimental benchtop XRF/XFCT imaging system [13]. These characteristics in turn enabled the detection and fast processing of high photon flux, up to 4 × 106 counts per second (cps), without sacrificing spectral resolution [14]. The improved photon counting efficiency of the detector was expected to result in the performance enhancement of benchtop XRF/XFCT imaging systems that utilize similar single crystal CdTe detectors [13]. Thus, the primary goal of this investigation was to demonstrate the expected performance enhancement of such experimental benchtop XRF/XFCT imaging systems in a quantitative fashion. The results of this investigation may help further optimize the existing or future benchtop XRF/XFCT imaging systems.

2. Methods

Our experimental benchtop XRF/XFCT imaging system [13], as shown in Fig. 1, was slightly modified by replacing the existing AXR-CdTe detector system with the previous firmware version (FW5) by the one with the latest version of firmware (FW6). Although both detectors had the same size of CdTe crystal (5 mm × 5 mm × 1 mm) with an integrated digital pulse processor/multichannel analyzer (DP5), associated firmware and pulse processing software were notably different with slight changes on the hardware (not detailed here). Note, from here on, the systems with FW6 is referred to as the new detector system, whereas the one with FW5 is designated as the old detector system. The detectors in both systems were coupled with 5 cm-thick stainless-steel collimators (aperture diameter of 2 or 3 mm) and placed inside a 2.5 cm-thick lead housing.

Fig. 1.

Fig. 1.

Open in a new tab

Experimental benchtop XRF/XFCT imaging system. This figure is for illustrative purposes only and not drawn to scale.

A height-adjustable stage secured a rotational stage whose center (isocenter) was at 15 cm from the tungsten target of the x-ray source (XRS-160, COMET Technologies USA, Inc., San Jose, CA, USA). The rotational stage held the calibration phantoms (plastic tubes of 8 mm diameter) filled with 200 μL solutions of 15-nm-diameter gold nanoparticles (GNPs) (AuroVist - Nanoprobes Inc., Yaphank, NY, USA) at different GNP concentrations prepared using the method consistent with the previous studies [13, 15]. The detector was placed at 10 cm from the isocenter (i.e., isocenter-to-detector collimator entrance distance of 5 cm) and at 90° with respect to the incident x-ray beam direction.

For data acquisition, the existing software previously designed for the old detector system was unable to support the new detector system due to firmware/software incompatibility. Therefore, a new custom software, based on Matlab (MATLAB, 2015b; MathWorks, USA), was developed to seamlessly interface the new detector system to the latest digital pulse processing software, DPPMCA, and the linear/rotational stage automation scripts. This was done by modifying the manufacture provided software development kit (DP5_API_SDK) using Microsoft Visual Studio 2019 and combining it with the stage automation script developed by customizing the respective manufacturer provided software development kits.

The GNP-loaded calibration phantoms were irradiated by a cone beam x-ray source (125 kVp, 24 mA with a focal spot size of 5.5 mm) and Compton/XRF spectra were acquired using both old and new detector systems. For each GNP concentration, three sets of data were acquired for statistics. Consistent with the previous studies [13, 15], the acquired spectra were energy-calibrated, corrected for CdTe efficiency, and then analyzed for the spectral resolution at the gold Kα1 XRF peak. The deconvolution-based signal extraction algorithm [15], developed by the current research group, was implemented to extract net XRF signals from the Compton background. The net XRF photon signals (only at gold Kα XRF peaks) generated the system calibration curve. The net background counts at gold Kα XRF peak region estimated the standard deviation of the background (square root of the net counts per Poisson statistics). Consistent with the previous study [13], the lowest GNP concentration for which the net XRF signals were higher than 1.96 times the standard deviation of the background (95% confidence interval) was regarded as the detection limit achievable by the respective systems.

Additionally, XFCT of a small-animal-sized polymethyl methacrylate (PMMA) cylindrical phantom (3 cm in diameter and 3 cm in height) containing three cylindrical holes filled with GNP solutions (0.1, 0.05 and 0.02 wt.%) was performed using the new detector system. For straightforward comparison, the XFCT scanning protocol and image reconstruction/analysis methods were following our previous study performed with the old detector system coupled with a 2 mm aperture collimator [13]. Briefly, the phantom was scanned at 30 rotational angles (in 12° interval) in combination with 11 translational positions (in 3 mm step size) using 10 s irradiation time at each combination (or projection). After the net XRF signal extraction, the XFCT image was reconstructed using a standard filtered back-projection algorithm and analyzed by visual inspection and the contrast-to-noise ratio (CNR) metric [13].

In all XRF experiments, the old detector system was operated at a bias voltage of 500 V and peaking time of 32 μs as optimized through previous experimental studies [13, 15], whereas the new detector system was characterized to operate at a higher bias voltage of 700 V and fast (3.2 μs) peaking time. The temperature on both detectors was set to 214 K. Compton/gold XRF spectra registered by both detectors and their dead time fractions under high-flux x-ray imaging conditions were evaluated to compare their photon counting efficiency. The GNP detection limits achieved by each detector system were compared to determine the enhancement in the benchtop XRF/XFCT imaging system sensitivity. The new detector system was also coupled with a collimator containing larger (3 mm) aperture to operate it to its full potential, and the detection limit achieved by such a detector collimator combination was analyzed to determine further enhancement in the system sensitivity. These comparisons allowed for quantitative assessment of the performance enhancement in the current benchtop XRF/XFCT imaging system, due to the optimization of an AXR-CdTe detector system with the latest firmware version, i.e., the new detector system for high flux x-ray operations.

3. Results

Under the identical setting using the 2 mm aperture detector collimator, the x-ray flux experienced by each detector system was relatively low, where the input count rates reported by the old detector system and the new detector system were respectively 3.5 × 103 and 4.6 × 103 cps. Even under such a condition, the old detector system suffered from nearly 5.5% dead time fraction, whereas the new detector system showed no dead time. These results indicated a better photon counting efficiency of the new detector system. This was also corroborated by higher photon counts in the Compton/gold XRF spectra registered by the new detector system, compared with the old detector system (Fig. 2(a)), during the same acquisition time (10 s). The overall effect as a consequence of this difference was the enhanced system sensitivity - the new detector system was able to collect nearly 50% more gold XRF photons, given the same acquisition time (10 s), than the old detector system for all GNP concentrations tested (Fig. 2 (b)). These improvements resulted in a lower (better) GNP detection limit (0.02 wt.%) achievable with the new detector system, compared with the GNP detection limit (0.03 wt.%) [13] achievable with the old detector system (Fig. 2 (b) inset). In terms of spectral resolution, however, both detector systems were able to perform comparably, providing ~0.7 keV FWHM at 69 keV photon energy (~ 1%) [13].

Fig. 2.

Fig. 2.

Open in a new tab

(a) The gold XRF/Compton spectra from 1 wt.% GNP calibration sample (only spectra > 40 keV is shown), and (b) the calibration curve. Data acquisition time was set to 10 s. The inset shows the lower concentration region of the calibration plot, where the horizontal lines represent 1.96 times the standard deviation of the backgrounds.

The difficulty in handling high x-ray flux using our experimental benchtop XRF/XFCT imaging system adopting the old detector system was evident from the reported greater than 50% dead time during the imaging of our small-animal-sized PMMA cylindrical phantom using a 2 mm aperture collimator [13]. To test the ability of both detector systems to detect and process high photon count rate, the detector collimator containing a larger (3 mm diameter) aperture was also tested. Under this setting, the photon count rates, per the new detector system, increased to as high as 3.5 × 104 cps. As expected, the old detector system quickly saturated (with greater than 93% dead time), suffered from significant photon count loss, and spectral resolution degradation (Fig. 3(a) inset). Thus, it was unusable for the imaging purpose. On the contrary, the new detector system was able to detect and fast process such a higher input rate as shown in Fig. 3, which demonstrated its capability in handling high x-ray flux input. Furthermore, the new detector system was found to work well with less than 8% dead time (well under the ideal operating condition of less than 50% dead time fraction) without any compromise in the spectral (energy) resolution.

Fig. 3.

Fig. 3.

Open in a new tab

(a) Comparison of the gold XRF plus Compton spectra from 1 wt.% GNP calibration sample obtained using the old detector system with a 2 mm aperture collimator for 10 s, and the new detector system with a 3 mm aperture collimator for 2.5 s and 10 s. The inset shows the spectrum obtained using the old detector system with a 3 mm aperture collimator for 10 s. (b) Calibration curve for the new detector system with a 3 mm aperture collimator and 10 s of data acquisition time. The inset shows the region of the calibration curve corresponding to lower GNP concentrations. As shown, the horizontal line represents 1.96 times the standard deviation of the background.

Fig. 3 also demonstrates further sensitivity enhancement of our experimental benchtop XRF/XFCT imaging system incorporating the new detector system and possibility of operating it to its full potential by allowing more x-ray photons passing through a large (3 mm) collimator. While the use of such a large collimator increased the Compton background, the efficient collection of more gold XRF photons reaching the detector facilitated the gold XRF peaks to rise faster from the background. This is evident in Fig. 3 (a), where the number of gold XRF photons scored by the new detector system using a 3 mm aperture collimator was found to be significantly greater than that scored by the old detector system using a 2 mm aperture collimator for the same 10 s of acquisition time. Most importantly, the number of gold XRF photons scored by the former setup for 2.5 s was higher than that scored by the latter setup for 10 s (Fig. 3 (a)). Consequently, the new detector system produced drastically higher gold XRF signal at a given GNP concentration (e.g., 9 times more for 1 wt.% GNP solution), leading to further reduction in the GNP detection limit (i.e., 0.01 wt.%) (Fig. 3 (b)). Thus, with the larger collimator, the net XRF signal increased thereby increasing the sensitivity of benchtop XRF/XFCT imaging system, which facilitated the detection of GNPs at lower concentrations (Fig. 3 (b)). Overall, the new detector system, characterized to work at higher bias voltage of 700 V and fast peaking time (3.2 μs), enhanced the sensitivity of benchtop XRF/XFCT imaging system.

As shown in Fig. 4 (a), the reconstructed XFCT image, which was obtained using the new detector system with a 2 mm aperture collimator, properly identified the regions loaded with low concentrations of GNPs (down to 0.02 wt.%). The CNR values corresponding to all three GNP-loaded regions within the reconstructed XFCT image were above the detectability threshold (CNR = 3) [13], which was in line with the GNP detection limit (0.02 wt.%) as determined from this study. Fig. 4 (b) shows the linearity (R2 = 0.999) between the average signal intensities within the GNP-loaded holes in the reconstructed XFCT image and GNP concentrations.

Fig. 4.

Fig. 4.

Open in a new tab

(a) Reconstructed XFCT image of a small-animal-sized phantom, after an XFCT scan with the new detector system coupled with a 2 mm aperture collimator. The table provides the CNR of the regions containing GNP solutions in the reconstructed XFCT image. (b) Linearity (R2 = 0.999) as shown in average signal intensity within the region of interest (10 × 10-pixel area within each GNP-loaded hole in the reconstructed XFCT image) vs. GNP concentration. The error bars represent the standard deviations of the pixel intensities within the regions of interest.

4. Discussion

As reported previously [13], the sensitivity of our original benchtop XRF/XFCT imaging system was enhanced considerably by incorporating a high-power x-ray source and optimizing the operation of a single crystal CdTe detector system. However, the low detection efficiency of the detector due to high dead time was a major hardware limitation that prevented the system from achieving further improvements. Therefore, the use of efficient detector systems capable of handling higher photon count rates were needed to take advantage of high photon flux from a high-power x-ray source.

The new detector system used in this study equipped with the latest firmware/software combination offered more flexibility in adjusting the detector operating parameters than the old detector system, enabling the detector further optimized for spectral data acquisition under higher photon flux. As a result, it allowed for further sensitivity enhancement of our original benchtop XRF/XFCT imaging system under the identical experimental conditions. Specifically, compared with our previous study conducted with the old detector system [13], the GNP detection limit was improved from 0.03 to 0.02 wt.%. As summarized in Fig. 4, this improvement in the system sensitivity (i.e., GNP detection limit) translated into the ability to image lower GNP concentration than before (i.e., the cylindrical hole loaded with 0.02 wt.% GNP was visible in the current work). Besides, the improved CNR was evident for the other two holes loaded with GNPs (at 0.05 and 0.1 wt. %) which could also be imaged with the old detector system but with lower CNR as shown in our previous study [13].

The 2 mm aperture detector collimator used in our experimental setup, however, restricted many photons from reaching the detector, preventing the new detector system from operating at its full capacity. Therefore, this study also investigated a collimator with larger (3 mm) aperture to test the ability of the new detector system in terms of handling high photon count rates. With the 3 mm aperture detector collimator, the new detector system was capable of further improving the system sensitivity by at least three times, given the same dose/data acquisition time. Thus, we would anticipate three times lower scan-time and delivered dose if we were to keep the same system sensitivity as that achievable from the existing system (adopting the 2 mm aperture detector collimator). The improved sensitivity (albeit reduced image resolution) from this approach will be beneficial for applications that focus on the detection of metal probes present at lower concentrations within the imaging objects.

5. Conclusion

The firmware/software for the new detector system, compared with that for the old detector system, allowed further optimization of the CdTe detector setting, resulting in improved photon counting efficiency under high photon flux. Consequently, the system sensitivity of our experimental benchtop XRF/XFCT imaging system was improved by at least 1.5 times, after adopting the new detector system. It was further improved, at least by three times, after incorporation of the detector collimator with a larger (3 mm) aperture (vs. 2 mm aperture). These improvements are noteworthy, contributing to further capacity enhancement of benchtop XRF imaging systems utilizing the same type of single crystal CdTe detector systems as investigated in this work.

Acknowledgments

This investigation was supported by the NIH under the award number R01EB020658.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Credit Author Statement

Hem Moktan: Investigation, Writing – Original Draft.

Sandun Jayarathna: Investigation, Writing – Review & Editing.

Sang Hyun Cho: Writing – Review & Editing, Supervision.

References

  • [1].Roberts WM, Estimation of strontium in animal bone using x-ray fluorescence analysis, Nature, 183 (1959) 887. [DOI] [PubMed] [Google Scholar]
  • [2].Beckhoff B, Kanngießer B, Langhoff N, Wedell R, Wolff H, Handbook of Practical X-Ray Fluorescence Analysis, Springer, Berlin: 2006. [Google Scholar]
  • [3].Pushie MJ, Pickering IJ, Korbas M, Hackett MJ, George GN, Elemental and chemically specific X-ray fluorescence imaging of biological systems, Chemical Reviews, 114 (2014) 8499. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [4].Cheong SK, Jones BL, Siddiqi AK, Liu F, Manohar N, Cho SH, X-ray fluorescence computed tomography (XFCT) imaging of gold nanoparticle-loaded objects using 110 kVp x-rays, Physics in Medicine & Biology, 55 (2010) 647. [DOI] [PubMed] [Google Scholar]
  • [5].L Jones B, Manohar N, Reynoso FJ, Karellas A, Cho SH, Experimental demonstration of benchtop x-ray fluorescence computed tomography (XFCT) of gold nanoparticle-loaded objects using lead- and tin-filtered polychromatic cone-beams, Physics in Medicine & Biology, 57 (2012) N457. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [6].Manohar N, Jones BL, Cho SH, Improving x-ray fluorescence signal for benchtop polychromatic cone-beam x-ray fluorescence computed tomography by incident x-ray spectrum optimization: A Monte Carlo study, Medical Physics, 41 (2014) 101906. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [7].Ren L, Wu D, Li Y, Wang G, Wu X, Liu H, Three-dimensional x-ray fluorescence mapping of a gold nanoparticle-loaded phantom, Medical Physics, 41 (2014) 031902. [DOI] [PubMed] [Google Scholar]
  • [8].Hertz JC, Larsson U, Lundstrom DH, Larsson C, Vogt, Laboratory x-ray fluorescence tomography for high-resolution nanoparticle bio-imaging, Optics Letters, 39 (2014) 2790. [DOI] [PubMed] [Google Scholar]
  • [9].Sjölin M, Danielsson M, Improved signal-to-noise ratio for non-perpendicular detection angles in x-ray fluorescence computed tomography (XFCT), Physics in Medicine & Biology, 59 (2014) 6507. doi: 10.1088/0031-9155/59/21/6507. [DOI] [PubMed] [Google Scholar]
  • [10].Ahmad M, Bazalova-Carter M, Fahrig R, Xing L, Optimized Detector Angular Configuration Increases the Sensitivity of X-ray Fluorescence Computed Tomography (XFCT), IEEE Transactions on Medical Imaging, 34 (2015) 1140–1147. [DOI] [PubMed] [Google Scholar]
  • [11].Bazalova-Carter M, Ahmad M, Xing L, Fahrig R, Experimental validation of L-shell x-ray fluorescence computed tomography imaging: phantom study, Journal of Medical Imaging, 2 (2015) 043501. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [12].Manohar N, Reynoso FJ, Diagaradjane P, Krishnan S, Cho SH, Quantitative imaging of gold nanoparticle distribution in a tumor-bearing mouse using benchtop x-ray fluorescence computed tomography, Scientific Reports 6 (2016) 22079. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [13].Manohar N, Reynoso FJ, Cho SH, Technical Note: A benchtop cone-beam x-ray fluorescence computed tomography (XFCT) system with a high-power x-ray source and transmission CT imaging capability, Medical Physics, 45 (2018) 4652. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [14].Pantazis T, Pantazis J, Huber A, Redus R, The historical development of the thermoelectrically cooled X-ray detector and its impact on the portable and hand-held XRF industries, X-Ray Spectrometry, 39 (2010) 90. [Google Scholar]
  • [15].Ahmed MF, Yasar S, Cho SH, A Monte Carlo Model of a Benchtop X-Ray Fluorescence Computed Tomography System and Its Application to Validate a Deconvolution-based X-Ray Fluorescence Signal Extraction Method, IEEE Transactions on Medical Imaging, 37 (2018) 2483. [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES