Comparison of quantitative imaging characteristics between a new, larger-FOV 1000 fps high-speed angiographic (HSA) photon-counting detector (Aries) with a smaller HSA detector (Actaeon)

Abstract

High Speed Angiography (HSA) requires imaging detectors with both high-temporal and high-spatial resolution. Both the Aries and Acteon detectors by Direct Conversion (Stockholm, Sweden) are CdTe direct photon-counting detectors (PCD) that have acquisition frame rates of up to 1000-fps and a 100-micrometer pixel pitch; however, the new Aries detector offers a larger field of view (512 × 768 pixels) compared to the smaller Actaeon detector (256 × 256 pixels). An expanded field of view is required for imaging of larger vasculature, thus the Aries offers this advantage. Evaluations were performed of both detectors under Anti-Coincidence Circuitry (ACC-ON) mode, which corrects for charge sharing between pixels. Initial evaluations of instrumentation noise and detector energy-threshold calibration using Am-241 gamma spectroscopy were performed for the new Aries detector. Linearity was also evaluated for the Aries for each of the 12 individual modules that compose the detector field to check for homogeneity in response to exposure throughout the detector. Finally, Normalized Noise Power Spectrum (NNPS), Modulation Transfer Function (MTF) and Detective Quantum Efficiency (DQE) were then compared between the Aries and Actaeon detectors at two different exposures and detector energy thresholds. The detectors are linear up to approximately 1000 μR and have no instrumentation noise above a threshold of 15 keV. As expected, the MTF’s and DQE’s are similar between the Aries and Actaeon detectors, and there are thus no tradeoff’s in image quality to achieve the larger FOV.

1. INTRODUCTION

In traditional angiography, frame rates of 30 fps limit the vascular flow information that can be obtained from images. With 1000 fps High-Speed Angiography (HSA) detectors, detailed flow data can be produced; however, given the same exposure rate, each HSA detector pixel absorbs fewer photons over the 1-ms acquisition time than would at lower frame rates with longer acquisition such as in traditional angiography. This then requires HSA detectors to additionally provide high spatial resolution and low noise. Furthermore, larger and complex anatomical structures also require a detector with a large field of view. Compared to the 256 × 256 100-micron pixel pitch Actaeon PCD CdTe detector, the Aries 100-micron pixel pitch PCD CdTe detector offers the same 1000 fps frame rate with a larger field of view of 768 × 512 pixels. Evaluation of detector metrics such as MTF, NNPS, and DQE are essential for quantifying the spatial resolution, noise, and performance of high-speed detectors [5].

2. MATERIALS AND METHODS

2.1. High-Speed Photon-Counting Detectors

Both the Actaeon and Aries (Direct Conversion, Sweden) are photon-counting direct detectors (PCD) with a 0.75-mm thick CdTe photon to electron convertor layer. PCD’s count the number of photons between a number of energy windows. The energy discrimination abilities of PCD’s allow for more tissue specific images and the capability of simultaneous dual energy images. [1]. These detectors also both have a 100 micron pixel pitch and can acquire up to 1000 images per second. Their difference then lies in their field of views, with the Aries having a 768×512 pixel readout and the Actaeon having a 256×256 pixel readout. A range of lower energy thresholds can be set for both the Aries and Actaeon, from 0 keV to 100 keV, where photons registered at energies below a set threshold are not counted. Evaluations of the detectors were made under anti-charge coincidence mode, which prevents charge sharing of a single event between the neighboring pixels [6].

2.2. Instrumentation Noise

For initial evaluations of the new Aries detector, we looked at the instrumentation noise energy threshold. Two different 1000-frame sets of 1-ms dark field images were taken with varying detector thresholds. One set of dark field images had no x-ray beam on, while the other set had a pulsed fluoroscopic x-ray beam on at 80 kV, 20 mA, 15 exposures per second with a 10-ms pulse width, and RQA5 spectrum with 5-mm lead shielding between the x-ray source and detector. Photon counts of each sequence were summed and Log10 mean counts of the summed sequence was then graphed at each threshold. After instrumentation noise evaluation, a 20 keV threshold was set as a conservative value for the instrumentation noise threshold for the rest of the analyses.

2.3. Gamma Spectroscopy

Verification of the energy threshold calibration was then done using an Am-241 button source, which has a monoenergetic peak of 60 keV. The Am-241 source was placed 30 cm away from the detector. Acquisitions of 500 frame images were taken with counts taken for every 2-keV threshold interval, and counts for each bin were compared to the Am-241 peak [2].

2.4. Linearity

Linearity of exposure and photon counts was evaluated for the entire field of view, as well as for each of the 12 modules that compose the Aries detector to check that response was both linear and uniform throughout. The x-ray tube was set on fluoroscopy mode at 70 kVp with an RQA5 spectrum whilst the mA varied from 2 to 400. Each acquisition had 1000 frames at 1-ms per frame. Average counts per frame for the total field of view and for each module were graphed versus exposure.

2.5. Standard Metrics

For detector metric comparison of the Aries detector and the Actaeon detector, the images and exposures were taken at 70 kV with an RQA5 spectrum for a 500-ms duration. Images were obtained at two thresholds, 20 keV and 32 keV, and two exposure levels, 100 mA and 160 mA. The 20 keV threshold reflects the need to eliminate instrumentation noise while preserving most counts, while the 32 keV level tests a higher threshold level, especially for the use of its dual energy capabilities that can also enable material decomposition. The modulation transfer function (MTF) was obtained using the Edge Spread Method [3]. An edge was placed in the middle of the field of view at an angle of 1.2 degrees, and the images acquired were flat field corrected. The Line Spread Function (LSF) was obtained by taking the derivative of the edge spread function (ESF). The Fourier Transform (FT) of the normalized LSF was taken to get the MTF for frequencies up to the Nyquist frequency of 5 mm⁻¹. The normalized noise power spectrum (NNPS), or Weiner spectrum, was obtained from the FT of flat field-corrected images [4]. Normalized horizontal, vertical, and radial NPS results were averaged to find the overall NNPS up to the Nyquist frequency. For the calculation of detector quantum efficiency (DQE), input exposure measurements were taken with the PTW TN34069 6 cc ionization chamber for both 100 mA and 160 mA, and the q-value for the RQA5 spectrum was obtained from IEC 62220-1.

$$\mathit{\text{DQE}}=\frac{\mathit{\text{MTF}}{\left(f\right)}^2}{\mathit{\text{NNPS}}\left(f\right)*\mathit{\text{Exposure}}\left(\mathit{\text{mR}}\right)*q}$$ (1)

3. RESULTS

3.1. Instrumentation Noise and Energy Calibration

As shown in Figure 1, instrumentation noise for the Aries is negligible at thresholds above 15 keV for both dark field measurements. A significant drop-off in Log10 counts occurs between 12 keV and 14 keV. To be precautionary, a 20 keV threshold is set for the instrumentation noise threshold for future evaluations.

Figure 1:

Graph of Log10 total counts with varying detector threshold settings for the Aries PCD.

Spectroscopy measurements are shown in Figure 2; compared to the monoenergetic gamma spectrum peak at 60 keV, the peak for the detector is in the 54 keV to 56 keV range. Counts in the detector become negligible at counts above 60 keV. The energy resolution was determined by the Full Width at Half Maximum (FWHM) to be around 10 keV or 17%.

Figure 2:

Graph of total counts from the Am-241 source within increasing 2-keV intervals.

3.2. Detector Linearity

With regards to linearity, Figure 3A shows R-squared values calculated from a linear fit with the y-intercept set to zero 0. The detector is relatively linear up to around 1000 μR, where then it increasingly becomes more saturated from pulse-pileup and becomes less linear. In Figure 3B, detector response is relatively linear below this point with an R-squared of .99, indicated by the blue portion in that figure. The individual modules are relatively homogenous in their response and start deviating more so only at very high exposures. In Figure 4, we took the Coefficient of Variation (COV) of total counts for each exposure amongst the 12 modules of the Aries detector to quantify homogeneity across modules. We see that COV increases linearly with exposure.

Figure 3:

Linearity of mean counts versus exposure for the entire (3A) Aries FOV and 3(B) individual Aries modules. In 3A, the black lines are the lines of best fit, with R-squared values displayed. The blue line indicates points up to where R-squared is above 0.99.

Figure 4:

Coefficient of variation of total counts for all 12 Aries detector modules per exposure.

3.3. Standard Metric Evaluation Comparison

As shown in Figures 5, 6 and 7, the MTF’s and DQE’s are similar between both the Aries and Actaeon detectors, with the QDE or DQE(0) for both detectors between .72 and .81. Higher exposure mA’s decrease the NNPS for each frequency and did not significantly impact the final DQE for the detectors in any particular direction when factoring in increased exposure in the DQE calculation, as expected. DQE’s of all 32-keV threshold were all slightly lower than their 20 keV counterparts, as more photons are excluded from the higher thresholds, resulting in a higher NNPS.

Figure 5:

Graph of MTF versus spatial frequency with varying detector threshold and exposure

Figure 6:

Graph of NNPS versus spatial frequency with varying detector threshold and exposure mA.

Figure 7:

Graph of DQE versus spatial frequency with varying detector threshold and exposure mA.

4. DISCUSSION

The new Aries detector is confirmed to be relatively linear for 1000 uR or below for the entire detector FOV. At the higher limits, it deviates more from a linear curve in a similar fashion to other non-paralyzable detectors, but these exposure rates are not typical for medical settings nor other in-vivo studies. The 12 separate modules appeared to show uniform response, with a covariance that is less than 0.06 even at high exposure rates. The differences in their responses are due to inherent module-by-module differences in saturation rate, and so there would be higher pulse pileup in certain modules as compared to others. A linear relationship with COV occurs under the given exposure conditions as all modules have increased probability of a detector pixel having pulse pileup, given their slightly different saturation rates when exposure increases. We have not reached a point close to full saturation for all pixels per module, where the COV would remain unchanged with increased exposure. The COV of counts for the modules can be easily corrected dividing the ratio of modules’ average counts given in figure 3B and the full detector average counts given in figure 3A. Compared with the Actaeon, the performance of the Aries was extremely close. The MTF and DQE was very similar for the two detectors, with the main differences in measured DQE being accountable to the differences in exposure and threshold such that the lower threshold provided a higher DQE, while the different exposures and thresholds accounted for the difference in NNPS, with very little difference between detectors when the same parameters were used. Given the close agreement between detectors in the measured image quality metrics, it is reasonable to conclude that the Aries detector can provide images of larger anatomical structures without sacrificing performance.

5. CONCLUSION

The Aries 1000 fps PCD detector has a relatively accurate energy threshold setting, a linear response to exposure, and has comparable spatial resolution, Weiner spectrum, and overall performance to the Actaeon detector. With the larger field of view, the Aries has the added advantage of imaging larger and more complex structures.