Photon-counting detector CT: early clinical experience review

Abstract

Since its development in the 1970s, X-ray CT has emerged as a landmark diagnostic imaging modality of modern medicine. Technological advances have been crucial to the success of CT imaging, as they have increasingly enabled improvements in image quality and diagnostic value at increasing radiation dose efficiency. With recent advances in engineering and physics, a novel technology has emerged with the potential to surpass several shortcomings and limitations of current CT systems. Photon-counting detector (PCD)-CT might substantially improve and expand the applicability of CT imaging by offering intrinsic spectral capabilities, increased spatial resolution, reduced electronic noise and improved image contrast. In this review we sought to summarize the first clinical experience of PCD-CT. We focused on most recent prototype and first clinically approved PCD-CT systems thereby reviewing initial publications and presenting corresponding clinical cases.

Introduction

X-ray CT has emerged as a landmark diagnostic imaging modality of modern medicine since its development in the 1970’s. The benefits and clinical demand for CT imaging continue to rise due to continued technical advancements facilitating improvements in image quality and diagnostic value at constantly higher radiation dose efficiency, together with broadening of the clinical indications. ^1–4

The X-ray detector constitutes a decisive component of a CT system that critically influences image quality and dose efficiency. Accordingly, the detector design has undergone radical transformations and improvements. Currently, most available CT systems use solid-state energy-integrating detectors (EIDs) with third-generation rotate–rotate designs to convert X-rays to electrical signals. The indirect conversion technology of EIDs is based on a scintillator layer converting X-ray photons to visible light, which is then detected by a photodiode layer and converted into an electric output signal that is proportional to the total energy deposited during a measurement interval. ^4–8 As the detector element integrates the energy from all photons, the output electrical signal does not convey any information on the energy of individual photons. Ultimately, in a substrate layer the generated signal is transmitted to analog electronics for amplification. ^6,9

Despite its continuous merits, EIDs exhibit some inherent limitations. First, the detector design and more importantly the detector element size of current CT systems limits its maximum achievable spatial resolution. The maximum achievable spatial resolution of a CT system depends on both the size of the focal spot of the X-ray tube and the size of the detector elements—both must be matched and approximately equal. The spatial resolution as a function of the spatial frequency (in line pairs per cm) is described by the so-called modulation transfer function (MTF) and shows the relative contrast with which small periodic structures are represented in the image. The MTF can be modified by choosing different convolution kernels, but in doing so, the resolution limit of the measurement system cannot be exceeded. The ultimate resolution limit is reached at a spatial frequency equal to 1/(2*pixel size) (Nyquist theorem)—so detectors must be made smaller to improve spatial resolution. This is a problem for EID detectors, because the individual elements have to be separated by thin septa to avoid optical cross-talk. With smaller detector elements, more area is occupied by the optically isolating septa, which absorb X-rays without contributing to the detector signal, resulting in lower dose efficiency. ^5,7,10 Second, electronic readout noise stemming from the analog electronic circuits remains a problem of EIDs. In case of high photon flux on the detector, the effect of electronic readout noise on image quality is negligible. At low to very low radiation doses, however, the number of detectable photons is low and electronic readout noise may become apparent thus degrading image quality. ⁶ Third, each photon contributes to the total input detector signal with an amount that is proportional to its own energy. Thus, the relative contribution of high-energy photons to the total signal is higher than that of low-energy photons. This underweighting of low-energy photons can be suboptimal as low contrast differences are most prominent at low X-ray energies. Contrast-enhanced CT scans in particular show suboptimal iodine contrast. ^5,6 Fourth, most current high-end CT systems offer dual-energy (DE) applications. DECT enables functional imaging by exploiting material-specific differences in X-ray attenuation at different X-ray energies. ^11,12 DECT data can be generated by means of various technologies including dual-source (DS), rapid kV switching, and dual-layer (DL) detector technology. However, each of these technologies has its specific set of inherent limitations such as imperfect spatial or temporal registration of data sets, field of view (FOV) restrictions and limitations in terms of tube voltage selection and the use of tube current modulation. Furthermore, for most of these technologies, DECT data are only available if a specific workflow has been selected prior to image acquisition. ¹³

With recent advances in engineering and physics, a novel technology has emerged with the potential to surpass many of the shortcomings and limitations of current CT systems. Photon-counting detector (PCD-) CT is a promising technology that might substantially improve and expand the applicability of CT imaging.

In this review, we want to summarize the basic technical features and the initial experience with PCD-CT by reviewing first publications with prototype and clinical systems and try to illustrate with case examples how this emerging technology may translate into improved clinical diagnostics.

Photon-counting detector CT

Detailed reviews about the technical principles of PCD-CT systems have been published elsewhere. ^5–7,10 Therefore, we will not provide too much detail on these aspects but will rather focus on basic specifications and the first experience with the new technology.

Basic physical principles of the detector

PCDs differ considerably from EIDs. In contrast to EIDs, which require a separate scintillator layer to convert X-rays to light, PCDs use a single layer of a semi-conductor made of cadmium telluride (CdTe), cadmium zinc telluride (CZT), or silicon. A large bias voltage is applied between a cathode on top of and pixelated anodes at the bottom of the semi-conductor. Each incident X-ray photon produces a cloud of positive and negative charges which are separated in the strong electric field and pulled away from each other rapidly. The electrons move towards the anodes to generate an electric signal that is registered by an attached electronic readout circuit. Thus, with PCDs X-ray photons are directly converted into an electrical signal and each photon leads to an electrical pulse whose amplitude is directly proportional to the energy of the photon. The PCD then counts the number of pulses to quantify the number of incident X-ray photons and compares the amplitude of each pulse to several pre-set threshold levels that are set by means of multiple electronic comparators and counters. Specifically, an initial threshold is set at a level that is higher than the electronic noise level but lower than the pulses of incident photons (e.g. at 25 keV). Furthermore, as all pulses are additionally compared to further threshold levels, photons can be assigned to energy bins depending on their energy. ^5–7,10

The detector design of PCDs relying on a direct conversion technology overcomes the above-mentioned issues of EIDs. First, by eliminating the scintillator material and consequently the optically isolating septa, the detector elements of PCDs can be made much smaller, thus offering improved spatial resolution. Second, by thresholding incoming photons according to their energies, electronic noise can be eliminated and spectral (i.e. dual- or multienergy) imaging becomes inherently available. Third, as the detective quantum efficiency (DQE) of a photon counting detector is approximately constant as a function of X-ray energy, ⁵ there is no underweighting of low-energy X-ray photons as with EIDs, and image contrasts can be improved. ^5,6 In a practical detector design, the thickness of the semi-conductor layer of a PCD has to be chosen large enough to provide a total DQE similar to EIDs. Thin layers of about 1.4–2 mm are sufficient for CdTe or CZT because of their high atomic number. Si with its low atomic number and low absorption efficiency in the X-ray energy range relevant to medical CT requires thick layers >30 mm. ^5,6 Still, the exact total DQE strongly depends on the detector design.

Temporal evolution

Computational power

Until recently, the use of PCDs has mainly been limited to nuclear imaging as the X-ray photon-count rate in CT imaging is much higher than in nuclear medicine. ¹⁴ Specifically, the PCD has to register each incoming X-ray photon before the next one arrives. If the PCD does not manage this, a pile-up occurs where the photons can no longer be separately registered and the count rate is no longer proportional to the X-ray flux, leading to image quality degradation. A further challenge concerns the cross-talk between detector elements. X-ray photons hitting the detector close to the border of a detector element produce charge clouds that spread out and may be wrongfully registered by more than one detector element. This results in a loss of spatial resolution and spectral separation. By implementing faster readout electronics, smaller pixel sizes, subdivided detector elements, and optimized electronic circuits that can detect coincidental registration of photons, these hurdles were recently overcome, making industrial production of PCD-CT possible. ^5,6,10

PCD-CT systems

A series of prototype PCD-CT systems enabled pre-clinical research but were all limited in one way or another. One prototype system was a whole-body research PCD-CT system built (SOMATOM Count, Siemens Healthineers, Forchheim, Germany) using a modified DS CT platform (SOMATOM Definition Flash, Siemens Healthineers), with the B-subsystem equipped with a CdTe PCD array. 2 × 2 subpixels of the photon counting detector can be binned to a “sharp pixel” or “ultra-high resolution (UHR) pixel” with a pixel size of 0.45 × 0.45 mm² (0.25 × 0.25 mm² at the isocenter), 4 × 4 subpixels can be binned to a “MACRO pixel” with a size of 0.9 × 0.9 mm² (0.5 × 0.2 mm² at the isocenter) comparable to today’s medical CT systems. A detailed description of this system can be found elsewhere. ^5,15–20 The main limitations of this scanner included a detector z-coverage of 8–16 mm depending on the acquisition mode, an in-plane FOV of the PCD array of 27.5 cm at the isocenter, and a lack of angular tube current modulation (for comparison: a conventional SOMATOM Definition Flash scanner has a detector z-coverage of 38.4 mm and a maximum in-plane FOV of 50 cm ²¹ ). In addition, due to the hybrid design, DS PCD scanning at high temporal resolution was not possible. Later, a further investigational whole-body full FOV single-source PCD-CT system (SOMATOM Count Plus, Siemens Healthineers, Forchheim, Germany) was developed thereby overcoming most of the key limitations of the hybrid DS scanner by offering a 50 cm scan FOV, 57.6 mm longitudinal detector coverage as well as automatic exposure control in both angular and longitudinal directions. ²² Recently, another prototype single-source CT scanner with a full FOV silicon-based PCD (modification of a commercial Lightspeed VCT scanner, GE Healthcare, Chicago, Illinois, USA) was presented with a phantom study. ²³ Another, further advanced PCD-CT prototype system based on a CZT detector (modified Brilliance iCT scanner, Philips Healthcare) and capable of human imaging was recently developed. ^24,25 This system enables cardiac imaging on a level and beyond the technical performance of a comparable EID-CT system. ^24,26 This prototype relies on a single-layer of energy-sensitive PCDs of 2 mm thick CZT with five adaptable energy thresholds set at 30, 51, 62, 72 and 81 keV and a detector element size of 0.27 × 0.27 mm² at the isocenter. An in-plane FOV of 50 cm and a z-coverage of 1.76 cm (64 × 0.275 mm at the isocenter) is offered. Tube voltage can be set at 80, 100, 120 or 140 kV, and tube current can be modulated between 10 and 500 mAs. The system has a focal spot of 0.6 × 0.7 mm and a gantry rotation time of 0.33–1 s for 2400 projections per rotation.

Importantly, however, a PCD-CT system has recently been cleared for clinical use (NAEOTOM Alpha; Siemens Healthineers/ FDA approval September 30, 2021). A detailed description of this CT system has been published recently. ⁸ In brief, this scanner exhibits a DS geometry with a minimum gantry rotation time of 0.25 s offering a temporal resolution of 66 ms. The system uses two dedicated PCDs with 1.6 mm thick CdTe. Each detector element has a size of 0.151 × 0.176 mm² projected to the isocenter. The detector pixels can be read out either independently, thus allowing for UHR imaging with a z-coverage of 24 mm (120 × 0.2 mm at the isocenter) or can be binned into 2 × 2 groups (resulting pixel size 0.302 × 0.352 mm² at the isocenter) for standard imaging with a z-coverage of 57.6 mm (144 × 0.4 mm at the isocenter). Scans acquired in the spectral mode use fixed energy bin thresholds of 20/35/55/70 keV and offer full spectral image information. The system is equipped with two Vectron tubes each with 120 kW power output. To match the spatial resolution of the detector, the Vectron tubes offer several focal spots with dimensions down to 0.4 mm x 0.4 mm² (0.181 × 0.181 mm² at the isocenter) for UHR scanning.

Images can be reconstructed with a range of slice thicknesses as low as 0.2 mm depending on the protocol and with various matrix sizes (512 × 512, 768 × 768, 1024 × 1024). ^8,22,27 A novel iterative reconstruction (IR) algorithm named Quantum Iterative Reconstruction (QIR) that is suitable for the reconstruction of spectral imaging data ²⁸ has been introduced. As with previous IR algorithms, higher strength levels lead to greater noise reductions. ^27,28 A representative image example illustrating the performance of the novel IR algorithm is provided in Figure 1.

Figure 1.

63-year-old male patient (body weight 77 kg) with multifocal hepatocellular carcinoma. Contrast-enhanced abdominal portal venous phase scans were acquired on a clinical PCD-CT system (NAEOTOM Alpha, Siemens Healthineers) in the spectral imaging mode at 120 kV with a tube current of 86 mAs. The CTDI_vol was 6.7 mGy. Virtual monoenergetic images at 60 keV were reconstructed with a 2 mm slice thickness without Quantum Iterative Reconstruction and with all strength levels of QIR (QIR 1–4). CTDI_vol, volume of CT dose index; PCD, Photon-counting detector; QIR, Quantum Iterative Reconstruction

Translating the benefits of PCD-CT into clinical routine

Improved spatial resolution without dose penalty

The lack of a scintillator layer and the use of subdivided detectors in a PCD enables dose-efficient UHR imaging. A range of technological approaches have previously enabled UHR imaging for EID-CT including the use of very small detector element sizes or the implementation of comb filters to reduce detector aperture in the z-axis and/or in-plane. All these approaches, however, were associated with manufacturing difficulties or reduced dose efficiency of up to 50%. ^27,29,30

The detector design of the current PCD-CT systems bypass these difficulties, by exhibiting a 125 µm limiting (4.00 lp/mm) (NAEOTOM Alpha, Siemens Healthineers) ⁸ to 178 µm limiting (2.81 lp/mm) (Philips Healthcare prototype) in-plane spatial resolution. ^24,25 The limiting spatial resolution is defined as the spatial frequency at 0% MTF for the sharpest convolution kernel available at the CT system. For comparison, the limiting spatial resolution of a comparable EID system (SOMATOM Force, Siemens Healthineers) is 240 µm (2.08 lp/mm).

Based on the UHR mode of a prototype PCD-CT system with 150 µm limiting spatial resolution (3.33 lp/mm) (SOMATOM Count) , Leng et al were able to show that the UHR mode with a pixel size of 0.25 × 0.25 mm² at the isocenter exhibited a 87% improvement in spatial resolution (10% MTF) using the sharpest available convolution kernel as compared to the so-called MACRO scan mode of the same system with 0.5 × 0.5 mm² pixel size (i.e. standard-resolution PCD-CT) comparable to today’s medical CT systems. Alternatively, a 15% reduction in image noise could be achieved at the same in-plane spatial resolution (sharpest kernel available in the MACRO mode) for scans acquired with the UHR scan mode as compared to the MACRO scan mode. ^6,15 Klein et al and Pourmorteza et al confirmed that the UHR mode achieves lower noise levels than the standard MACRO mode at comparable spatial resolution because detector cell binning is avoided. This further highlights the high dose efficiency of the PCD-CT UHR mode. ^18,31

Importantly, using dedicated phantom experiments, cadaveric and patient scans, both Leng et al ^15,32 and Zhou et al ³³ provided evidence that the UHR PCD-CT mode outperforms the comb-filter UHR EID-CT mode when using comparable imaging and reconstruction parameters. The latter study even demonstrated 40% less noise for PCD-CT UHR-based temporal bone imaging as compared to comb-filter UHR EID-CT imaging at matching radiation doses and reconstruction settings. ³³

A variety of further studies have been published on HR and UHR PCD-CT imaging highlighting its value for lung assessment, ^{15,25,27,34–39} cardiac imaging ^{20,24,40–44} and musculoskeletal applications. ^{19,33,45–48}

UHR imaging is of particular interest for these anatomical areas because the clear depiction of fine details may effectively enhance diagnostic accuracy. In case of lung imaging, fine parenchymal changes may be better detected, delineated and characterized. ^15,36–38 Inoue et al scanned 30 patients with suspicion of interstitial lung disease on a research PCD-CT system in the UHR mode with optimized reconstruction settings and on standard-of-care EID-CT systems. PCD-CT improved reader’s confidence for the presence of imaging findings of reticulation, ground-glass opacities, and mosaic pattern as determined by three thoracic radiologists. Furthermore, reader confidence in the probability of usual interstitial pneumonia increased for one of the three thoracic radiologists. Lastly, overall image quality and sharpness of PCD-CT images was deemed improved despite the slightly lower radiation dose (median CTDIvol of 6.49 mGy for PCD-CT vs 7.88 mGy for EID-CT). ³⁸ A further clinical study with 80 systemic sclerosis patients showed that the UHR mode of the first clinical PCD-CT system maintains image quality and diagnostic accuracy for the assessment of interstitial lung disease at only 33% of the dose of comparable EID-CT scans performed on a third generation (i.e. latest generation) DS EID-CT scanner. ³⁹

In regard to coronary artery imaging, plaque visualization and stent imaging can be improved. ^24,40 Two recent studies assessing the UHR scan mode of clinical PCD-CT in patients referred for coronary CT angiography have shown that coronary arteries can be visualized in excellent quality with improved visualization of non-calcified plaque components and with reduced blooming of calcified plaques. ^43,44 For bone imaging, fine bone details and previously occult hairline fractures may become apparent. ^33,46,49,50

In a clinical feasibility study including 32 patients who underwent both UHR PCD-CT imaging on a prototype CT system with optimized reconstruction settings and standard-of-care EID-CT imaging, Baffour et al reported improved visualization of osseous structures of the pelvis and shoulder for UHR PCD-CT imaging at a 31–47% lower radiation dose. A further study on 29 multiple myeloma patients showed that the visualization of lytic bone lesions, medullary lesions and fat attenuation in myeloma lesions could significantly be improved with a clinical PCD-CT in the UHR mode as compared to standard-of-care EID-CT systems with matching reconstruction parameters and radiation dose. ⁵¹ Further clinical studies will demonstrate whether or not the improvements in anatomic display will also impact on therapeutic management and ultimately, on patient outcome. ⁵²

Representative PCD-CT cases scanned with the UHR mode illustrating the benefits for coronary CT angiography and upper ankle joint imaging can be found in Figures 2–3.

Figure 2.

65-year-old male patient (body weight 94 kg) with calcified coronary plaques in the left main and left anterior descending coronary artery with an Agatston Score of 245. Coronary CTA images were acquired in the UHR mode (z-coverage of 24 mm) on a clinical dual source PCD-CT system (NAEOTOM Alpha, Siemens Healthineers) at 120 kV with a tube current of 61 mAs. The gantry rotation time was 0.25 s, with a temporal resolution of 66 ms. The CTDI_vol was 46.3 mGy. Images at different slice thicknesses and kernels were reconstructed. Note the improved sharpness of anatomical structures, vessels, and calcified coronary plaques on UHR images reconstructed with the Bv64 kernel and 0.2 mm section thickness. CTA, CT angiography; CTDI_vol, volume of CT dose index; PCD, Photon-counting detector; UHR, ultra-high resolution.

Figure 3.

32-year-old male patient (body weight 64 kg) presenting with a depression fracture of the talus. The fracture gap extends to the lateral part of the talus. Images were acquired in the UHR mode (z-coverage of 24 mm) on a clinical PCD-CT system (NAEOTOM Alpha, Siemens Healthineers) at 120 kV and with a tube current of 40 mAs. Radiation dose was 3.23 mGy CTDI_vol. Images were reconstructed with a Br56 kernel, 2 mm section thickness and a 512 × 512 matrix size, with a Qr68 kernel, 1 mm section thickness and a 512 × 512 matrix size and with a sharp Qr72 kernel, 0.2 mm section thickness (UHR image) and 1024 × 1024 matrix size. The left image represents the default setting for bone imaging, while the right image leverages the potential of PCD-CT UHR imaging. Note the excellent visualization of bone features and trabeculae and the exquisite delineation of the small fracture gap. CTDI_vol, volume of CT dose index; PCD, Photon-counting detector; UHR, ultra-high resolution.

Improved CNR and noise properties

Improved contrast-to-noise ratio (CNR) and noise performance stemming from factors such as lack of electronic noise or equal weighting of all X-ray photons (no underweighting of low energy photons) are a key feature of PCD-CT systems enabling radiation and contrast media dose reduction for routine clinical imaging. ⁵³

Using the first clinical PCD-CT system, Liu et al quantified the ability of PCD-CT to eliminate electronic background noise, with it achieving mean percent noise reductions of up to 74% at a radiation dose level of 0.4 mGy CTDI_vol compared to a third generation (i.e. latest generation) DS EID-CT system. ⁵⁴ Rajagopal et al investigated the technical performance of a prototype PCD-CT system for low dose abdominal CT imaging in a phantom ⁵⁵ and found that both for PCD-CT and EID-CT spatial resolution as a function of noise and contrast remained unaffected by dose while PCD-CT achieved a 22–24% improvement in noise across four radiation dose levels ranging from 1.7 to 6 mGy CTDI_vol. Consequently, this improved noise performance could be translated to a 29–41% improvement in CNR and a 20–36% improvement in detectability index. Using the first clinical PCD-CT, Racine et al confirmed that PCD-CT outperforms third generation (i.e. latest generation) DS EID-CT for the detection of hypo- and hyperattenuating focal liver lesions across a wide range of radiation dose levels. ⁵⁶ Gutjahr et al showed that iodine CNR was improved by 11–38% for a prototype PCD-CT system relative to EID-CT at matching scan and tube voltage settings of 80–140 kV. ²¹ This was further confirmed by Sawall et al who showed that with a prototype PCD-CT, dose-normalized iodine CNR could be improved by up to 37% relative to EID-CT, thus potentially enabling a radiation dose reduction of up to 46%. ⁵⁷ More recently, Booij et al investigated the iodine CNR benefits of the first clinical PCD-CT on VMI relative to a third generation (i.e. latest generation) DS EID-CT system run in DE mode with largely matching imaging and reconstruction settings. The authors confirmed that the CNR benefits of PCD-CT were also applicable for VMI, with low keV images below 60 keV exhibiting a 55–75% higher CNR than their EID-CT based counterparts depending on the tube voltage settings. ⁵⁸

Further experiments on sinus and temporal bone imaging with a prototype PCD-CT have shown significant radiation dose reductions between 56 and 85% relative to EID-CT depending on the exact imaging protocol without compromising image contrast and image noise. ¹⁹ Specifically, experiments were performed using an UHR PCD-CT mode with an additional tin (Sn) filter at 100 kV tube voltage. Besides the use of the tin filter, the authors also cite the advantages of the UHR PCD-CT mode with better intrinsic detector resolution and the detector design of PC detectors that eliminates the need for comb/grid filters for UHR imaging as reasons for the improved performance of the PCD-CT system. More recent studies performed on the first clinical PCD-CT system support these previous results, with two experimental studies showing that the clinical PCD-CT outperforms third generation (i.e. latest generation) DS EID-CT systems across a variety of widely matching protocol parameters, reconstruction settings and radiation dose levels in terms of objective and subjective image quality. ^59,60

The improved performance of PCD-CT was also confirmed in-vivo in a variety of clinical studies. Exemplarily, Symons et al showed that a prototype PCD-CT exhibited between 15 and 17% lower noise than EID-CT for 120 kV and 100 kV dose-matched chest CT scans. ⁶¹ Importantly, low dose PCD-CT imaging of the lungs was also associated with improved HU stability relative to EID-CT. ⁶² Specifically, across tube voltage settings of 80, 100 and 120 kV, attenuation values of lung equivalent foams of a dedicated phantom as measured on PCD-CT remained stable while the attenuation values of EID-CT decreased by up to approximately 5 HU when decreasing the dose level from 3 to 0.75 mGy CTDI_vol.

Pourmorteza et al demonstrated improved image quality of a prototype PCD-CT for brain imaging relative to EID-CT exhibiting 12.8–20.6% less image noise and 15.7–33.3% improved soft-tissue CNR. ⁶³ Lastly, in a combined phantom and in-vivo study Symons et al demonstrated improved coronary artery calcium scoring (CACS) accuracy at low radiation doses for PCD-CT relative to EID-CT owing to the improved noise and CNR properties of PCD-CT. ⁶⁴ The improved detection and quantification accuracy of CACS at low radiation doses was later confirmed by van der Werf et al using a different PCD-CT concept. ⁶⁵

Recently, two patient studies have been published on the performance of clinical PCD-CT operated at 120 kV compared to EID-CT scans performed on a third generation (i.e. latest generation) DS EID-CT scanner with automated tube voltage selection. These were intraindividual comparison studies in whom the same patients underwent scans both with PCD-CT and EID-CT within a relatively short time period. For abdominal CT, 50 keV VMI reconstructions of PCD-CT exhibited higher vascular and parenchymal CNR than polychromatic EID-CT reconstructions at similar subjective image quality. ⁶⁶ For high-pitch CT angiography of the aorta, 45 keV VMI reconstructions of PCD-CT showed improved vascular CNR and similar subjective image quality as compared to polychromatic EID-CT reconstructions. ⁶⁷

In a recent study, Higashigaito et al could show that CT angiography of the aorta using PCD-CT and administering 20% reduced contrast media volume provides non-inferior image quality compared to CT angiography of the aorta using EID-CT in the same patients and at matched radiation dose. ⁵³

Representative images of patient studies illustrating the improved noise and CNR characteristics of PCD-CT are shown in Figures 4–6.

Figure 4.

47-year-old male patient (body weight 85 kg) with COVID-19-associated pneumonia presenting with ground-glass opacities and mild reticular abnormalities. Non-enhanced chest CT was acquired on a clinical PCD-CT system (NAEOTOM Alpha, Siemens Healthineers) in the UHR mode at 120 kV; the CTDI_vol was 0.55 mGy. CTDI_vol, volume of CT dose index; PCD, Photon-counting detector; UHR, ultra-high resolution.

Figure 5.

82-year-old male patient (body weight 81 kg) undergoing CTA for follow-up after endovascular treatment of an abdominal aortic aneurysm. Images were acquired on a third generation dual-source EID-CT system (SOMATOM Force, Siemens Healthineers) with automated tube voltage selection (80 kV) and CTA with a clinical dual-source PCD-CT system (NAEOTOM Alpha, Siemens Healthineers) (120 kV) at matched radiation dose (CTDI_vol 6.1 mGy) and using the same contrast media protocol. Note the reduced noise and improved contrast on PCD-CT images. CTA, CT angiography; CTDI_vol, volume of CT dose index; PCD, Photon-counting detector; EID, energy-integrating detector; PCD, Photon-counting detector.

Figure 6.

71-year-old female patient (body weight 75 kg) with a cyst in liver segment VII. Contrast-enhanced abdominal portal venous phase images were acquired on a third generation dual-source EID-CT system (SOMATOM Force, Siemens Healthineers) with automatic tube voltage selection (120 kV) and with a clinical PCD-CT system (NAEOTOM Alpha, Siemens Healthineers) in the spectral imaging mode at 120 kV at matched radiation dose (CTDI_vol 7.24 mGy) and using the same contrast media protocol. Note the improved iodine contrast and lesion conspicuity on PCD-CT images. CTDI_vol, volume of CT dose index; EID, energy-integrating detector; PCD, Photon-counting detector.

Intrinsic spectral capabilities

A key feature of PCD-CT is its ability to provide spectral information from every scan due to the detector being able to count and characterize individual photons according to their energy. In contrast, most EID-CT systems require the user to choose between single-energy or DE scan modes prior to image acquisition. The intrinsic spectral capabilities of the first clinical DS PCD-CT system with 0.25 s rotation time allows for multienergy imaging at 66 ms temporal resolution and high pitch multienergy imaging with helical pitch values up to 3.2. When implemented as an ECG-triggered high-pitch scan mode for cardiac imaging, the latter may enable radiation dose reductions of up to a factor of 2 as compared to other DSCT scan techniques such as ECG-triggered sequential step-and-shoot and ECG-gated spiral with X-ray pulsing. ^8,67,68

Currently, ECG-gated UHR imaging represents an exception to this rule: due to the vast amount of image information that has to be processed, spectral image information is not yet available for this imaging mode on the first clinical DS PCD-CT system. However, additional updates to the scanner’s soft- and hardware should be able to fix the problem in the near future.

The spectral capabilities of PCD-CT allow the user to reconstruct iodine maps, virtual non-contrast (VNC) and virtual monoenergetic images (VMIs) from every acquisition. The use of VMI for routine clinical PCD-CT imaging has exemplarily been shown for abdominal imaging, ^28,66 cardiac CT ^69–71 and coronary calcium quantification, ⁷² high-pitch CT angiography of the aorta, ⁶⁷ lung ^35,73 and brain imaging. ⁷⁴ The benefits of low keV VMI include better visualization of small low-contrast structures and increased iodine signal. For CT angiography examinations, the latter may be particularly promising as high pitch CTA together with low keV VMI may set new standards in terms of radiation and contrast media dose reductions. ^8,53

Concerning VNC images and iodine maps, Rajendran et al ⁸ and Sartoretti et al ⁷⁵ have demonstrated the feasibility of reconstructing these images from routine clinical PCD-CT scans. Quantitative accuracy of reconstructions was high with VNC showing mean absolute errors of 4 HU ⁷⁵ and iodine maps exhibiting a root mean squared error of 0.5 mg/cm³ for iodine concentration. ⁸ For VNC imaging, the high quantitative accuracy was later further confirmed in a larger clinical study encompassing 100 patients who underwent a triphasic examination on the first clinical PCD-CT system. Attenuation errors of VNC images were less than 5 HU in 76% and less than 10 HU in 95% of measurements compared with true non-contrast images across a variety of abdominal organs and regions. Furthermore, diagnostic image quality of VNC images as determined by two independent readers was achieved in 99 and 100% of cases, respectively. ⁷⁶

The clinical benefits of the routine availability of VNC images and iodine maps from PCD-CT include, among others, emphysema quantification from VNC images, ⁷⁷ assessment of adrenal adenomas from VNC images, ⁷⁸ anemia detection and quantification from VNC images ⁷⁹ and myocardial extracellular volume quantification based on iodine maps from a single cardiac late enhancement scan. ⁷⁰

Beyond the opportunities discussed above, PCD-CT harbors further potential in terms of improved and novel spectral imaging applications. The spectral data allow for the reconstruction of further spectral images such as calcium-only or virtual non-calcium (VNCa) images. In this regard, a novel vascular calcium removal algorithm has been recently introduced that aims to surpass the performance of previous similar algorithms designed for DECT capable EID-CT systems. ⁸⁰ With this novel algorithm, high quality VNCa images can be reconstructed thus counteracting the problem of blooming artifacts from heavily calcified plaques on standard monoenergetic or polychromatic images. ⁸¹ A representative image example illustrating the performance of the novel algorithm for high quality VNCa imaging is provided in Figure 7.

Figure 7.

67-year-old male patient (body weight 71 kg) with atypical chest pain. Coronary CTA was performed on a clinical dual source PCD-CT system (NAEOTOM Alpha, Siemens Healthineers) at 120 kV (CTDI_vol 10.1 mGy). Virtual monoenergetic images at 55 keV and VNCa images using a novel vascular calcium removal algorithm (PureLumen) were generated. A calcified plaque can be seen in the distal right coronary artery, which is subtracted on a dual-energy basis in the VNCa (PureLumen) images. CTA, CT angiography; CTDI_vol, volume of CT dose index; PCD, Photon-counting detector; VNCa, virtual non-calcium.

With PCD-CT, improved material decomposition can potentially be achieved, as the selection of energy thresholds can be tailored towards the spectral behavior of the materials that are to be separated. For example, by selecting optimized energy thresholds high quality VNCa and contrast media maps could be computed from coronary and carotid CTA images acquired with a range of contrast media including iodine and experimental contrast media such as bismuth, tungsten, holmium or hafnium. ^82,83 Furthermore, the energy threshold capabilities of PCDs can theoretically also be leveraged for improved simultaneous dual-contrast agent imaging as shown in experimental studies ^84–87 or to further reduce metal artifact burden. ^88,89

Although the latter applications are still in the preclinical phase of testing, PCD-CT opens up a range of new opportunities that may inspire a new momentum for clinical CT imaging.

Representative image examples highlighting the applications discussed above are provided in Figures 8–9.

Figure 8.

54-year-old male patient (body weight 76 kg) with hepatocellular carcinoma and vascular invasion. Late arterial scans were performed with a clinical PCD-CT (NAEOTOM Alpha, Siemens Healthineers) in the spectral mode with reconstruction of virtual monoenergetic images at 70 keV, virtual non-contrast images, and iodine maps from a single acquisition (CTDI_vol3.6 mGy). Although tumor enhancement is seen also on the monoenergetic images, the iodine maps allow for a better appreciation of the carcinoma along with the possibility of quantification of iodine uptake. CTDI_vol, volume of CT dose index; PCD, Photon-counting detector.

Figure 9.

59-year-old male patient (body weight 81 kg) with chronic occlusion of the right pulmonary artery and partial occlusion of left pulmonary artery branches resulting in severely reduced perfusion of the right and, to a lesser extent, of the left lung. Axial and coronal thick maximum intensity projection images and coronal PBV map computed from a routine contrast-enhanced chest CT scan acquired on a clinical PCD-CT (NAEOTOM Alpha, Siemens Healthineers) at 120 kV (CTDI_vol 2.37 mGy) illustrate both the anatomical and functional situation in the lungs. CTDI_vol, volume of CT dose index; PBV, perfused blood volume; PCD, Photon-counting detector.

Conclusion

In this review, we aimed to summarize basic technical principles and potential advantages of PCD-CT. Potential clinical benefits as evidenced by recent publications with this technology are outlined and representative clinical cases from our experience were added to illustrate the added value of the technique.

While the current possibilities of PCD-CT already considerably enhance our diagnostic capabilities, we foresee developments with spectrally optimized contrast media in combination with adaptable energy threshold imaging to enter clinical CT imaging in the near future.