Photon-counting CT: Review of initial clinical results

.


Introduction
After its clinical introduction, computed tomography (CT) has become an integral part of contemporary medicine [1,2].Over the years, the technique gradually improved in various aspects including scan speed, dose efficiency, dual-energy acquisition methods and temporal and spatial resolution.Today, conventional CT still has limitations regarding tissue contrast, image quality and spatial resolution for specific applications, which are mostly related to the design of current energy-integrating detectors (EIDs) [3,4].These detectors consist of scintillator material, such as cesium iodide or gadolinium oxysulfide, which absorbs X-ray photons and emits visible light in response (Fig. 1).The emitted light is then detected by a photodiode and converted into an electrical signal.Septa are integrated in the detector to prevent crosstalk between detector elements reducing the effective detector area.In addition, all incoming photons are integrated to generate the signal.
Low-energy photons are responsible for generating contrast, however, due to integration, they contribute relatively less to the final image, resulting in lower contrast-to-noise images with conventional CT (Fig. 1A).
Photon-counting CT (PCCT) is a new technique that has the potential to overcome the limitations of EIDs by using photon-counting detectors (PCDs) (Fig. 1B).PCDs consist of a semiconductor in which no septa are needed to separate individual detector elements [5].The detector elements of PCDs are smaller, whilst remaining geometrically efficient.This leads to an improved spatial resolution, creating new opportunities for clinical applications.PCDs convert single incoming photons directly into an electrical pulse proportional to the photon's energy.This enables the detector to categorize photons into energy bins based on the heights of the incoming electric pulses, allowing for spectral imaging, filtering out the electronic noise of the final image, and improved material decomposition [5].Several of these advanced image reconstruction Abbreviations: CACS, Coronary artery calcium score; (C)CTA, (Coronary) computed tomography angiography; CNR, Contrast-to-noise ratio; CT, Computed tomography; CTDI vol , Computed tomography dose index (volume); ECV, Extracellular volume; EIDs, Energy-integrating detectors; EVAR, Endovascular aneurysm repair; HU, Hounsfield unit; IR, Iterative reconstruction; IQR, Interquartile range; keV, Kiloelectronvolt; PC, Pure calcium; PCCT, Photon-counting computed tomography; PCDs, Photon-counting detectors; SNR, Signal-to-noise ratio; SPCCT, Spectral photon-counting computed tomography; TNC, True-non-contrast; UHR, Ultra-high resolution; VNC, Virtual-non-contrast; VMI, Virtual-mono-energetic image.
techniques are illustrated in Fig. 2.
PCCT is starting to be gradually employed in clinical practice and more research giving insight into this matter has become available.Therefore, this article reviews the literature on the current PCCT systems and is followed by an overview of the potential clinical applications and current challenges.

Data collection
A literature search was performed in the following databases: Embase, Medline, Web of Science, Cochrane and Google Scholar for articles written in English published until September 2022 (appendix A).One author (JvdB) assessed all retrieved articles based on title and abstract.References were scanned to ensure no relevant articles were missed.Articles were included if they described clinical application of photon-counting CT in patients.Overall, 51 studies were included in this review.From all studies the following characteristics were extracted: first author, publication year, journal, clinical application, sample size, PCCT system, EID system, acquisition mode and conclusions.All characteristics are summarized in supplementary table 1.A summary of this table is provided in Table 1.

Available PCCT systems
Until now, two PCCT systems by different vendors are described for patient use.Other manufacturers are currently developing their system but there is only scant information available on system design, capabilities and, to our best knowledge, no use in humans.The different PCCT systems and properties are briefly discussed below.

Siemens
Most studies are conducted with the NAEOTOM Alpha (n = 29) which is currently commercially available or a prototype version (n = 17) of this system (Siemens Healthineers).The clinical system contains a dual-source PCD setup with a maximal temporal resolution of 66 ms and a cadmium-telluride detector.In high-resolution mode, the PCCT system can acquire images with a z-coverage of 144x0.4mm.In ultra-highresolution (UHR) mode a maximal coverage of 120x0.2mm is achieved albeit for dual-source applications without spectral information as of yet due to the amount of data to be registered for these acquisitions [6].

Philips
Three studies from the same center use the spectral photon-counting CT (SPCCT) system from Philips.The system is equipped with a detector yielding a pixel pitch of 0.3 × 0.3 mm (UHR) at the isocenter and a zcoverage 64 x 0.3 mm [7][8][9].

Other vendors
Two other vendors (GE Healthcare, and Canon) are developing a PCCT system, to our best knowledge, Canon has one prototype installed in Japan and GE has multiple prototypes installed but none have published results of its use in patients [10,11].GE Healthcare uses a different type of semiconductor material (silicon) whereas Canon uses cadmium-zinc-telluride.

Cardiovascular imaging
In cardiovascular imaging, thirteen studies were found focusing on enhancement of contrast agent signal, depiction of coronary arteries, calculation of coronary artery calcium scores (CACS) on virtual-noncontrast reconstructions (VNC), stent imaging and myocardial assessment.

Coronary computed tomography angiography (CCTA)
Si-Mohamed et al. showed improved contrast-dependent spatial Fig. 1.Schematic overview of CT detectors.A. Energy-integrating detectors convert incoming photons to visible light, which is directly followed by a conversion to an electrical signal by photo-diodes.All measured photons in one projection are integrated to create the signal of that projection.The detector's spatial resolution depends on the use of reflecting septa to prevent cross-talk.B. Photon-counting detectors create an electron-hole pair when a photon strikes the semiconductor material of the detector.The anode attracts the electrons from the electron-hole-pair resulting in an electric pulse proportional to the photon's energy.resolution and noise reduction in UHR-CCTA which allowed for improved subjective visualization of small arteries, stents, calcifications, and non-calcified plaques compared to EID-CT (n = 14) using the Philips system [12].An additional study using the Siemens system demonstrated UHR-CCTA (n = 20) with a sharp kernel (Bv89) depicted coronary arteries with less blooming artifacts due to calcifications compared to a less sharp kernel Bv40 (40%vs53%, respectively) despite lower SNR (2.4 ± 0.4) and CNR (3.0 ± 0.4) for Bv89 compared to Bv40 (SNR:18.7 ± 3.8, CNR:22.5 ± 3.3) [13].UHR acquisition also allowed for sharp depiction of partially calcified plaques, discriminating between calcium and non-calcified components, which might improve the identification of vulnerable plaques [13].Considering the subjective image analysis, Bv64/Bv72 kernels were recommended for lumen and adjacent plaques assessment by the authors for this specific vendor.
Virtual-mono-energetic images (VMI's) can increase luminal iodine contrast at lower keV levels (40-45 keV) [14,15].In aortic imaging, lower keV's also showed a significant increase in CNR (p < 0.025) and similar subjective overall image quality but increased image noise (p < 0.001) in 40 patients compared with EID-CT [15].The increase of iodine contrast allows for reducing the administered iodine dose while preserving visualization of small low-contrast structures [14].

Coronary artery calcium score
CACS is an important predictive marker for future cardiovascular events.[16].True-non-contrast (TNC) scans are made to determine CACS and represent a considerable part of the total radiation dose of a clinical CCTA [17].In addition, blooming artifacts and the presence of small calcifications may lead to CACS over-or underestimation.Symons et al. investigated low-dose PCCT (25% of the standard dose) to compute CACS in 10 patients [17].Low-dose PCCT had a mean difference of − 10.8 in Agatston CACS compared to EID-CT.VMI's at higher keV levels and stronger iterative reconstruction (IR) strengths also resulted in lower mean Agatston scores [18].The lower Agatston scores could lead to underdiagnosis of patients with coronary calcifications, incorrectly classifying them with no calcifications or in a lower risk group.
By using the spectral information of PCCT, images can be reconstructed without the signal of iodinated contrast, obviating the need for a non-enhanced scan.In 67 patients, CACS derived from two reconstructions algorithms (VNC and Pure calcium (PC)) showed underestimation of Agatston scores compared to standard non-enhanced scans [16].These algorithms aim to replace TNC scans in the future, however further optimization of these algorithms seems needed to derive more accurate calcium scores of CCTA scans.

Stent imaging
On CT, metal stents appear thicker due to blooming artifacts Fig. 2. PCCT-scan of a Patient with endovascular aortic repair.A/B.Virtual-mono-energetic images (VMIs) can be created without additional scanning from the spectral dataset to generate images for virtual photons at a single kiloelectronvolt (keV) energy level.This can be helpful to improve differentiation between tissues, visualization of contrast-enhanced tissues, or reducing artifacts.C. Virtual-non-contrast (VNC) reconstructions remove the signal of iodinated contrast agents from contrast-enhanced scans, with the idea to potentially make non-enhanced scans obsolete.D. Iodine maps are useful to measure perfusion but hinge on their quantification accuracy [2].E. It is possible to fuse VNC and iodine maps for additional information.hampering in-stent lumen assessment.Additionally, stents have struts that are spaced close to each other which complicates visualization of stents.UHR-PCCT of eleven coronary stents showed smaller external and larger internal stent diameters for PCCT than EID-CT (p < 0.05).Less blooming artifacts were present in PCCT images, resulting in more accurate stent diameter measurements [7] (Fig. 3).
In patients after endovascular aneurysm repair (EVAR), routine follow-up is often done by CT using TNC and CTA scans for diagnosing endoleaks and accurate discrimination from calcifications.In 20 EVAR patients, two contrast removing algorithms (VNC and PC) were reconstructed from CTA images to investigate if they can replace TNC images [19].Complete iodine removal was obtained by both algorithms, but only the PC algorithm obtained minimal erroneous elimination of calcification and stent struts [19].Five-point Likert scales showed more accurate image quality for PC (4.2 ± 0.9) compared to VNC (2.5 ± 0.6).Lower SNR (PC = 2.5 ± 1.3, VNC = 1.9 ± 0.9) compared to enhanced scans (3.3 ± 1.6) was found for both algorithms.

Other cardiac imaging
Four studies investigated the advantages of PCCT to evaluate the myocardium.First, Ayx et al. compared myocardial radiomics features between PCCT and EID-CT [20,21].Radiomic features incorporate pixelbased quantitative data to reveal characteristics that are not observed with the human eye.Both scanners showed comparable first-order features but second-order features (textures) were more heterogeneous [21].This heterogeneity might indicate improved spatial resolution, detection of lower-energy photons, and signal-to-noise ratio.The differences between EID-CT and PCCT image textures indicate the images cannot always be used interchangeably for radiomics assessment.Texture-features analysis in the assessment of myocardium is still experimental and has yet to prove its clinical value.
Extracellular volume (ECV) quantification is a technique to measure the expansion of myocardial tissue and usually measured on magnetic resonance imaging [22].In most pathologies involving the myocardium, ECV enlarges.For ECV quantification, a non-contrast and contrastenhanced scan are needed.Mergen et al. studied the feasibility of iodine maps from late-enhancement PCCT scans to quantify ECV.The use of iodine maps is not subjected to misregistration and might obviate the need for a TNC image like in CAC scoring.Further evidence is required to establish the clinical efficacy of quantifying ECV utilizing PCCT.
VMIs were investigated for the quantification of epicardial adipose tissue attenuation [23].Images at 70 keV were suggested for the optimal epicardial adipose tissue attenuation assessment due to corresponding CT numbers compared to conventional CT.Lastly, one case report showed the feasibility of myocardial perfusion assessment through iodine mapping for myocardial infarction with obstructive coronary artery disease in a single patient [24].

Thoracic imaging
In thoracic imaging, high spatial resolution is of vital importance to assess lung parenchyma and surrounding low-contrast structures especially in patients with interstitial or alveolar lung diseases [25,26].Spectral properties may provide information on vessel occlusions and lung perfusion differences with the aid of blood volume maps (Fig. 4).Nine articles investigated the possibilities of PCCT in thoracic imaging.valuable for patients with interstitial disease or lung cancer (screening/ treatment) as their radiation exposure can accumulate rapidly when undergoing regular CT scans for treatment monitoring and follow-up [25,26].Four studies compared image quality of low-dose high-resolution PCCT to EID-CT [25][26][27][28][29]. PCCT allowed for scanning with lower radiation dose (43%-50% reduction) while retaining equal image quality.One study found lower SNR in PCCT (CTDI vol = 2.5 mGy) compared to EID-CT (CTDI vol = 3.0 mGy) in the evaluation of lung pathologies [28].The higher noise levels could be explained by increased spatial resolution and the use of sharper kernels in PCCT [28,30].The lower SNR in PCCT did not affect subjective visual assessment.

Lung emphysema
For the quantification of lung emphysema, a non-contrast-enhanced scan is needed.Jungblut et al. investigated the impact of different VMIs (n = 60) and VNC (n = 65) reconstructions to quantify lung emphysema [31,32].For VMI (40-80 keV), the smallest difference in the amount of emphysema (-16 %) was found at 80 keV compared to VNC.Monoenergetic levels at 40 keV showed improved iodine CNR (46.5 ± 10.7 vs 29.1 ± 7.5) and vessel evaluation.Considering the assessment of pulmonary vessels and emphysema simultaneously, objective and subjective analysis showed the best results for images reconstructed between 60 and 70 keV.For VNC reconstructions, emphysema quantification was comparable for both VNC reconstructions derived from arterial (p = 0.41), and venous (p = 0.09) scan phases compared to TNC, demonstrating the feasibility of replacing TNC scans with VNC reconstructions for lung emphysema quantification.

Ultra-high resolution thoracic imaging
Six feasibility studies with ultra-high-resolution acquisition (0.2 mm) in combination with a sharp kernel, showed excellent visibility of 4th and 5th-order bronchi, bronchial walls, vessels and lung parenchyma as well as retaining visibility of lung nodules [27,28,30,[33][34][35].Proper kernel selection is important for optimal usage of high spatial resolution PCCT.Two studies investigated UHR-PCCT imaging in interstitial lung disease [36,37].In 30 patients with usual interstitial pneumonia, low-contrast tasks like assessing absence or presence of ground glass opacities (p = 0.019) and mosaic pattern (p = 0.013) demonstrated higher readers confidence compared to EID-CT [37].For systemic sclerosis, a dose reduction of 66% was established while maintaining equal diagnostic accuracy compared to EID-CT [36].

Abdominal imaging
Clinical studies in abdominal PCCT focused on three main topics: the need for anatomical detail in abdominal contrast-enhanced scans, improvement of image quality in obese patients, and spectral postprocessing in abdominal malignancies, liver, spleen and pancreas imaging.A/B.Patient with bilateral central pulmonary embolism as seen on the axial CT images.C. Spectral properties of PCCT allow to reconstruct perfusion blood volume maps to determine perfusion loss of the lung.The darker area in the left lower lung indicates less perfusion.The Red/blue overlay enhanced the differences between anatomical structures filled with contrast (blue) and thrombi and less perfused tissues (red).(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Image detail advancement
In contrast-enhanced abdominal imaging, three studies observed equal objective and subjective image quality compared to EID-CT with a low or reduced dose up to 57% for PCCT [27,[38][39][40].Sartoretti et al. examined the effect of increased IR strength on noise magnitude and texture in patients with a contrast-enhanced abdominal scan [41].Increasing IR strength did not affect the noise texture.However, the area of the noise power spectrum decreased with an increase in IR strength corresponding with less noise (45%) and a better subjective/objective (p < 0.001) image quality.Higher diagnostic accuracy for abdominal pathologies could be achieved due to this image quality improvement.For example, Marcus et al. found the number of detected uric acid and non-uric acid containing normal and small size renal stones (<3mm) was higher for PCCT compared to DECT (70% vs 54.4%) [42].

Obese patient population
To achieve high image quality in obese patients, an increase in kV is required leading to a more substantial reduction of iodine image contrast compared to other tissues.Using spectral information, PCCT offers the potential to compute low-keV VMI.Decker et al. investigated the feasibility of low-dose non-contrast abdominal PCCT compared to EID-CT in 20 dose-matched patients [43].They concluded that an increase in BMI had a greater impact on noise in EID-CT (+39 %) than PCCT (+2%).Abdominal structures such as mesenteric vessels, ureters and the renal pelvis were better delineated by PCCT compared to EID-CT (overall image quality: p < 0.001).Low-dose non-contrast abdominal PCCT enables a more reliable diagnosis of abdominal pathologies with higher SNR at equal low-dose EID-CT.Another strategy is to keep image quality constant to conventional EID-CT but decrease the radiation dose for the patient [38].Hagen et al. confirmed this strategy with a cohort of obese patients scanned for oncological reasons (n = 51) with a dose reduction of 25% [44].
Niehoff et al. investigated diagnostic accuracy of VNC reconstructions compared to TNC images for hepatic steatosis.VNC images showed lower HU values for both the liver (mean difference; 8.7 ± 4.7HU) and the spleen (mean difference; 5.5 ± 4.7HU) resulting in more false positives when only liver HU values were included for diagnosis [46].Considering the mean CT numbers for both liver and spleen were lower on VNC compared to TNC, the ratio between the two with an adjusted cut-off value was considered useful for diagnosis.Sartoretti et al. found stable HU values for hypodense liver lesions in VNC and TNC images with a mean error of 3.7 ± 2.2HU [47].Mergen et al. found an absolute error (VNC/TNC) < 10 HU in 95 % of 100 patients [48].Furthermore, attenuation values remained stable when IR strength increased for both conventional and 60 keV images [41].In another study in 29 patients with adrenal adenomas, VNC reconstructions showed over-and underestimation of the HU values compared to TNC resulting in a non-significant mean difference of 9.2 ± 8.6HU vs 7.3 ± 8.4HU respectively [49].But, despite the small absolute difference in HU values this lead to seven misclassified adrenal lesions.
In 30 patients with liver lesions, iodine quantification of the liver parenchyma and lesions was investigated.Iodine concentration in PCCT was measured and compared to values of EID-CT found in literature.For the liver parenchyma (1.7 ± 0.4 vs 2.1 ± 0.8 mgIodine/mL) and cysts (0.2 ± 0.1 vs 0.2 ± 0.3 mgIodine/mL) concentrations were comparable, opening opportunities for iodine mapping with PCCT [50].The use of VNC/iodine mapping can reveal the perfusion status of tissue and might play a role in detection of bowel infarction.Improvement in CNR can detect subtle density differences in bowel wall enhancement [51].

Neuro-imaging
For the initial assessment of the brain with conventional CT, softtissue contrast remains low, beam-hardening arises from the skull and small anatomic structures are difficult to visualize.Reconstructions with high mono-energetic levels can reduce beam hardening artefacts without additional scanning.A total of nine studies assessed opportunities of PCCT head imaging.

Gray matter -White matter differentiation
Differentiation between white and gray matter is troublesome with conventional CT considering tissue attenuations only differ by approximately 10HU [52].Subjectively, neuroradiologists could make more accurate distinction between gray matter and white matter with PCCT compared to EID-CT in 21 patients [53].Objectively, in 49 patients, conventional PCCT images had higher white-gray matter CNR (2.90 ± 0.92, p < 0.001) compared to mono-energetic reconstructions (40 keV [2.26 ± 0.96] − 120 keV [0.71 ± 0.54]) [52].The CT numbers of white and gray matter still only differed by ≈10HU (WM = 33HU, GM = 42HU) for conventional images.Images at 40 keV depicted a more prominent differentiation with a mean signal difference of ≈20HU but with larger image noise.Low keV images are also prone to more noise Fig. 5. Spectral advantages in abdominal imaging.A/B by lowering the virtual-mono-energetic level, hyperdense liver lesions with iodine uptake stand out more clearly compared to 70 keV images.and beam-hardening artifacts caused by the skull leading to a decrease in SNR and CNR.PCCT showed less beam-hardening artifacts compared to EID-CT even at low keV's, but optimization is needed.Currently, images at high resolution (0.4/0.2 mm) are more favorable to assess brain matter but not optimal [53,54].

Neurovascular imaging
Beam-hardening artifacts attributable to the skull base and cervical vertebrae greatly impact the assessment of vascular disease (i.e., atherosclerotic lesions) in the carotid and intracranial vessels.Three studies investigated CTA of the brain with PCCT.Conventional images had comparable results in terms of SNR/CNR compared to VMI's but subjectively, were more favorable for the assessment of brain arteries with CTA (p < 0.001) [54].In two studies describing a total of seventeen patients, higher iodine contrast attenuation (20.8%) with VMI reconstructions and less beam-hardening in regions with extensive bone involvement (C2 segment of the carotid artery) compared to EID-CT were found [55,56] (Fig. 6A/B/C).

Temporal bone
Temporal bone imaging seems to benefit from UHR-PCCT because of the submillimeter size of the anatomical structures [33][34][35]57].UHR-PCCT in thirteen patients showed subjectively sharp anatomic details because of the reduced partial volume effect.This resulted in improved visualization of abnormalities of the ossicular chain, postoperative changes related to ossicular prostheses, and the integrity of the incudostapedial articulation at a reduced radiation dose (-31%) compared to EID-CT [57] (Fig. 6D/E).

Musculoskeletal imaging
The ability to visualize trabecular bone and the reduction of metal artifacts near orthopedic implants has created interest for musculoskeletal imaging with PCCT.Until now, four studies with a small study population described a clinical evaluation in this field.

Bone structure
Four patients with bone lesions of metastasized breast cancer were scanned with UHR-PCCT [58].Fine structures such as single trabeculae could be visualized.In conventional CT, the partial volume effect affects the differentiation between small metastases and benign bone focal changes.PCCT allows for an accurate depiction of tumor margins and spiculae resulting in a more definite evaluation of bone metastasis [58].Three studies investigated the image quality of bone in patients undergoing UHR-PCCT for shoulder/pelvic/wrist assessment [33,59,60].Fine cortical and trabecular bone together with bridging callus and sequalae of hypertrophic degenerative arthritis had greater subjective reader confidence at 31%-49% lower radiation dose compared to EID-CT.confidence (65%) of periprothestic fractures, implant loosening, and enhanced delineation of cortical bone, bone trabeculae, and implantbone interface compared to EID-CT.UHR-PCCT also demonstrated less metal artifacts (p < 0.001) compared to EID-CT.This effect was most evident in patients with spinal implants.The noise level was ≈12% lower in PCCT, yielding a 23% improvement in dose efficiency [61].Fig. 7 demonstrates a clinical case example of metal artifact reduction.

Current challenges of photon-counting CT
Besides the benefits of PCCT, there are also some limitations.In PCCT, achievable image quality is now limited by the X-ray tube that has a maximum filament current, maximum power and an inadequate focal spot size.Second, many studies mentioned potential dose reduction of PCCT.The aims of these studies are often to demonstrate non-inferior image quality compared to conventional CT.To profit from the spectral properties of PCCT, a relatively wide energy spectrum is needed.To create such a wide spectrum, a relatively high tube voltage is needed which is often correlated with a higher dose and lower contrast, depending on the clinical task.The question remains whether this leads to an increased dose and if this is justified by the advantages of spectral image analysis [62].In addition, the large amount of data generated by PCCT requires significant computational resources or prior to scanning accurate selection of the acquisition protocol and reconstructions to limit the data, especially in UHR mode.Lastly, and not insignificantly, PCCT undoubtedly enables high-quality images and allows physicians to post-process images to their preferences to increase diagnostics confidence.However. the impact of PCCT on clinical decision-making is still unanswered and the question remains if PCCT will lead to for example earlier diagnosis or make invasive interventions obsolete.

Conclusion
PCCT is a promising technique to overcome the shortcomings of energy-integrating CT detectors currently used in clinical practice.In cardiovascular imaging, PCCT improves image quality of small arteries, stent visualization and allows for deriving CAC scores from the contrastenhanced CCTA's.In thoracic imaging, PCCT can play a role in radiation dose reduction for patients who frequently undergo CT scans.
Challenges of abdominal imaging, like imaging obese patients, can be mitigated by PCCT.Additional spectral post-processing has the potential to improve visualization of liver and renal pathologies but further research is needed to identify more advantages of abdominal PCCT.For neuro-imaging, the sharp depiction of temporal bone can improve diagnostics in this small and complex anatomic region, differentiation of white and gray matter, however, remains challenging.Lastly, a detailed assessment of bone structure and osseointegration of orthopedic implants in musculoskeletal imaging can be achieved with PCCT.The number of studies describing the clinical applications of PCCT is still limited and most studies include small sample size and scan acquisition protocols vary widely.A vast increase in the number of publications on PCCT is expected in the (near future).There is a need for studies assessing the capabilities of PCCT in larger patient cohorts, using standardized acquisition settings to investigate the influence of photoncounting CT on the sensitivity and classification of various diseases.

Declaration of Competing Interest
The authors declare institutional support by Siemens Healthineers to Erasmus MC.  7. Spectral shaping in combination with high energy (keV) provides artifact-free imaging of bone with a metal elbow implant.High X-ray attenuation of orthopedic implants causes metal artifacts expressed by streaks and dark/bright regions in the image.Elimination of these artifacts is crucial to assess the position, osteointegration and infection of orthopedic implants.A. Using the spectral dataset, low-energy photons can be excluded from the final image, which might improve the performance of metal artifact reduction thanks to less beam-hardening and photon starvation compared to conventional dual-energy CT.B cinematic rendering of CT scan.

Fig. 3 .
Fig. 3. Dual-source ultra-high resolution PCCT scans of a 73-year-old patient suspected of obstructive coronary artery disease A. Demonstrate calcifications in circumflex coronary artery with no significant stenosis as confirmed by angiography.B/C.Depiction of a stent in the left anterior descending with calcification outside of the stent is compressing the lumen.Mid-LAD a suspected stenosis was observed.

Fig. 4 .
Fig. 4. Imaging of pulmonary embolism by PCCT.A/B.Patient with bilateral central pulmonary embolism as seen on the axial CT images.C. Spectral properties of PCCT allow to reconstruct perfusion blood volume maps to determine perfusion loss of the lung.The darker area in the left lower lung indicates less perfusion.The Red/blue overlay enhanced the differences between anatomical structures filled with contrast (blue) and thrombi and less perfused tissues (red).(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

6. 6 .Fig. 6 .
Fig. 6.Applications of PCCT for neuro-imaging.A/B/C.A patient with an occluded flow diverter.Thanks to the increased spatial resolution of PCCT, the lumen of the flow diverter is clearly depicted, allowing for assessment of the lumen.D/E.Labyrinthitis ossificans in a 63-year-old patient.Again, the high spatial resolution of PCCT, enables differentiation between intracochlear calcification and partial volume effect of the surrounding otic capsular bone, which could be differentiated less reliably before with conventional CT.
Fig.7.Spectral shaping in combination with high energy (keV) provides artifact-free imaging of bone with a metal elbow implant.High X-ray attenuation of orthopedic implants causes metal artifacts expressed by streaks and dark/bright regions in the image.Elimination of these artifacts is crucial to assess the position, osteointegration and infection of orthopedic implants.A. Using the spectral dataset, low-energy photons can be excluded from the final image, which might improve the performance of metal artifact reduction thanks to less beam-hardening and photon starvation compared to conventional dual-energy CT.B cinematic rendering of CT scan.

Table 1
Summary of articles included.
* Mixed articles describe multiple clinical applications in various fields.J. van der Bie et al.