A scoping review of photon-counting detector computed tomography in cardiovascular imaging

Cardiovascular computed tomography (CT) is a critical tool for the non-invasive assessment of coronary artery disease, myocardial tissue pathology, structural/valvular disease, pulmonary vascular disorders, and aortic/peripheral arterial pathology. Photon-counting detector CT (PCD-CT) represents a shift in detector design. By registering the arrival and energy of individual X-ray photons, PCD-CT can reduce electronic noise and support high-resolution (HR), multi-energy imaging more effectively than energy-integrating detector CT (EID-CT).

PCD-CT can achieve submillimeter spatial resolution at routine radiation doses, enabling coronary CT angiography (CCTA) that depicts 1–2-mm distal vessels and fine morphologic features (e.g., non-calcified plaques, napkin-ring signs, ostial lesions). In parallel, energy-based photon weighting can mitigate calcium and stent blooming and improve in-stent lumen depiction and calcium quantification compared with EID-CT.1,2,3,4 Beyond morphology, the multi-energy output allows retrospective monoenergetic reconstructions, iodine maps, and (where available) K-edge material decomposition. These capabilities have supported single-acquisition protocols capturing coronary and myocardial phases, in vivo plaque imaging, and angiographic strategies that have reported a higher contrast-to-noise ratio (CNR) with iodine dose reductions of up to 40% in selected settings. Low-keV reconstructions have been reported to maintain diagnostic quality despite suboptimal bolus timing and to facilitate single-scan endoleak evaluation after aortic repair.5,6,7,8,9,10

These capabilities extend beyond coronary imaging: iodine-based maps and delayed-enhancement surrogates enable myocardial tissue characterization; HR sizing and peripheral access planning support structural/valvular workflows; iodine mapping supports pulmonary perfusion assessment alongside embolus detection; and low-keV reconstructions can improve vascular conspicuity in aortic and peripheral arteries, enabling contrast-saving protocols.5,6,7,8,9,10 Given heterogeneity in protocols, reconstructions, and endpoints across studies, a structured synthesis is warranted.

Accordingly, we present a state-of-the-art scoping review of peer-reviewed adult cardiovascular PCD-CT studies published through September 2025, synthesizing findings across five domains—coronary arteries, myocardial tissue, structural heart/valves, pulmonary–cardiopulmonary function, and aortic/visceral/peripheral arteries—and complementing these with clearly labeled, illustrative examples from our single-center experience [>1.000 PCD-CT examinations on a (Siemens, Forchheim, Germany), not included in the current synthesis].

A scoping literature review was conducted to map and summarize clinical studies of PCD-CT in cardiovascular imaging. The search covered January 2021 (onset of clinical availability) through September 1, 2025. Searches were performed in PubMed/MEDLINE, Embase, and Scopus using a broadened keyword set combining PCD-CT terms and cardiovascular terms. The core logic included PCD-CT synonyms (e.g., “photon-counting detector computed tomography,” “photon-counting CT,” “PCD-CT,” “SPCCT,” and device names where reported) AND cardiovascular concepts (cardiac, cardiovascular, coronary, vascular, myocardial, aorta, valvular/structural, TAVI/TAVR, pulmonary).

The initial search yielded 828 records. After deduplication and title/abstract screening, 467 records remained for full-text assessment. Original clinical studies in adults that used PCD-CT for any cardiovascular indication, were published in peer-reviewed journals, and reported at least one imaging, diagnostic, or workflow outcome were included. Phantom-, animal-, or simulation-only studies; pediatric or congenital cohorts; neurovascular studies; abstract-only publications; and purely technical reports without a clinical cohort were excluded. A total of 59 studies met the criteria and were included in the review. The evidence-identification flow is summarized in Figure 1.

Two reviewers independently screened titles/abstracts and full texts, resolving disagreements by consensus. For each included study, the following was charted: study design and setting; patient characteristics; scanner/vendor; acquisition and reconstruction parameters [e.g., kilovolt peak (kVp), pitch, ultra-HR (UHR) protocols, virtual monoenergetic image (VMI) keV levels, kernels/iterative settings]; dose metrics as reported [CT dose index volume (CTDIvol), dose-length product (DLP), size-specific dose estimate (SSDE), and/or effective dose]; contrast volume; image-quality measures [objective and subjective, including the signal-to-noise ratio (SNR)/CNR when provided]; and diagnostic performance vs. the stated reference standard [e.g., area under the curve (AUC), sensitivity, specificity].

Given heterogeneity in populations, protocols, endpoints, and reporting formats, no quantitative pooling or meta-analysis was performed. Findings are synthesized narratively, with numeric ranges reported by application domain (coronary, myocardium, structural/valvular, pulmonary, aortic/peripheral).

Coronary computed tomography angiography

A total of 31 investigations evaluating CCTA with PCD-CT were identified (Table 1). Study designs included retrospective and prospective diagnostic accuracy cohorts, protocol-optimization studies for radiation/contrast reduction, and quantitative assessments using virtual monoenergetic imaging and iterative reconstruction. Several studies used UHR (0.2 mm) acquisitions.11^, 12

In a large retrospective routine-practice cohort (n = 7.833), per-patient specificity with PCD-CT was 98% and 93% with EID-CT, positive predictive value was 83% and 63%, invasive coronary angiography (ICA) referral was 9.9% and 13.1%, and revascularization yield was 43.4% and 35.5%, respectively.13 In a high-risk cohort with heavy calcification or stents (n = 68), sensitivity and specificity were 96% and 84%, respectively.14 Two UHR studies reported a mean stenosis-measurement error of 6% and a vessel-level AUC of up to 0.99.11^, 12

Low-dose protocols achieved diagnostic image quality at a CTDIvol of 1.72 mGy (DLP:29.1 mGy·cm; effective dose:0.41 mSv), with >95% of segments rated diagnostic.15 A prospective comparison for calcium scoring showed a 56% dose reduction (1.0 vs. 2.3 mGy) with 1.5-mm slices and low-strength iterative reconstruction.16 Contrast-saving protocols delivered diagnostic CCTA with 30 mL of iodinated agent using 50-keV reconstructions,17 and preserved image quality after 40% contrast reduction using 45-keV reconstructions.18 Kernel and iterative-reconstruction optimization maintained a coronary CNR with finer matrices.19^, 20 An energy-sweep analysis (40–180 keV) documented systematic shifts in quantified plaque components across keV levels.21

Virtual non-contrast (VNC) approaches for calcium scoring showed high agreement with true non-contrast (TNC) scans [intraclass correlation coefficient (ICC): 0.98; kappa (κ): 0.88] and contributed to dose reduction.22 Additional evaluations reported risk-category misclassification for low-density calcium,23^, 24 with improved agreement using thin-slice reconstructions and higher tube potentials.25

Functional and data-driven analyses included on-site CT fractional flow reserve (FFR) with 100% negative-predictive value and a projected reduction of invasive angiography in 39% of cases,26 deep-learning stenosis detection with a vessel-level AUC of 0.92,27 and radiomics of pericoronary adipose tissue and plaque texture differentiating hypercholesterolemia and coronary disease status.28,29,30

A Monte Carlo economic model estimated reductions in downstream functional testing (18.9%), invasive angiography (6.0%), and major complications (9.4%), with an approximate cost saving of USD 800 per patient in stable chest pain pathways.31

Figure 2 presents the CCTA of a patient with a stent in the left anterior descending artery obtained with PCD-CT in our center.

Myocardial tissue characterization

Four clinical studies evaluated PCD-CT against cardiovascular magnetic resonance imaging (MRI) for late iodine enhancement (LIE) and/or extracellular volume (ECV) mapping (Table 2). In this review, LIE denotes CT-based delayed-phase iodine-related myocardial hyperenhancement acquired 5–10 minutes after iodinated contrast and quantified on iodine maps/VMIs; it is analogous to -but distinct from- cardiovascular magnetic resonance (CMR) late gadolinium enhancement (LGE).

In a diagnostic accuracy study of 27 patients (459 myocardial segments), Tremamunno et al.32 used dual-source PCD-CT with electrocardiogram (ECG)-triggered sequential acquisition 5 minutes post-contrast (120 kVp; 144 × 0.4 mm collimation; iodine maps reconstructed with Qr40 and iterative reconstruction). For two readers, per-patient sensitivity was 100% and 91.7%, specificity was 73.3% and 80.0%, and accuracy was 85.2%. Per-segment sensitivity was 74.7% and 66.7%, specificity was 94.9% and 96.4%, and accuracy was 91%. Inter-reader agreement was κ = 0.70 at the patient level and κ = 0.63 at the segment level.32

In a prospective series of 17 patients (24 CT/MRI pairs), Klambauer et al.33 performed spectral dual-source PCD-CT with a 5-minute delayed LIE and atlas-based ECV mapping. Agreement with LGE-MRI was κ = 0.832 in the acute setting; agreement with combined LGE + edema was κ = 0.944; and at 3-month follow-up, κ = 0.944.33

In 30 patients with systemic amyloidosis, Popp et al.34 used first-generation PCD-CT with CCTA and a delayed phase. Global ECV was 42.93% ± 10.14% (CMR), 42.51% ± 9.07% [single energy (SE)], and 40.69% ± 9.24% [dual energy (DE)]. Compared with CMR, SE showed a mean difference of 0.43% [95% confidence interval (CI): -1.83 to 2.68], whereas DE was -2.24% (95% CI: -4.42 to -0.06); DE vs. SE was -1.82% (95% CI: -2.70 to -0.94). Bland–Altman analysis: the mean bias for DE vs. CMR was -2.28% (limits of agreement: -11.16 to 6.59); for SE vs. CMR, it was -0.42% (-9.77 to 8.92); and for DE vs. SE, it was -1.82% (-5.46 to 1.83). Both CT approaches correlated strongly with CMR (r: 0.892 for DE; r: 0.882 for SE).34

In a retrospective CT–MRI comparison of 32 patients with acute myocarditis, Gkizas et al.35 reported a DLP of 96 ± 32 mGy·cm. The global ECV on PCD-CT was 29.4% ± 4.5% and 30.0% ± 4.1% on CMR (P = 0.69); correlation with LGE was r: 0.82, and the AUC for segment-level inflammation was 0.95 at a 26.9% threshold.35

Structural heart and valvular assessment

Six clinical studies (Table 3) assessed PCD-CT across components of transcatheter aortic valve replacement/implementation (TAVR/TAVI) evaluation, including access planning, annulus sizing, valve-calcium quantification, and concomitant coronary assessment.

In a retrospective comparative cohort of 300 patients (202 PCD-CT; 100 dual-source EID-CT), Dirrichs et al.36 performed aorto-ilio-femoral contrast CT for TAVI. The SNR was 33 ± 10.5 vs. 47.3 ± 16.4, and the CNR was 47.3 ± 14.8 vs. 59.3 ± 21.9 (both P < 0.001). The visual image-quality scores were 4.8 vs. 3.3 (P < 0.001), suitability for TAVI planning was 99% vs. 85%, and the effective radiation dose was 8.8 ± 4.5 vs. 15.3 ± 5.8 mSv.36

In a prospective study with 30 patients (plus 30 matched controls), Yang et al.37 implemented ECG-gated high-pitch PCD-CT at 30% R–R for annulus sizing. Correlation with spiral CT was strong (r ≥ 0.94). Mean paired differences (bias) with 95% CIs between high-pitch PCD-CT and spiral CT were 0.16 mm (-0.10, 0.42) for mean diameter, 0.22 mm (-0.70, 1.13) for the perimeter, and 5.35 mm² (-22.02, 32.72) for the annular area; Bland–Altman plots showed minimal bias. Additionally, CTDI was 4.52 vs. 24.10 mGy (P < 0.001); systolic capture occurred in 90% vs. 50%.37

In a paired-scan comparison of 64 pre-TAVR candidates, Hagar et al.38 compared UHR PCD-CT CTA (120 × 0.2 mm) with high-pitch spiral CTA (144 × 0.4 mm). The effective dose was 12.6 (UHR) vs. 4.1 mSv (high pitch); annulus image-quality scores were median 4 vs. 3 (P < 0.001). Area-derived annulus measurements were highly correlated (r²: 0.857), and prosthesis size selection was identical in 91% of patients; the distribution of ±1 size and ≥±2 sizes was not reported in the paper.38

In 123 patients (56 with aortic-valve calcification), Feldle et al.39 evaluated ECG-gated cardiac PCD-CT with 70 keV virtual non-iodine (VNI) images against 70 keV TNC. Sensitivity, specificity, and accuracy were 69%, 100%, and 85%, respectively, under prospective gating. Correlation between VNI-derived and TNC Agatston scores was r: 0.983–0.986 (P < 0.001).39

In a retrospective single-center cohort of 260 patients during TAVR workup, Brendel et al.40 performed dual-source PCD-CT CCTA with artificial intelligence (AI) stenosis quantification (CorEx) and AI-derived FFR (Spimed-AI), both referenced to ICA. For ≥50% stenosis, sensitivity was 96.0%, specificity was 68.7%, and the AUC was 0.82. For FFR ≤ 0.80, sensitivity was 96.8%, specificity was 87.3%, and the AUC was 0.92. Decision analysis indicated that 46.5% vs. 37.3% of cases were classified as not requiring ICA with AI-based FFR vs. diameter-based stenosis.40

During TAVI planning in 60 patients, Sharma et al.41 compared HR (120 kV), UHR (120 kV), and adjusted UHR (90 kV, IQ65) PCD-CT protocols against quantitative coronary angiography. Per-patient AUCs were 0.57 (HR), 0.80 (UHR), and 0.80 (adjusted UHR); per-vessel AUCs were 0.73, 0.69, and 0.87, respectively (UHR vs. adjusted UHR, P = 0.04).41

Figure 3 presents the PCD-CT angiography for a patient undergoing evaluation for TAVR.

Pulmonary and cardio-pulmonary functional imaging

Four clinical studies (Table 4; n = 447 patients) evaluated PCD-CT for pulmonary and cardiopulmonary functional assessment. Scharm et al.42 acquired contrast-enhanced inspiratory PCD-CT and expiratory PCD-CT 5 minutes later in 166 patients, producing technically successful functional maps in 84.7% of patients; mean enhancement values were 325 HU for the pulmonary trunk, 260 HU for the left atrium, and 252 HU for the aorta, with per-phase dose indices of approximately CTDI: 3 mGy and DLP: 110 mGy·cm.

Kerber et al.43 performed a retrospective comparison in 26 patients who received both PCD-CT iodine maps and single photon emission computed tomography (SPECT)/CT for suspected or known chronic thromboembolic pulmonary hypertension (CTEPH). Using multidisciplinary clinical diagnosis as the reference standard and per-patient CTEPH classification, two blinded PCD-CT readers achieved accuracies of 0.85 (95% CI: 0.66–0.94) and 0.88 (0.71–0.96), with sensitivities of 0.90 (0.60–0.98) and 0.90 (0.60–0.98) and specificities of 0.81 (0.57–0.93) and 0.88 (0.65–0.97). The SPECT/CT consensus had an accuracy of 0.73 (0.54–0.86), sensitivity of 0.80 (0.49–0.94), and specificity of 0.69 (0.44–0.86), and between-modality differences were not significant (P > 0.688). The lobar-level perfusion-defect extent on PCD-CT showed moderate correlations with right-heart catheter measures, and the dose was markedly lower with PCD-CT (1.19 ± 0.33 mSv) than with SPECT/CT (6.34 ± 1.68 mSv).43

Saeed et al.44 conducted a retrospective dose-reduction series of 105 patients undergoing high-pitch fast low-angle shot PCD-CT pulmonary CTA with 35, 45, or 60 mL of contrast. Subjective image quality was 4.6 vs. 4.1 for 35 vs. 60 mL (P < 0.001), pulmonary-trunk attenuation was 320–347 HU, and all segmental arteries were assessable.44

Yalon et al.45 performed an evaluation of a three-arm comparative cohort (n = 150) of multi-energy high-pitch PCD-CT pulmonary CTA compared with high-pitch and routine dual-source DE-CT, reporting a CTDIvol of 8.1 vs. 9.6/16.2 mGy and a CNR (P < 0.001), a subjective artery-contrast score of 4.7/5 vs. 4.4/5 and 4.3/5, and fewer motion artifacts with PCD-CT.

Figure 4 illustrates the use of PCD-CT perfusion maps in a patient with a segmental pulmonary embolism.

Aortic, visceral, and peripheral arterial disease

Fourteen clinical studies (Table 5; n = 851 patients) evaluated PCD-CT across the thoracoabdominal aorta, renal–visceral branches, and lower-extremity run-off.

Euler et al.7 performed an intra-individual comparison (n = 40) at a matched dose using high-pitch PCD-CT with 40–55 keV VMIs and reported a CNR of 22 ± 7 at 40 keV vs. 17 ± 8 on EID-CT, with the greatest CNR gain in patients who were overweight; subjective noise increased at 40–45 keV, whereas overall image quality was similar.

Dillinger et al.46 prospectively evaluated arterial-phase PCD-CT of the abdomen (n = 20) with VMI reconstructions from 40–190 keV: a cohort CTDIvol of 7.90 ± 3.92 mGy, a DLP of 330.6 ± 198.5 mGy·cm, and an effective dose of 4.92 ± 2.97 mSv (Radimetrics v3.4, ICRP-103 Monte Carlo; verification with k-factors of 0.015 for the abdomen and 0.014 for the chest). The CNR peaked at 60 keV and the SNR at 70 keV (no significant difference vs. 60 keV, P = 0.294), and subjective image quality was rated optimal at 70 keV. Acquisition and reconstruction settings were automatic 100–120 kVp, a pitch of 0.80, and 1-mm VMI in Qr40.46

Hennes et al.47 conducted an intra-individual comparison (n = 57) of ECG-triggered, high-pitch aortic CTA using PCD-CT [120 kVp; 144 × 0.4 mm; VMI 55 keV; kernel Bv36, quantum iterative reconstruction 3 (QIR-3)] and EID-CT with automatic tube voltage selection (ATVS) at 90/100 kVp [effective collimation: 192 × 0.6 mm; Bv36, advanced modeled iterative reconstruction 3 (ADMIRE-3)]. The CTDIvol was 3.95 ± 0.54 (PCD-CT) vs. 4.97 ± 0.57 mGy (EID-CT) (P < 0.001), and the SSDE was 4.88 ± 0.48 vs. 6.28 ± 0.50 mGy (P < 0.001); the DLP and effective dose were not reported. The CNR was 41.11 ± 8.68 vs. 27.05 ± 6.73 (P < 0.001), with higher overall image quality and luminal contrast on PCD-CT; vessel sharpness was similar, whereas blooming and beam hardening were less pronounced on EID-CT.47

Rippel et al.48 (prospective matched-cohort, ECG-gated high-pitch thoracoabdominal CTA; n = 50) reported an exam-level CTDIvol of 4.0 [interquartile range (IQR): 3.1–4.9] vs. 6.5 mGy (5.5–9.7) and a DLP of 288 (207–402) versus 466 mGy·cm (365–681) (both P < 0.001). On PCD-CT, the SNR was higher at 40 and 70 keV VMIs, and the CNR was higher at 40–45 keV (each P < 0.001) than with EID-CT, and low-keV VMIs salvaged low-contrast studies (diagnostic acceptability 50% → 75% at 40 keV). Acquisition and reconstruction: PCD-CT at 120 kVp, 144 × 0.4-mm collimation, 3.2 pitch, 0.25-s rotation time, Bv36 + QIR-3; EID-CT with ATVS at 100/120/140 kVp, 123 × 0.6-mm collimation, 3.2 pitch, 0.28-s rotation time, I26s + ADMIRE-3. The effective dose was not reported.48

In a retrospective matched run-off CTA cohort (40 PCD-CT vs. 40 EID-CT), the exam-level CTDIvol and DLP were 3.9 (IQR: 3.0–7.6) vs. 3.5 mGy (2.4–5.7) (P = 0.024) and 499 (353–1060) versus 456 mGy·cm (268–753) (P = 0.029), respectively. The SNR on PCD-CT exceeded EID-CT for 40–70 keV VMIs, whereas the CNR exceeded EID-CT at 40–45 keV (vs. 80 kVp EID) and 40–50 keV (vs. 100 kVp EID). Subjective image quality was optimal at 40–60 keV and not significantly different from EID-CT overall. Acquisition and reconstruction: PCD-CT at 120 kVp (QuantumPlus; 144 × 0.4-mm collimation, 0.8 pitch, Qr36 + QIR-3, 1-mm slices, 512 × 512 matrix) with VMI 40–120 keV; EID-CT with ATVS at 80/100 kVp (128 × 0.6 mm, 0.5 pitch, I26s + ADMIRE-3), with identical slice thickness and matrix.49

An in vitro/in vivo study (n = 20) of lower-leg PCD-CT reconstructed at 0.4-mm isotropic resolution found that a sharp quantitative kernel (Qr60) combined with the highest QIR level (QIR-4) best reduced noise without degrading edge definition, yielding the highest qualitative scores. In vivo CTDIvol at the lower-leg level was 2.51 mGy (IQR: 2.50–2.57); the DLP, SSDE, and effective dose were not reported. Acquisition/reconstruction: 120 kVp, CARE Dose4D (image-quality index 145), VMI 55 keV, 512 × 512 matrix, field of view of 205 × 205 mm, kernels Qr44/Qr60/Qr72 with QIR-2/-3/-4; inter-reader reliability was substantial overall (Krippendorff’s α: 0.70–0.71) and excellent for noise (α: 0.84–0.86).50

In a portal-venous intra-individual comparison (n = 50), PCD-CT used 20 mL less intravenous contrast than weight-based EID-CT (90.9 ± 23.0 vs. 111.0 ± 24.0 mL; P < 0.001) and, at 70-keV VMIs, showed no significant differences in hepatic or portal-vein attenuation, noise, SNR, or CNR (all P > 0.0016), with similar qualitative scores and metastasis-detection confidence [odds ratios: 0.58 (95% CI: 0.33–1.01), 1.25 (0.61–2.56), and 1.17 (0.54–2.52), respectively]. Exam-level dose metrics were a CTDIvol of 9.4 ± 4.0 vs. 11.1 ± 7.4 mGy (P = 0.005) and a DLP of 458.7 ± 219.9 vs. 534.6 ± 391.7 mGy·cm (P = 0.01); the effective dose was not reported. Inter-reader agreement for metastasis identification was κ = 0.86 (95% CI: 0.70–1.00) for PCD-CT and 0.78 (0.59–0.98) for EID-CT. Acquisition/reconstruction (fairness): PCD-CT at 120 kVp, 144 × 0.4 mm, CARE Dose4D/Care kV (IQ 145), 70-keV VMI, Br44; EID-CT at 120 kVp, Br44; 4-mm axial and 3-mm coronal/sagittal reconstructions.51

In a within-patient comparison (n = 74) of portal-venous VNC images, PCD-CT vs. EID-CT showed an exam-level CTDIvol of 9.2 ± 3.5 vs. 9.4 ± 9.0 mGy (P = 0.06) and a DLP of 417.9 ± 162.8 vs. 523.4 ± 290.9 mGy·cm (P = 0.026) (32-cm phantom for both). Qualitatively, PCD-CT VNC had higher overall image quality, lower perceived noise, better small-structure delineation, improved noise texture, and fewer artifacts (all P < 0.00001). Quantitatively, PCD-CT VNC had lower attenuation (all P < 0.05), lower noise (P = 0.006), and a higher CNR (P < 0.0001–0.04); the SNR was lower for enhancing structures (reflecting greater iodine removal) but higher in fat. Acquisition/reconstruction (fairness): PCD-CT 120 kV (QuantumPlus), 144 × 0.4 mm, 0.8 pitch, 0.5-s rotation, CARE Dose4D/CARE kV; EI-DECT of 80–90/Sn150 kV, 0.6 pitch, 0.5-s rotation; VNC recon at 4 mm (Br44) for both.52

For infrapopliteal evaluation, a same-day intra-individual study (n = 32) used 60.0 ± 11.0 mL contrast on PCD-CT vs. 139.6 ± 11.8 mL on EID-CT and reported an exam-level CTDIvol of 6.6 ± 2.2 vs. 4.6 ± 3.0 mGy (DLP and effective dose not reported). Acquisition/reconstruction parameters were as follows: PCD-CT UHR mode 120 kV, 120 × 0.2-mm collimation, 0.5 pitch, 0.25-s rotation, 1,024 matrix, Br68, IR-3; EID-CT SE with CARE kV (variable kV), 192 × 0.6-mm collimation, 0.4 pitch, 0.5-s rotation, 512 matrix, Bv44, IR-2. PCD-CT yielded more visualized fibular perforators (R1: 6.4 ± 3.2 vs. 4.2 ± 2.4, P < 0.001; R2: 8.8 ± 3.4 vs. 7.6 ± 3.3, P = 0.04) and greater arterial sharpness (both readers 3.2 vs. 1.7–1.8, P < 0.001), with fewer total occlusions for one reader (0.5 ± 1.3 vs. 0.9 ± 1.7, P = 0.04) and similar subjective noise.53

In abdominal arterial-phase imaging, a retrospective comparison (25 + 25 patients) of UHR PCD-CT vs. EID-CT reported a median CTDIvol of 4.7 (IQR: 3.9–5.1) vs. 7.3 mGy (4.6–12.6) and a DLP of 229 (187–262) vs. 295 mGy·cm (233–595); the effective dose was 3.4 (2.8–3.9) vs. 4.4 mSv (3.5–8.9), calculated as DLP × 0.015 mSv·mGy^-1·cm^-1. PCD-CT showed higher SNR/CNR (significant for renal arteries, P = 0.0432) and higher subjective image quality (P < 0.0001). Acquisition/reconstruction details: PCD-CT UHR 120 kV, 0.25-s rotation, 0.8 pitch, 0.2-mm collimation; 0.6 /0.6 mm axial reconstruction, Bv40 kernel; EID-CT 80–140 kV with automatic dose modulation, 0.28-s rotation, 0.6 pitch, 0.6-mm collimation; I30f/Bv38 reconstruction; non-ECG gated.54

Diagnostic performance against digital subtraction angiography (DSA) was assessed per segment in 109 patients (933 arterial segments): sensitivity was 91% (95% CI 87–94), specificity was 95% (92–96), and accuracy was 93% (≈95% CI: 91–95) overall; territory-level accuracies were 98% (iliac), 92% (femoro-popliteal), and 93% (calf). Inter-reader agreement was good (weighted κ: 0.791; κ: 0.829 for pure-lumen reconstruction). Agreement with DSA grading was κ: 0.905 (CTA) and κ: 0.825 (pure lumen); 95% CIs for κ were not reported.55

Exploratory biomarker research included a radiomics study (n = 55) in which two gray-level co-occurrence matrix features from periaortic adipose tissue distinguished Agatston ≥ 100 vs. 0; ClusterProminence showed the most stable performance under 10-fold cross-validation.29 In 200 patients, Ota et al.56 derived %-calcification on VNCa maps of the abdominal aorta with an AUC of 0.94 (vs. an ACV AUC of 0.90); a 14.8% cut-off yielded 73% sensitivity and 99% specificity for high cardiovascular-risk classification.

A histology-matched study of ascending thoracic aortic aneurysms (n = 14) compared per-patient minimum and maximum aortic-wall thickness on PCD-CT with ex vivo histology. ECG-gated UHR-CTA of the aortic root (120 × 0.2-mm collimation; 66-ms temporal resolution) was followed by non-gated thorax–abdomen–pelvis CTA (144 × 0.4-mm collimation). The effective-dose model, CTDIvol, DLP, and SSDE were not reported. The results (unit of analysis = patient-level paired measures) showed a PCD-CT mean minimum/maximum wall thickness of 1.05/1.69 mm versus histology of 1.66/2.82 mm. Bland–Altman (PCD-CT−histology) analysis revealed a mean bias of −0.61 (minimum) and −1.10 mm (maximum); the authors stated no systematic bias, and numerical limits of agreement were not tabulated. Inter-/intra-observer ICCs were not performed or reported.57

Figure 5 presents aortic PCD-CT and peripheral angiography in a patient with an iliac stent and multiple stenotic segments in the peripheral arterial system.

Evidence summary

Fifty-nine clinical studies published between January 2021 and September 1, 2025, evaluated PCD-CT across the cardiovascular spectrum: 31 coronary, 14 aortic–visceral–peripheral, 4 pulmonary/cardiopulmonary functional, 4 myocardial tissue characterization, and 6 structural-heart/valvular-planning investigations. Across domains, the included studies frequently reported at least one advantage of PCD-CT over EID-CT, including higher spatial resolution at a routine or reduced dose, opportunities for radiation and contrast savings, and robust diagnostic performance in real-world cohorts. Reported examples include vessel-level AUCs of up to 0.99 with UHR modes; reductions in CTDIvol of approximately 40%–60% in matched coronary comparisons; effective doses as low as 0.41 mSv for selected coronary protocols; and 40%–60% contrast-volume reductions in thoracoabdominal CTA, infrapopliteal run-off, and pulmonary embolism protocols without loss of diagnostic confidence.

In routine coronary practice, a large comparative cohort of 7.833 examinations reported an increase in per-patient specificity from 93% to 98% and a reduction in invasive angiography referrals from 13.1% to 9.9%. In the peripheral circulation, diagnostic performance approached that of DSA, with sensitivity around 91% and specificity around 95%. Quantitative capabilities extend beyond morphology: VNC calcium scores have shown ICCs of 0.97–0.99 compared with TNC; iodine-derived ECV estimates differ from cardiac MRI by less than 2% in selected studies; and decision-analytic modeling suggests potential cost savings by reducing downstream testing.

Context within prior syntheses

Since 2023, several narrative or semi-structured reviews have highlighted the clinical promise of PCD-CT but have generally not aggregated core metrics and often predate workflow and economic data emerging in late 2024–2025. Flohr et al.58  presented a seminal cardiac-focused overview in 2023, illustrating early findings such as an approximate 50% calcium-scoring dose reduction and an approximate 11% decrease in blooming-related stenosis overestimation while calling for multicenter outcome and economic evaluations.

Sharma et al.59 offered a clinician-oriented digest combining phantom, animal, and human data, reporting ranges of 29%–41% in noise reduction, 20%–36% in CNR improvements, and 100%/87% in sensitivity/specificity for in-stent restenosis but without formal synthesis and with limited attention to non-coronary applications.

Hagen et al.60 broadened the scope to oncology, cardiovascular, and pediatric imaging with a qualitative three-pillar framework. Hagar et al.61 introduced a more structured approach but included only around 20 cardiac studies up to August 2024. In the vascular domain, Wildberger and Alkadhi62 reviewed feasibility-level studies and emphasized prototype-related bias and the need for prospective surveillance in endovascular aortic repair follow-up. Van der Bie et al.63 provided a focused systematic review on stent imaging.

A separate review by Van der Bie et al.63 addressed clinical aspects of PCD-CT utilization not only in cardiovascular imaging but also in abdominal, thoracic, musculoskeletal, neuro, and pediatric imaging. They specifically investigated stent imaging, coronary stenosis measurements, coronary calcium quantification, plaque component quantification, ECV quantification, TAVI planning, and calcium scoring in the cardiovascular imaging section, following a largely narrative methodology with limited comparative analysis.64

Clinical impact and emerging signals

In coronary imaging, UHR and low-keV reconstructions sharpen lumen–plaque interfaces, mitigate blooming in calcified and stented segments, and enable contrast-sparing protocols while maintaining diagnostic performance. Real-world cohorts suggest improved specificity, fewer unnecessary invasive angiographies, and operational efficiencies when functional adjuncts can be derived from the same dataset. In myocardial tissue characterization, delayed iodine maps and ECV estimates demonstrate high concordance with MRI at segment and patient levels, supporting single-session coronary-plus-tissue assessment when MRI is contraindicated or impractical.

For structural-heart/TAVR planning, PCD-CT supports accurate annular measurements and peripheral-access assessment at reduced exposure; VMI reconstructions can facilitate calcium quantification without additional native scans. Within the pulmonary circulation, multi-energy datasets enable iodine-based perfusion mapping alongside embolus detection, often with lower radiation and reduced contrast loads. In aortic and peripheral vascular applications, low-keV VMIs improve vascular conspicuity—particularly in small-caliber infrapopliteal vessels—while supporting dose- and contrast-efficient protocols; early research also indicates potential quantitative biomarkers.

Challenges and evidence gaps

The evidence base remains dominated by single-center experiences with heterogeneous acquisition/reconstruction (tube potentials, matrix/slice thickness, kernels/iterative strengths, and VMI energies) and variable reference standards (ICA, CMR, and DSA). Quantitative thresholds—for example, plaque-component cut-points and calcium-score categories in VNC/VNI workflows—require harmonization. Health economic and workflow data are encouraging but largely model based; prospective utilization and cost-impact studies across health systems are needed. Finally, multicenter trials linking PCD-CT-guided decisions to hard clinical endpoints remain limited.

Future directions

Priorities include the following: (i) multicenter prospective studies with standardized acquisition and reporting checklists; (ii) consensus on recommended VMI energies and quantitative thresholds by indication; (iii) reproducibility studies for ECV/perfusion metrics and for VNC calcium scoring across vendors and sites; (iv) prospective evaluations of AI-enabled adjuncts (e.g., CT-FFR, radiomics) anchored to outcomes; and (v) robust cost-utility and budget-impact analyses in diverse clinical contexts.

Limitations of the evidence and of this review

Most included studies are observational, single-center studies, with heterogeneity that precludes formal pooling. Accordingly, we report study-level results and observed ranges rather than pooled effects. External validity across institutions and vendors and longer-term outcomes require further study.

In conclusion, across cardiovascular applications, PCD-CT has been reported to demonstrate higher spatial resolution, improved tissue/contrast characterization, and greater opportunities for radiation and contrast reduction. These technical gains, together with early signals of diagnostic and workflow efficiency, support an expanding clinical role for PCD-CT, contingent upon confirmation in multicenter outcome and economic evaluations.