RPS 1503 - AI and deep learning in cardiac imaging: artificial intelligence, motion correction and photon-counting optimisation

Author Block: J. Osoria-Velasquez¹, S. Hyska¹, N. Fink², M. Vecsey-Nagy³, T. S. Emrich⁴, A. Varga-Szemes¹, M. T. Hagar¹; ¹Charleston, SC/US, ²Munich/DE, ³Budapest/HU, ⁴Mainz/DE
Purpose: To assess the incremental benefit of post-hoc convolutional neural network (CNN)–based denoising for coronary CT angiography (CCTA) performed at low radiation doses, compared with conventional iterative reconstruction.
Methods or Background: Consecutive patients who underwent clinically-indicated CCTA on a third-generation dual-source CT system were included. All examinations used a fast-pitch acquisition and a fully individualized contrast protocol comprising 56 tailored combinations of contrast volume and injection rate. Axial datasets were reconstructed with a vascular kernel (Bv40) using iterative reconstruction (Admire 3), with and without additional CNN-based denoising. Two blinded readers assessed signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR), and image noise across proximal and distal coronary segments. Subjective image quality and sharpness were rated on a 4-point scale (4 = excellent). Linear mixed-effects models were used to analyze the impact of body mass index (BMI) on denoising performance.
Results or Findings: Fifty-six patients (51.7±12.8 years old; 57% women) were included, 35 (62.5%) were obese. The mean dose-length product was 99±86 mGy·cm. Compared with iterative reconstruction, CNN-denoising decreased image noise (21 ± 5 HU vs 35 ± 8 HU, p<0.001) and enhanced SNR (22 ±6 vs 12 ±3, p<0.001) and CNR (19 ±5 vs 11 ±3, p<0.001), without altering attenuation values (p>0.1). Subjective image quality improved (median 3 [3–4] vs 3 [2–3], p<0.001), while sharpness remained unchanged. Poor-quality segments decreased from 14.7% to 9.0%, and good-to-excellent quality rose from 52.6% to 75.0%. Noise reduction (−10.3 HU, p=0.007) and gains in SNR (+ 8.0, p=0.008) and CNR ( + 6.9, p=0.011) were consistent across BMI categories (p≥0.29).
Conclusion: Post-hoc CNN-based denoising enables high-quality, low-dose CCTA -even in obese patients- by markedly reducing image noise and improving SNR/CNR while preserving attenuation accuracy and vessel sharpness.
Limitations: The used denoising algorithm is proprietary.
Funding for this study: No funding was received for this study.
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: This retrospective study was IRB-approved.

Author Block: L. Zou; Beijing/CN
Purpose: A novel super-resolution deep learning reconstruction (SR-DLR) technique has been developed for coronary CT angiography (CCTA). The purpose of this study was to compare SR-DLR against conventional hybrid iterative reconstruction (HIR) in coronary stenosis assessment, using invasive coronary angiography (ICA) as a reference, and explore the possible impact on patient-level CAD-RADS classification.
Methods or Background: From September 2023 to November 2024, patients who underwent clinically indicated CCTA and ICA within 2-month interval were prospectively enrolled from 10 hospitals across China. CCTA images were reconstructed with both HIR and SR-DLR and percentage diameter stenosis of calcified, noncalcified, and mixed plaques were quantified and bias was determined from the ICA-derived quantitative coronary angiography measurements. Changes in patient-level CAD-RADS categories across the reconstructions were also assessed.
Results or Findings: 204 patients (mean age, 64.3 years ± 9.1 [SD]; 137 male patients) were included and 605 plaques (175 calcified, 140 noncalcified, and 290 mixed) were identified. Higher agreement with ICA was obtained with SR-DLR as compared to HIR across the plaque type(mean bias:11% vs. 17% for calcified plaques, 0% vs. -4% for noncalcified plaques, and 4% vs. 5% for mixed plaques). While 163 patients remained constant, 25 patients were assigned a lower CAD-RADS category using SR-DLR than assigned using HIR, whereas 16 patients were assigned a higher category.
Conclusion: SR-DLR outperformed HIR for coronary stenosis assessment and led to 20% patient-level CAD-RADS reclassification, potentially enhancing the role of CCTA in the diagnosis and patient management of CAD.
Limitations: Due to major differences in the image characteristics, observers may have been able to visually distinguish between the reconstruction methods. And further dedicated studies are warranted to investigate the possible impact of CAD-RADS reclassifications enabled by SR-DLR on downstream management and patient outcomes.
Funding for this study: This study was supported by the Beijing Natural Science Foundation [Grant No. Z210013, 2021], the National Science Fund for Distinguished Young Scholars [Grant No. 82325026, 2024], the CAMS Innovation Fund for Medical Science[2023-I2M-C&T-A-004], the Non-profit Central Research Institute Fund of Chinese Academy of Medical Sciences[2024-RC320-03] and the National High Level Hospital Clinical Research Funding [2022-PUMCH B-027, 2022-PUMCH B-068]
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: Peking union medical college hospital

Author Block: J. Ma, M. Jing, H. Zhang, X. Ming; Lanzhou, China/CN
Purpose: To construct and validate a predictive model for in-stent restenosis (ISR) following percutaneous coronary intervention (PCI) .
Methods or Background: This retrospective study enrolled patients who underwent coronary computed tomography angiography (CCTA) within one week prior to PCI and follow-up CCTA or invasive coronary angiography (ICA) post-PCI. Fat attenuation index (FAI)
was measured at the target lesion(FAIp), both proximally and distally(FAIpp). CT-derived fractional flow reserve (CT-FFR) at the lesion and the translesional gradient (ΔCT-FFR) were computed. The LASSO was used to screen for factors influencing ISR occurrence. These factors were incorporated into a multivariate Cox regression analysis to construct a nomogram model.
Results or Findings: CCTA-derived FAIp (HR=1.090, 95% CI 1.051–1.131, P < 0.001) and ΔCT-FFR (HR=5.335, 95% CI: 1.085-26.229, P < 0.001) were independent risk factors for ISR. The constructed nomogram model achieved a C-index of 0.869 (95% CI: 0.797-0.941), with good calibration curve fit. The clinical decision curve demonstrated high clinical utility of the model.
Conclusion: The integrated model combining FAIp and ΔCT-FFR based on CCTA enables early prediction of ISR after PCI.
Limitations: First, as a single-center retrospective study, it is potentially subject to selection bias. The extended follow-up period also contributed to a relatively small sample size. Despite this, the study yielded favorable outcomes. Future investigations will incorporate larger cohorts to enhance model stability and include prospective validation. Second, the use of a fixed tube voltage may limit generalizability; future work should explore other voltage levels. It is noteworthy, however, that consistent scanning equipment and parameters eliminated their potential confounding effects on FAI measurements. Finally, semi-automated quantification of FAI and CT-FFR (using Shukun software) may introduce some bias, although all results were reviewed and corrected by physicians, thereby ensuring a reasonable level of reliability.
Funding for this study: This work was supported by the Gansu Provincial Health Industry Research Program for Outstanding and Key Young Talents [Grant No. GSWSQN2023-04], the "Cuiying Graduate Supervisor" Mentorship Program of the Second Hospital of Lanzhou University [Grant No. CYDSPY202003], and the Natural Science Foundation of Gansu Province [Grant No. 23JRRA0997].
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: This study complied with the Declaration of Helsinki and was approved by the Institutional Ethics Committee (Approval No. 2025A-703).

Author Block: G. Coutat¹, F. Tatsugami², D. Touitou-Gottenberg¹, W. Fukumoto², T. Higaki², A. Taniguchi², K. Haioun², Y. Nakamura², M. Ohana¹; ¹Strasbourg/FR, ²Hiroshima/JP
Purpose: Reproducibility of aortic annulus sizing and aortic valve opening area planimetry on pre-TAVR cardiac CT is crucial for procedural planning.
Whether a DLR-based whole heart Motion Correction algorithm (MC-DLR) combined with Super Resolution Deep Learning Reconstruction (SR-DLR) could affect inter and intra reader reproducibility of these measurements, particularly in heavily calcified aortic cusps, remains unknown.
This study aimed to compare inter and intra reader reproducibility of aortic annulus and aortic valve opening planimetry measurements on CT reconstructions without and with MC-DLR.
Methods or Background: 60 consecutive pre-TAVR CT scans stratified by heart rate (30 with HR<75bpm, 30 with HR>75) were retrospectively selected from 2 tertiary centers. Systolic phase was reconstructed with SR-DLR (1024 matrix size) without and with MC-DLR.
4 radiologists independently and randomly reviewed all 120 datasets to assess qualitative image quality, aortic annulus area and aortic valve planimetry.
2 readers repeated all measurements following a 4 weeks delay.
Statistical analysis was performed using Bland-Altman plots and intraclass correlation coefficient (ICC).
Results or Findings: Image quality was higher with MC-DLR (mean 2.82/3 vs 2.67, p=0.07)
Inter-reader agreement for aortic annulus area was excellent and similar without (ICC 0.81, 95% CI 0.78-0.83) and with MC-DLR (ICC 0.80, 95% CI 0.77-0.82).
Inter-reader agreement for aortic valve planimetry was higher with MC-DLR (ICC 0.90, 95% CI 0.86-0.92) than without (ICC 0.83, 95% CI 0.80-0.85).
Intra-reader agreement for both measurements was marginally improved with MC-DLR.
Conclusion: DLR-based whole heart MC may improve the reproducibility of aortic valve area planimetry in systolic pre-TAVR CT.
Limitations: Potential clinical implications of MC-DLR on device selection were not analyzed in this study.
Funding for this study: No funding was received for this study.
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: Approved by the ethics committee of Strasbourg University Hospital

Author Block: H. Shih¹, M-C. Liu¹, S-W. CHAN¹, W-H. Chen¹, J-H. Chen²; ¹Taichung/TW, ²Kaohsiung/TW
Purpose: Blooming artifacts from calcified plaques may lead to overestimation of coronary arteries stenosis (OCAS) using virtual monochromatic imaging (VMI). The aim of this study was to test the ability of virtual non-calcification (VNC) of photon-counting detector CT (PCD-CT) to reduce OCAS in VMI.
Methods or Background: Retrospective analysis of subjects underwent coronary CT angiography from 2025 Jan. to Mar. was performed. Images acquired by PCD-CT were reconstructed by VMI and VNC. Percentage diameter stenosis (PDS) was compared between the two algorithms. Subgroup analyses by stenosis extent, CAD-RADS and calcium score were also performed.
Results or Findings: 344 subjects were initially included. 180 subjects were excluded, 159 due to zero calcium scores, and 21 with stent history. Totally, 164 subjects with 562 plaques were studied. VNC failed to remove calcium in 124 plaques (22.1%). Most of these plaques had density < 1000 HU (average density 887 HU). In 27 dense plaques, VNC erroneously removed intravascular contrast medium. PDS significantly decreased in VNC compared to VMI (VMI: 18.9 ± 16.7; VNC: 8.7 ± 10.5. P<0.01). VNC also showed significant decrease of PDS in three subgroup analyses (all Ps<0.01). For example, subjects with PDS> 50 in VMI decreased from 58.7 ±8.3 in VMI to 26.1 ±13.2 in VNC. Further, VMI may show false positive findings. A 67-year-old male subject showed moderate to severe stenosis (>50%) in VMI. While invasive angiography and VNC images showed no significant stenosis. Overall, subjects with more severe calcified plagues showed more prominent overestimation of stenosis in VMI images.
Conclusion: Our preliminary results showed the benefit of VNC to reduce the overestimation of stenosis evaluated by VMI. VNC may potentially reduce the unnecessary invasive coronary angiography.
Limitations: Our study lacked invasive angiography for most of the cases.
Funding for this study: No funding was provided for this study.
Has your study been approved by an ethics committee? Not applicable
Ethics committee - additional information:

Author Block: M. J. S. Irding; Jönköping/SE
Purpose: In patients referred for TAVI prospectively evaluate the determination of left ventricular ejection fraction (LVEF) and global longitudinal strain (GLS) measured by feature‑tracking multi‑detector CT (MDCT) versus transthoracic echocardiography (TTE).
Methods or Background: 77 consecutive TAVI candidates (mean age 80 ± 7 years; 54% women) underwent clinical TTE and multiphase MDCT. LVEF and GLS were analysed using commercially available software.
Results or Findings: Mean GLS −13.9 ± 4.2% (TTE) and −11.2 ± 4.1% (MDCT) where MDCT underestimated GLS versus TTE (mean bias 2.7 percentage points; p < 0.001). Moderate correlations were observed for LVEF (r = 0.603, p < 0.001) and GLS (r = 0.452, p < 0.001); correlations improved in a subgroup excluding 12 examinations with compromised image quality (LVEF r = 0.688; GLS r = 0.622). Reproducibility of MDCT measurements was high (intra‑reader ICC for LVEF 0.90 and GLS 1.00; inter‑reader ICC for LVEF 0.67 and GLS 0.96).
Conclusion: Our findings indicate that cardiac MDCT is a feasible and repeatable method for assessing LVEF and GLS in TAVI candidates. However, MDCT shows a tendency to underestimate GLS compared to TTE, warranting the establishment of modality-specific threshold values. Stricter quality criteria yield a higher agreement, suggesting that diagnostic modality selection should consider patient-specific factors. Future research should explore the integration of MDCT-derived GLS in clinical practice, particularly in cases where TTE quality is compromised.
Limitations: The main limitations include the variable time interval between echocardiography and MDCT and the relatively small sample size for reproducibility testing. Despite these, the study's prospective design and low intra- and inter-reader variability support the reliability and clinical relevance of our findings.
Funding for this study: Funding for this study was generously provided by Futurum, The Academy for Healthcare Region Jönköping County, and FORSS, the Medical Research Council of Southeast Sweden.
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: It has been granted authorization from the Swedish ethical review authority.

Author Block: T. T. Demmert¹, K. Klambauer¹, B. Schmidt², V. Mergen¹, L. J. Moser¹, T. Allmendinger², T. Flohr², M. Eberhard¹, H. Alkadhi¹; ¹Zürich/CH, ²Forchheim/DE
Purpose: Blooming artifacts from calcified plaques obscure the vessel lumen and cause stenosis overestimation. Spectral coronary angiography with photon-counting CT (PCD-CT) provides virtual monoenergetic images (VMI). While higher VMI energies reduce blooming, iodine contrast is diminished. This study evaluated whether contrast agents with higher atomic numbers (high-Z) preserve vascular contrast at high energies and improve stenosis quantification.
Methods or Background: A phantom with 4 mm and 6 mm rods mimicking vessels containing eccentric calcified plaques (25%, 50%, 75% stenoses) was scanned with a dual-source PCD-CT. Five contrast agents (Iodine, Tungsten, Holmium, Hafnium, Bismuth) were tested. VMI were reconstructed from 40–190 keV in 1 keV increments. Vessel attenuation, contrast-to-noise ratio (CNR), and stenoses were measured. Image quality was qualitatively assessed.
Results or Findings: Iodine attenuation was high at low energies but dropped below 250 HU at >100 keV. Tungsten, Holmium, Hafnium, and Bismuth maintained >250 HU across the range. Iodine CNR was high at low but decreased at high energies, similar to Holmium and Bismuth. Tungsten and Hafnium showed lower CNR at low energies but stable at high keV; Tungsten rose to ~40. Stenoses were overestimated at low energies (24–32.5% at 40 keV) but decreased at high energies (0–13.5% at 190 keV). At 190 keV, Tungsten, Hafnium, and Bismuth showed ≤2.5% overestimation versus Iodine (10–13.5%). Image quality varied: very high-Z agents achieved highest scores, iodine peaked at 55–70 keV but performed worst overall at high energies.
Conclusion: Compared with iodine, very high-Z agents enable superior lumen definition and more accurate stenosis assessment at high VMI energies, minimizing calcium blooming.
Limitations: We used a static coronary phantom with uniform vessel diameters and calcified stenoses, not reflecting in vivo complexity. Patient factors including motion, tortuosity, and heterogeneous plaque morphology were not considered.
Funding for this study: There was no funding.
Has your study been approved by an ethics committee? Not applicable
Ethics committee - additional information:

Author Block: C. R. Jeukens, B. Martens, J. Vandewall, S. Jasper, G. M. Schrijnemaekers, J. E. Wildberger, T. Flohr; Maastricht/NL
Purpose: It is recognized practice to reduce the contrast media (CM) dose in CT as much as clinically feasible. For conventional CT a practical 10-to-10-rule indicates that a 10% lower CM dose can be used for each 10 kV reduction in tube voltage, while maintaining constant iodine enhancement.
This study aims to develop a practical rule how to reduce CM dose in photon-counting detector CT (PCD-CT) when lowering the energy of the reconstructed virtual mono-energetic images (VMI) for parenchymal CT and CTA.
Methods or Background: Spectral abdominal and chest CT phantoms, containing a range of iodine concentrations and ICRU muscle tissue, were scanned using a portal venous phase (PVP) abdominal and a high-pitch CTA protocol on a dual-source PCD-CT. Two fat equivalent rings were used to mimic different patient sizes. Iodine Contrast-to-Noise Ratio (CNR) was measured in VMIs at energies from 40 to 60 keV in 5 keV steps. In 15 abdominal and 15 CTA patient scans (body-mass-index 17-37 kg/m2), the CNR at different VMI energies was retrospectively determined.
Results or Findings: Each 5 keV reduction in the VMI energy range 60-40 keV maintained a similar CNR when reducing the iodine concentration by 11.7-14.4% for PVP scans and 11.8-14.5% for CTAs. Patient scan analysis showed that each 5 keV reduction resulted in a mean 11.4% and 13.7% increase in CNR for PVP and CTA scans. This can be translated to a corresponding reduction in CM dose at constant CNR. A robust, simple 10-to-5-rule was derived: for each 5 keV reduction of VMI energy, the CNR can be maintained with about 10% less CM dose.
Conclusion: Based on this phantom and retrospective proof-of-principle patient study, a practical 10-to-5-rule was derived for contrast-enhanced PCD-CT.
Limitations: Validation in larger patient study needed.
Funding for this study: N/A
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: The proof-of-principle patient study received a waiver of written informed consent from the local ethical committee and institutional review board (METC 2024-0471), due to its retrospective nature.

Author Block: K. Hou, L. Peng, J. Shen, S. Xu, M. Zeng; Shanghai/CN
Purpose: To report the initial experience and evaluate the clinical feasibility of acquiring stress perfusion-derived coronary CT angiography (CCTA) on a newly introduced dual wide-coverage CT, in patients with heart rate >90 bpm.
Methods or Background: This retrospective study included 35 patients with heart rate >90 bpm who underwent both stress dynamic myocardial CT perfusion (CTP) and CCTA within one examination on a dual wide-coverage CT. CTP was followed by a routine rest CCTA within a 10-minute interval, using two separate contrast medium injections. Perfusion-derived CCTA was obtained by manually selecting the peak arterial phase from the dynamic perfusion dataset. Two radiologists jointly evaluated the presence of ≥50% stenosis on both perfusion-derived CCTA and routine CCTA. The diagnostic performance was compared between the two CCTA datasets on a per-segment and per-vessel basis, using invasive coronary angiography as the reference standard. The overall image quality was evaluated using a 5-point scale (1=poor, 5=excellent).
Results or Findings: The mean heart rates during CTP and routine CCTA acquisitions were 102.3 ± 10.3 bpm and 78.9 ± 7.8 bpm, respectively. There was no significant difference in sensitivity, specificity, and accuracy between two CCTA datasets for diagnosing ≥50% stenosis in both per-segment (perfusion-derived CCTA: 92.0%, 96.4% and 95.7% vs. routine CCTA: 93.3%, 97.0% and 96.5%; all p>0.05) and per-vessel analyses (perfusion-derived CCTA: 96.1%, 95.5% and 95.7% vs. routine CCTA: 98.0%, 95.5% and 96.4%; all p>0.05). No significant difference was found in overall image quality between two CCTA datasets (perfusion-derived CCTA: 4.5±0.5 vs. routine CCTA: 4.6±0.5, p=0.508).
Conclusion: It is feasible to obtain stress perfusion-derived CCTA in patients with heart rate >90 bpm using dual wide-coverage CT, where the diagnostic performance of perfusion-derived CCTA was comparable to that of routine rest CCTA.
Limitations: N/A
Funding for this study: N/A
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: This study was approved by the local IRB

Author Block: W. Zhang, Q. Zhou, X. Xu; Hangzhou, China/CN
Purpose: To evaluate the predictive value of CT-based longitudinal changes in EAT for obstructive CAD progression.
Methods or Background: We included 583 suspected CAD patients underwent coronary computed tomography angiography (CCTA) between 01.06.2010 and 30.06.2017, and received follow-up CCTA at an interval of ≥ two years . 107 patients progressed to obstructive CAD. EAT and coronary artery calcium score (CACS) were measured on non-contrast CCTA. Annualized EAT volume change was calculated as the difference between two scans divided by the follow-up duration. Obstructive CAD progression was defined as new-onset ≥50% stenosis in initially non-obstructive patients. The association of EAT with obstructive CAD progression was evaluated using univariate and multivariate logistic regression.
Results or Findings: Annualized EAT volume change was 3.47 cm³ (IQR, 0.33 to 6.73 cm³) in the obstructive progression group and 2.07 cm³ (IQR,–1.16 to 4.97 cm³) in the no obstructive progression group (P = 0.006). In univariate analysis, only age, CACS, baseline largest luminal diameter stenosis, and annualized EAT volume change were significantly associated with obstructive CAD progression. In multivariate analysis, after including variables significant in univariate analysis, annualized EAT volume change remained an independent predictor of obstructive CAD progression (adjusted OR 1.06, P = 0.007).
Conclusion: The annualized EAT volume change is an independent predictor of obstructive CAD progression, suggesting that dynamic EAT monitoring on routine CT scans may improve risk stratification for obstructive CAD.
Limitations: First, this was a retrospective study and used single-center data. Although patients were included consecutively, selection bias could not be avoided. Second, there were differences in the scan interval, with a median of 3.33 years (IQR, 2.67 to 4.33 years). Therefore, the annualized changes may to some extent adjust the potential differences induced by the various follow-up intervals.
Funding for this study: No funding was received for this study.
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: The study was conducted in accordance with the Declaration of Helsinki and approved by the Medical Ethics Committee of the Second Affiliated Hospital, Zhejiang University School of Medicine (No. 2024-0566). In our study, an informed consent waiver was obtained from the Clinical Research Ethics Committee.

Author Block: C. Schuppert¹, S. Barus¹, M. Soschynski¹, T. Allmendinger², F. Bamberg¹, T. Krauß¹, C. L. Schlett¹; ¹Freiburg/DE, ²Forchheim/DE
Purpose: Photon-counting CT (PCCT) introduces novel scan modes that assist coronary CT angiography (CCTA) but increase protocol complexity. We developed and evaluated a scanner-integrated decision support solution to optimize protocol selection.
Methods or Background: In this prospective, two-center study, adult patients referred for CCTA were scanned on identical dual-source PCCT systems (NAEOTOM Alpha, Siemens Healthineers) with standard preparation. One site used the decision support solution (myExamCompanion, Siemens Healthineers), whereas the control site relied on conventional operator-driven protocol selection. Patient flow was not controlled. The decision support solution considered pre- and intra-scan input parameters reflecting coronary artery disease risk and cardiac dynamics (age, BMI, stents, Agatston score, heart rate, heart rate variability). Guided by the principle of “minimized exposure, full diagnostic performance”, it automatically selected from Quantum (70/90 kVp), Quantum Plus (120/140 kVp), Ultra High-Resolution (UHR), and Spectral UHR acquisition modes, as well as ECG-gating strategies (high-pitch spiral, prospective sequence, retrospective spiral). Evaluation focused on radiation dose, diagnostic confidence, and image quality, as measured by coronary signal-to-noise ratio (SNR) in a low-risk subset.
Results or Findings: A total of 1,304 patients underwent CCTA, with 727 (56%) scanned using the decision support solution. Baseline patient characteristics were comparable between groups. Use of the decision support solution was associated with lower dose length product (median 269 vs. 370 mGycm), mainly through greater Quantum mode utilization (34% vs. 11%). SNR was slightly lower in Quantum vs. Quantum Plus mode (median 15.3 vs. 17.7, p=0.07). The incidence of CAD-RADS≥3 was 21% vs. 29% with vs. without the solution, while CAD-RADS N was <1% in both groups.
Conclusion: The decision support solution optimizes PCCT protocol selection for CCTA, preserving image quality and interpretability while lowering radiation dose.
Limitations: Single-vendor, two-center study with site-based, non-randomized allocation.
Funding for this study: No funding was provided for this study.
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: Ethics Committee of the Medical Center – University of Freiburg (No. 21-1469, approved on September 21, 2021, and amendments).

Author Block: G. Gambato, D. Vignale, A. Palmisano, A. Esposito; Milano/IT
Purpose: To compare photon-counting CT (PCCT) and conventional energy-integrating detector CT (EID-CT) in the in-vivo evaluation of coronary artery stents.
Methods or Background: This retrospective study was conducted in patients with coronary stents undergoing cardiac PCCT between May 2024 and September 2025 (NAEOTOM Alpha, Siemens; slice thickness 0.4 mm; kernels: Bv56, n=26; Br56, n=3; Qr56, n=1; Bv48, n=2; Br40, n=1), who had undergone a prior examination with EID-CT (SOMATOM Definition FLASH, Siemens; 120 keV, I36f kernel). Stent models were known. Images were evaluated according to qualitative parameters (Overall Image Quality, Sharpness, Noise, Blooming, and Diagnostic Confidence) by a single rater with three years of experience in cardiovascular radiology, using a five-point Likert scale. Blooming was quantified by comparing nominal stent diameter (median 3.00[2.75,3.50] mm) with measured stent diameters and by assessing intra- and extra-stent attenuation. The Wilcoxon signed-rank test was used for comparisons.
Results or Findings: Sixty stents in 33 patients (M:F 32:1, 80.5 [IQR 72.5,84.0] years) were analyzed. PCCT yielded higher scores than EID-CT across all qualitative parameters (p<0.001). Median improvements were +1 [IQR 0–2] for image quality, +1 [IQR 0–2] for sharpness, +1 [IQR 0–2] for noise, +2 [IQR 1–2] for blooming, and +1[IQR 0–1] for diagnostic confidence. All quantitative comparisons were significant (p<0.001). PCCT measurements were closer to the nominal stent diameter (median difference -0.77[IQR -1.10,-0.54]mm) than EID-CT (-1.05[IQR -1.40,-0.75]mm, and PCCT yielded larger in-stent luminal diameters than EID-CT (median difference +0.30[IQR 0.00,0.50]mm). Moreover, PCCT showed less attenuation difference between intra- and extra-stent measurements compared to EID-CT (median 139 [IQR 37–213] vs 236 [IQR 158–312] HU), both indicating a reduction of blooming artifacts.
Conclusion: Compared to EID-CT, PCCT improves the in-vivo evaluation of coronary stents by reducing blooming.
Limitations: Small sample.
Funding for this study: None.
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: Ospedale San Raffaele (CTMyoC)