KEY POINTS
- This experimental and retrospective clinical study compared four photon-counting detector CT (PCD-CT) acquisition modes for coronary artery stenosis assessment: high-pitch prospective Flash, prospective Sequence, retrospective Spiral, and retrospective ultra-high-resolution (UHR) imaging. Phantom experiments used six 3D-printed coronary stenosis models with known diameter reductions of 21–73%, while the clinical analysis included 41 patients undergoing coronary CT angiography with invasive coronary angiography as the diagnostic reference.
- Clinical acquisition was performed using a first-generation Siemens NAEOTOM Alpha scanner. Patients underwent UHR (n=17), Flash (n=17), or Sequence (n=7) imaging, whereas conventional retrospective Spiral mode was evaluated only in phantom experiments. UHR used finer 0.2-mm detector collimation with retrospective ECG gating, compared with 0.4-mm collimation and prospective gating for Flash and Sequence, making the comparison one of complete acquisition protocols rather than spatial resolution alone.
- UHR achieved the highest subjective image-quality score, averaging 4.12 ± 0.99 out of 5, compared with 3.00 ± 0.86 for Flash (p=0.002) and 3.57 ± 1.10 for Sequence. Nondiagnostic coronary segments were less frequent with UHR than Flash (8.0% vs 29.4%), although subjective blooming and diagnostic-confidence scores did not differ significantly between protocols.
- In the phantom experiment, quantitative calcium blooming was lowest with UHR. Mean blooming ratios were 2.66 ± 2.26 for UHR versus 5.05 ± 2.83 for Flash, but differences between acquisition modes were not statistically significant (all p>0.05). Correct CAD-RADS classification was achieved in 83.3% with UHR, 100% with Sequence, 66.7% with Flash, and 50.0% with conventional Spiral, based on only six stenosis phantoms.
- Better image quality did not consistently translate into higher diagnostic accuracy against invasive coronary angiography. For detecting ≥50% stenosis, patient-level accuracy was 64.7% with UHR versus 76.5% with Flash; for ≥70% stenosis, accuracy was 82.4% versus 76.5%, respectively. The Sequence cohort achieved 100% accuracy for ≥50% stenosis, but all seven patients had disease at that threshold, so specificity could not be estimated and the result cannot establish superior performance.
- Radiation exposure differed substantially between clinical protocols. Mean estimated effective dose was 26.44 ± 9.76 mSv for UHR, 13.65 ± 8.86 mSv for Sequence, and 1.40 ± 0.32 mSv for Flash, with significant differences between acquisition groups (all p≤0.011). These values reflect the complete acquisition configurations, including ECG-gating strategy, collimation and exposure settings; the higher UHR dose cannot be attributed solely to the finer spatial resolution.
- The clinical comparison was limited by small, nonrandomized and imbalanced groups, different indications for imaging, and differences in baseline heart rate and patient characteristics. UHR was predominantly used for pre-TAVR assessment, while other modes served different diagnostic populations. There was no intraindividual protocol comparison, interobserver agreement was not evaluated because readings were performed in consensus, and prospective UHR acquisition was not assessed.
CLINICAL TAKEAWAY
Ultra-high-resolution photon-counting coronary CT improves subjective image quality and can reduce calcium blooming, but its additional diagnostic value must be balanced against substantially higher radiation exposure under the acquisition protocols evaluated. The findings support individualized protocol selection rather than routine use of the highest-resolution setting, and controlled studies are needed to separate the effects of resolution, ECG gating and acquisition technique.