Accuracy of prospectively ECG-triggered very low-dose coronary dual-source CT angiography using iterative reconstruction for the detection of coronary artery stenosis: comparison with invasive catheterization.

Objective To evaluate the image quality and diagnostic accuracy of very low-dose computed tomography (CT) angiography (CTA) for the evaluation of coronary artery stenosis. Background Iterative reconstruction (IR) has shown to substantially reduce image noise and hence permit the use of very low-dose data acquisition protocols in coronary CTA. Methods Fifty symptomatic patients with an intermediate likelihood for coronary artery disease underwent coronary CTA (heart rate: 59 +/- 5 bpm, prospectively ECG-triggered axial acquisition, 100 kV, 160 mAs, 2 x 128 x 0.6 mm collimation, 60 mL contrast, 6 mL/s) prior to invasive coronary angiography. CTA images were reconstructed using both standard filtered back projection (FBP) and a raw data-based IR algorithm [Sinogram Affirmed Iterative Reconstruction (SAFIRE), Siemens Healthcare]. Subjective image quality (four-point Likert scale from 0 = non-diagnostic to 3 = excellent image quality), image noise, signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR), as well as the presence of coronary stenosis >50% were independently determined by two observers. Results The mean dose-length product was 46.8 +/- 3.5 mGy cm (estimated effective dose 0.66 +/- 0.05 mSv). IR led to significantly improved objective image quality compared with FBP (image noise: 41 +/- 12 vs. 49 +/- 11 HU, P < 0.0001; CNR: 16 +/- 8 vs. 12 +/- 4, P < 0.0001; SNR: 13 +/- 7 vs. 10 +/- 3, P < 0.0001). Four coronary segments were not evaluable on FBP data, whereas all segments showed diagnostic image quality with IR. To detect significant coronary stenosis, sensitivity, specificity, positive predictive value, and negative predictive value were 69% (11/16), 97% (175/180), 69% (11/16), and 97% (175/180) per vessel with FBP data sets, respectively. With IR data sets, the corresponding values were 81% (13/16), 97% (178/184), 68% (13/19), and 98% (178/181). These differences were not statistically significant (P = 0.617). Conclusions Raw data-based IR significantly improves image quality in very low-dose prospectively ECG-triggered coronary dual-source CTA when compared with standard reconstruction using FBP.