153 Automated Quantitative Perfusion CMR With Simplified Dual-Bolus Contrast Protocol observers graded the diagnostic confidence as good–excellent (4–5 points in Likert scale) in the majority of cases (92% for observer 1, 68% for observer 2) (Table 2, Figure 5). Figure 5. Results of qualitative assessment of image quality by two independent level 3 CMR experts Table 2. Image quality assessment of the dual-bolus protocol Measure Mean value Presence of artifacts 3.7 ± 0.7 Presence of noise 3.3 ± 0.6 Overall impression of image quality 3.4 ± 0.6 Certainty in diagnosis 4.3 ± 0.6 Diagnostic accuracy of QP CMR workflow ROC analysis showed the optimal cut-off value for regional stress MBF ≤ 1.84 ml/g/min to predict significant coronary artery obstruction with a sensitivity of 89% [95% Confidence Interval (CI): 52% to 100%)], specificity of 70% [95% CI 55% to 83%], positive predictive value (PPV) of 36% [95% CI 26% to 48%] and negative predictive value (NPV) of 97% [95% CI 84% to 100%]. Area under the curve (AUC) was 0.79 [95% CI 0.66 to 0.89]. Visual assessment of only color pixel QP maps by CMR experts yielded an accuracy of 80% [95% CI 59 to 93%], with sensitivity of 90% [95% CI 56% to 100%] and specificity of 73% [95% CI 45% to 92%], PPV of 69% [95% CI 39% to 91%], NPV of 92% [95% CI 62% to 100%]. In addition, combined visual analysis of both color pixel QP maps and conventional gray-scale first-pass perfusion images yielded an even higher diagnostic accuracy of 84% [95% CI 64% to 95%]; with sensitivity of 70% [95% CI 35% to 93%] and specificity of 7
RkJQdWJsaXNoZXIy MTk4NDMw