FIGURE 2.  (A and B) In patients with T1 – T2 SDS of more than 2, linear regression analysis (A) showed good correlation between T1 and T2 SDS (y = 0.89x +2.98; r = 0.67; SEE = 0.1; P < 0.0001), whereas Bland–Altman method (B) showed shift in mean value of difference of +2.67 ± 2.73. (C) ROC curve analysis identified cutoffs of 3.0 for T1 and 2.0 for T2 SDS as best separating patients with and without coronary artery disease. (D) With these thresholds, T1 resulted in global diagnostic accuracy significantly higher than that of T2 imaging (0.79 vs. 0.70, P < 0.001).