Oral Abstract Session
SCMR 22nd Annual Scientific Sessions
Large animal and single-center patient studies have shown that chronic infarctions can be detected and characterized without contrast agents using native T1 maps at 3T. We investigated the diagnostic accuracy of native T1 mapping at 3T in a multi-center setting with interobserver assessments.
Patients(n=105) with single prior MI were recruited in medical centers from US, South Korea and China respectively. Native T1 mapping and LGE were performed in patients with a prior MI (mean of 10.1 yrs; (IQR 5.3-16.5 yrs)) since the acute MI. Infarct size was measured as the percentage of total LV myocardial volume and transmurality was measured using the centerline chord method by two experienced observers.
Fig.1 shows representative findings in patients. LGE images and T1 maps were not different for measuring infarct size by both observer 1 (O1- LGE: 11.9±8.0 %; native T1: 11.9±7.4%; p=0.85) and observer 2 (O2 - LGE: 11.8±7.2%; native T1: 12.1±7.5%; p = 0.08). Both observer’s measurements showed good agreement between the two techniques for measuring infarct size: Results from linear regression analyses were: O1: R2=0.91; and O2: R2=0.93. There were no significant differences between the infarct size measurements determined by the two observers on LGE images（p=0.72) and native T1 maps (p=0.36). Linear regression analyses (LGE: R2=0.91; native T1: R2=0.90) and Bland-Altman analyses (LGE: bias=0.19±2.4%; native T1: bias=-0.18±2.4%) showed excellent agreement between the two observers for measuring infarct size. Sensitivity of the native T1 maps for detecting infarct size were: 84% (O1) and 88% (O2) and the corresponding specificity were 92% (O1) and 96% (O2). Area under the curve for each observer were: 0.92 (O1) and 0.97 (O2).
LGE images and native T1 maps were not different for measuring infarct transmurality [O1: LGE: 47.2±19.0%; native T1: 49.1±15.8%; p=0.06; and O2 - LGE: 49.0±16.1%; native T1: 49.1±16.1%; p=0.86. Both observers’ measurements showed good agreement between the two techniques for measuring infarct transmurality: Linear regression analysis: O1: R2 =0.71; O2: R2=0.70. There were no significant differences in the transmurality measurements by the observers on LGE images（p=0.06) and native T1 maps (p=0.99). Bland-Altman (LGE: bias=-1.79±9.7%; native T1: bias=-0.01±9.4%) and linear regression analyses (LGE: R2=0.73; native T1: R2=0.68;) showed good agreement between the two observers for measuring infarct transmurality. Sensitivity and specificity of the native T1 maps to detect infarct transmurality were 71% ,92%(O1) and 74%,93% (O2), respectively. Area under the curve for T1 maps detecting chronic MI were 0.82 (O1) and 0.85 (O2) respectively.
Native T1 mapping can accurately characterize chronic MI; thus, it is a reliable alternative for scar imaging in chronic MI patients, who are contraindicated for Gd-based contrast agents.