Magnetization-transfer flow-independent dark-blood delayed enhancement cardiac MRI optimizes discrimination of ST-elevation myocardial infarct borders

Figures

• 

  Fig. 1. Flow chart of the present study. STEMI, ST-elevation myocardial infarction; CMR, cardiac magnetic resonance; LGE, late gadolinium enhancement; MT-FIDDLE, magnetization-transfer flow-independent dark-blood delayed enhancement

• 

  Fig. 2. Confidence in delineation of infarct from blood for all infarct types combined between “bright-blood”-LGE and MT-FIDDLE. The number of patients with excellent confidence in infarct segmentation clearly increased overall, as well as for the individual infarct types. The difference in confidence in infarct segmentation between “bright-blood”-LGE and MT-FIDDLE was highly significant overall (p < 0.001), for subacute (p = 0.00167) and chronic (p < 0.001) infarcts, but not yet significant for acute infarcts (p = 0.083)

• 

  Fig. 3. Confidence in delineation of infarct from blood for all infarct types combined between “bright-blood”-LGE and MT-FIDDLE. The number of patients with excellent confidence in infarct-blood delineation clearly increased over all, as well as for the individual infarct types. The difference in confidence in infarct-blood delineation between “bright-blood”-LGE and MT-FIDDLE was highly significant overall (p < 0.001), significant for acute infarcts (p = 0.026) and highly significant for subacute (p < 0.001) and chronic (p < 0.001) infarcts

• 

  Fig. 4. Bias (mean difference) between infarct volumes obtained from “bright-blood”-LGE and MT-FIDDLE sequences by Bland-Altmann analysis for different SD-values used for scar segmentation based on the standard deviation method. Only patients were included where the confidence in scar segmentation for “bright-blood”-LGE was excellent. For 3-SD, the lowest bias between “bright-blood”-LGE and MT-FIDDLE sequence was found

• 

  Fig. 5. Difference in scar volume between “bright-blood”-LGE and MT-FIDDLE sequence as a function of the time delay between the sequences. Due to clinical workflow reasons, the “bright-blood”-LGE images were always acquired prior to the MT-FIDDLE images. No significant correlation between volume difference and time delay was observed both for the FWHM method (r = 0.04, p = 0.679) and the 5 SD method (r = 0.02, p = 0.8377)

• 

  Fig. 6. For patient 1, hyperenhancement by MT-FIDDLE clearly identified infarcted myocardium in the anterior wall and for patient 2 in the posterior wall with extension to posterolateral papillary muscle. MT-FIDDLE detected a small infarct in the anterior wall of patient 3 and a lateral wall infarct with extension to the posterolateral and the anterolateral papillary muscle in patient 4. In none of these four patient cases did “bright-blood” hyperenhancement clearly identify all infarcted regions. Infarct identification by “bright-blood”-LGE was particularly poor at the subendocardial border