Feasibility of fast cardiovascular magnetic resonance strain imaging in patients presenting with acute chest pain

Figures

• 

  Fig 1. a. Representative example of fSENC manual contouring in endsystole and final report. Left 6 images in clockwise direction: 2 chamber, 3 chamber, 4 chamber view (long-axes for circumferential strain calculation), short basal, mid and apical view (short axes for longitudinal strain calculation). Right image: final report after contouring with global and segmental strain results. b. Representative example of fSENC bull’s-eye maps of longitudinal strain data (left) and circumferential data (right). Color code: blue represents normal contractility (strain<-17), green represents slightly reduced contractility (-10> strain >-17) and yellow depicts dysfunctional segments (strain>-10). Colors vary according to the exact segmental strain value. c. Differentiation between multi-vessel disease and cardiac, non-oCAD using cine and fSENC raw images. A: oCAD—multi-vessel disease 1. cine end-diastole, 2. cine end-systole, 3. fSENC in end-systole/ B: cardiac, non-oCAD pathology 1. cine end-diastole, 2. cine end-systole, 3. fSENC in end-systole.

• 

  Fig 2. Study duration timeline.

• 

  Fig 3. a Comparison of ROC curves: Single diagnostic methods. The area under the curve (AUC) for GCS and GLS was 0.851 and 0.873 respectively. Both values were statistically significant (p<0.0001). GCS and GLS performed significantly better than ECG (AUC: 0.596), hscTnT (AUC:0.625) and EF (AUC: 0.503) (GCS vs. ECG: p<0.015; GCS vs. hscTnT: p<0.05; GLS vs. ECG: p<0.004; GLS vs. hscTnT: p<0.035; GCS vs. EF: p<0.0015; GLS vs. EF: p<0.0005). The AUC for the number of segments with a strain >-17 and >-10 was 0.769 (p<0.0001) and 0.795 (p<0.001), respectively. Segmental strain performed significantly better than ECG and EF (segments strain>-17 vs. ECG: p = 0.06; segments strain >-17 vs. hscTnT: p = 0.22; segments strain>-17 vs. EF: p<0.03; segments strain>-10 vs. ECG: p<0.05; segments strain>-10 vs. hscTnT: p = 0.16; segments strain>-10 vs. EF: p<0.01;) but was inferior to global strain (GCS vs. segments strain>-17: p = 0.25; GLS vs. segments strain>-17: p = 0.11; GCS vs. segments strain>-10: p = 0.34; GCS vs. segments strain>-17: p = 0.25). b Logistic regression analysis: Combined diagnostic methods. The area under the curve (AUC) for GCS and GLS in combination with routine clinical…

• 

  Fig 4. Nomogram. Nomogram for low (0.9–1.7% risk of MACE) and intermediate (12–16.6% risk of MACE) HEART score: Pre-Test Probability according to HEART score, Post-Test Probability as a combined diagnostic input of the HEART score and fSENC results (LR+ 6.24/ LR– 0.2). Accordingly, post-Test probability of low and intermediate HEART score with negative fSENC results was 0.2% and 3.5% respectively, whereas for positive fSENC results a 9% and 50% post-test probability could be assessed.

• 

  Fig 5. a and b Boxplot of GCS (3a) and GLS (3b) values in the fSENC triage groups 0) non-cardiac 1) oCAD 2) cardiac, non-oCAD. GCS was significantly better for differentiation between the non-cardiac group and the oCAD group, while there was no significant difference between the oCAD group and cardiac, non-oCAD group. Values for GLS differed significantly between all groups.

• 

  Fig 6. a and b Boxplot of number of dysfunctional segments in the three patient groups (4a: strain >-10 4b: strain >-17) based on 16 short axis segments and 21 long axis LV-segments. The number of severely dysfunctional segments was significantly different between the non-cardiac and oCAD group, differences between the oCAD group and cardiac, non- oCAD group were not significant. The number of slightly dysfunctional segments was a significant parameter for differentiation between all three patient groups.