Measuring the degree of fibrosity of myocardial scars from late gadolinium-enhanced cardiovascular magnetic resonance images

We seek to define and measure the degree of fibrosity of the scars in myocardial infarction (MI) patients, and to see if it helps in separating a group of MI patients who have had episodes of ventricular tachycardia (VT) after the infarction from a group of MI patients who have not. Late gadolinium-enhanced cardiovascular magnetic resonance images of 20 MI patients with VT and 34 without were taken, and the scars demarcated by cardiologists. Various fibrosity measures were defined through weighting the normalised signal intensity distributions of the scars. The fibrosity measures were used as input to a classifier together with the left ventricular ejection fraction (LVEF) and the relative scar size. The classification problem being separating the two patient groups and done in a leave-one-out fashion. At a sensitivity of 90%, combining the best fibrosity measure with LVEF gave a specificity of 91% (with 95% confidence interval (CI) 76-98%), whereas LVEF combined with relative scar size yielded 85% (CI: 6994%) and LVEF alone 82% (CI: 66-92%). Capturing the degree of fibrosity of myocardial scars through weighting the scars´ normalised signal intensity distributions might be of assistance in evaluating the risk of MI patients getting VT.