Systems hemodynamics approach for evaluating myocardial risk zone from carotid pressure waveforms in coronary occlusion /reperfusion rat models

Evaluation of the left ventricle (LV) myocardial risk zone (MRZ) after episodes of coronary occlusion and myocardial infarction, has recently gained significant interest. This evaluation can determine the extent of myocardial salvage and assess the effectiveness of therapeutic interventions. This study presents a hybrid intrinsic frequency (IF)-machine learning (ML) methodology for evaluation of MRZ from carotid pressure waveforms. We used standard occlusion/reperfusion rat models (female Sprague Dawley) where the proximal left coronary artery was occluded for 30 minutes, followed by 3 hours of reperfusion. The coronary artery was then reoccluded and 1 ml of a 50% solution of Unisperse Blue Dye was injected. Post-operation, the LV was transversely sliced and photographed. The risk area, visualized as tissue not stained by the blue dye, was traced manually for each LV slice. Subsequently, the overall risk zone was obtained via computerized planimetry. MRZ was quantified as mass percentage of the risk area over the left ventricle (LV) area. IF parameters were computed from carotid pressure waveforms 2 hours after reperfusion and fed into Random Forest classifiers. The cut-off value for mild and severe MRZ classification was set to 50%. The final model was externally tested on additional rats. Our results showed high accuracies for evaluating the true class of MRZ via an IF-ML method. This method has potential clinical impact in management and treatment of patients under risk especially post-MI patients.