Enhanced Phonocardiogram Analysis Through Troponin Test for the Myocardial Injury Detection

Myocardial Infarction (MI) is a condition that necessitates prompt medical intervention. Current examinations using ECG and Troponin-test have the disadvantage of diagnosing MI. Phonocardiograms (PCG) represent the physiological and pathological cardiac, with the data acquisition each cycle being 0.8 s. This research aims to develop a non-parametric correlation model between PCG-MI signal feature results and the troponin tests. The signals were obtained using an electronic stethoscope. The acquired signals in (.wav) were extracted into each cycle, time, and statistical domain features. Twelve selected features were tested for normality of distribution using SaphiroWilk. The Principal component regression was used to model the relationship between the feature and troponin test. Then, we found that 12 features were categorized into non-normal distributions. The correlation test between STEMI and NSTEMI features resulted in insignificant correlations, suggesting the model’s feasibility. The proposed principal component modeling achieved an accuracy of $\mathbf{9 1. 0 7 \%}$. Multivariate analysis, aligning with earlier studies on the features of PCG signals, has been represented by transfer functions. This study can serve as a basis for clarifying patients exhibiting pathological symptoms for MI detection.