Cardiac Ischemia Detection Using Self-Organizing Map Feature Selection Method

Cardiac ischemia, a prevalent cause of heart failure, remains the leading cause of death in Iran. Early diagnosis of this condition is crucial, and electrocardiogram (ECG) signal processing techniques offer valuable insights. By analyzing ECG signals from ischemic patients and comparing them to those of normal individuals, we can extract relevant features for classification. Our study extracted some features from ECG signals, including statistical, time-based, frequency, and time-frequency features. We employed the self-organizing maps method (SOM) to identify the most informative features. Subsequently, we evaluated the classification accuracy using several machine learning algorithms, including support vector machines (SVM), Naive Bayes, Linear Discriminant Analysis (LDA), and Logistic Regression. The results demonstrated the superiority of the SOM method in selecting the best features compared to PCA, with accuracies of $\mathbf{9 1. 2 \%}, \mathbf{8 4. 3 \%}, \mathbf{7 2. 5 \%}$, and $88.9 \%$ for SVM, LDA, NB, and LR methods, respectively. Our findings contribute to the early detection and management of cardiac ischemia, potentially improving patient outcomes.