Heart Disease Prediction Using Fuzzy Logic-Based Image Processing and Classification Techniques

Abstract

The medical field deploys heart disease prediction as a vital operation for early detection to minimize serious health risks. Researchers in this study introduce a new method for heart disease prediction which combines fuzzy logic with image processing together with classification methods. This methodology utilizes fuzzy methods to manage imprecision together with uncertainty found in medical images and numerical information for obtaining more accurate and interpretable outcomes. The initial stage requires fuzzy imputation for handling missing values and fuzzy scaling which transforms features into fuzzy sets for better representation of medical data uncertainties. Define relevant medical imaging regions through fuzzy C-Means clustering before evaluating tissue patterns for heart disease indicators by analyzing these fuzzy texture elements. The combination of fuzzy-genetic algorithms selects significant features through optimized feature space improvements while fuzzy decision trees provide clear means to rank and select features. The system utilizes Mamdani fuzzy inference systems as the final stage to classify heart disease severity based on expert model predictions. Through fuzzy support vector machine implementations the system minimizes data imprecision and overlaps to boost its classification precision. The proposed heart disease prediction method adopts fuzzy machine learning integration to optimize accuracy levels. Image segmentation occurs through Fuzzy C-Means clustering and Local Binary Patterns (LBP) extract texture features before Fuzzy Genetic Algorithms (FGA) select the features. The model received performance evaluation through assessment of its accuracy as well as sensitivity and specificity tests and AUC-ROC metric. The analysis reveals predictive strength through an AUC-ROC value of 0.96 as well as 96.4% accuracy and 93% sensitivity alongside 38% specificity. Cross-validation techniques produced average accuracy of 94% through five-fold validation tests. The integration of fuzzy logic with traditional machine learning proves effective for precise heart disease prediction as it deals effectively with medical data uncertainty and imprecision.