Pneumonia detection in chest X-ray images using Convolutional Neural Network and fuzzy VIKOR

Pneumonia is a life-threatening respiratory disease that requires early and accurate diagnosis for effective treatment and reduced complications. However, conventional diagnostic methods such as PCR are often time-consuming, equipment-dependent, and limited to specialized medical centers. This study introduces a novel anomaly detection framework for pneumonia diagnosis using advanced machine learning techniques applied to chest X-ray images. To enhance classification performance, the framework integrates several feature selection methods, including Correlation-based Feature Selection (CFS) to evaluate feature relevance, Fisher Score to rank features based on discriminative power, Maximum Information Coefficient (MIC) to capture complex dependencies, and Local Learning-based Correlation Feature Selection (LLCFS) to improve accuracy by considering local feature correlations. To further enhance classification performance, this study introduces the first-ever application of Fuzzy VIKOR in pneumonia detection. This fuzzy logic-based ensemble method effectively handles uncertainty in medical imaging data, leading to more balanced decision-making when dealing with conflicting information. The proposed model was trained on a chest X-ray dataset and evaluated using key classification metrics, including accuracy, recall, precision, and F۱-score. Experimental results confirm that the model outperforms baseline methods across all metrics, achieving an accuracy of ۹۸.۳۴%. These findings validate the effectiveness of the proposed framework and highlight its high potential for real-world deployment in AI-driven computer-aided diagnosis (CAD) systems, particularly in hospitals and telemedicine applications.

کلیدواژه ها:

Pneumonia Detection ، Anomaly detection ، Fuzzy MCDM ، Ensemble Feature Selection ، Machine Learning