Deep Learning-Based Medical Image Classification Using Convolutional Neural Networks for Clinical Diagnosis

Abstract

Computer-aided detection of diseases using machine learning mechanisms on medical images has been an interesting applied research topic in both academia and the healthcare sector. Practical studies aimed at improving clinical decision-making through accurate medical image classification can significantly assist medical professionals in disease diagnosis. This paper presents an investigation into the classification of chest X-ray images for clinical diagnosis using a proposed Convolutional Neural Network (CNN). The Chest X-Ray Images (Pneumonia) dataset containing 5,863 images was utilized, and preprocessing techniques including removal of corrupted and duplicate images, denoising, resampling, label encoding, data standardization, GLCM-based feature extraction, and SMOTE-based class balancing were applied to enhance data quality and model performance. The dataset was divided using a stratified 80:20 train-test split, and the proposed CNN model was trained to classify images into Normal and Pneumonia classes. Experimental results demonstrate that the proposed CNN achieved an accuracy of 99.5%, precision of 99.2%, recall of 99.6%, and an F1-score of 99.4%, outperforming ResNet50 (91.55%), InceptionV3 (94.70%), and VGG16 (96.75%). The findings indicate that the proposed CNN provides highly accurate and reliable medical image classification, making it a promising decision-support tool for early pneumonia detection and improved clinical diagnosis.