CNN - Based Medical Image Classification Models for the Identification of Pneumonia and Malaria

Across the globe, deep learning methodologies are rapidly transforming the medical field. Among the rapidly expanding domains, medical image categorization stands out as a crucial area for developing effective intelligent systems. Consequently, our research employs various CNN variations to establish a robust and reliable model for medical image categorization. Within this study, we utilize two distinct medical imaging datasets: one comprising malaria cell images and the other featuring chest X-ray images for pneumonia diagnosis. To enhance the model's accuracy, we fine-tune its performance by adjusting parameters such as layer count and activation functions. This approach empowers researchers to identify optimal CNN parameters for image classification and observe how model behavior evolves with changing image types. The presented models undergo validation using precision metrics, including F -score, specificity, and accuracy. Notably, in the malaria and pneumonia datasets, our model achieves accuracy rates of 96 % and 95%, respectively.