Convolutional Neural Networks (CNNs) represent a significant advancement in the field of medical imaging, redefining diagnostic and analytical methods. Their effectiveness is particularly notable in the identification and classification of abnormalities, promoting the early detection of conditions suchas cancer, brain injuries, and cardiac issues. In this thesis, CNN-based models were compared to decide the most efficient one for the task of classifying pulmonary X-ray images, aimed at diagnosing the presence or absence of pneumonia. This work highlighted the potential of CNNs in practical applications,underscoring their relevance and efficacy in image-based medical diagnosis.