Enhanced Pneumonia Detection Through Advanced AI-Driven Hybrid Models: A Comparative Study of Deep Learning Architectures
Honey Joseph, J. Anitha
Abstract
Pneumonia is an infection which is affected on lungs. The reason for infection can by bacteria, virus or fungi in humans. This is characterized by a severe cough with phlegm, chills, and fever. Pneumonia continues to be a significant global health concern. It is the necessity of early and accurate diagnosis is prudent to improve patient outcomes and reduce mortality rates. Now a days with the integration of advanced technologies, mainly artificial intelligence (AI) and medical imaging, we have observed promising results in enhancing pneumonia detection. The objective was to evaluate and examine different deep learning models to recognize pneumonia affected X-ray pictures in various instances, thereby identifying the best model. The proposed platform collects and processes datasets, builds different learning models for pneumonia classification, compares the models, and selects the best model. It delves into the potential of AI-driven models to analyze vast amounts of data, leading to faster and more accurate diagnoses. A comparative examination of each model in different settings may offer a deeper knowledge of each model's effectiveness in diagnosing pneumonia, resulting selection of deep learning models for pneumonia detection easier. In this study, we assess the effectiveness of pre-trained CNN models as feature extractors, followed by a variety of classifiers for distinguishing abnormal and normal chest X-rays. Statistical results show that pretrained CNN models combined with supervised classifier algorithms can be highly useful in analyzing chest X-ray images, particularly for detecting pneumonia. According to the results, the ResNet-50 model performed better on large datasets.