Deep Learning Technique for The Classification of Lung Diseases from X-ray Images

Pictures obtained from a chest X-ray can assist in diagnosing and screening various illnesses, including pneumonia, mass, lung opacity, TB, and nodules. The convolutional neural network (CNN) has been devised to predict pulmonary disease. In this investigation, we classify lung disease based on pictures from chest X-rays using a technique known as transfer learning. This method was chosen because it allows us to make more accurate diagnoses. Transfer learning has proven effective for detecting diverse anomalies in limited medical imaging datasets, leading to notable outcomes. This study involves utilizing a dataset comprising a collection of chest X-ray photos obtained from the Kaggle website, including nine different lung diseases (COVID, mass, effusion, lung opacity, nodule, pneumonia, pneumothorax, pulmonary fibrosis, and tuberculosis), as well as images of individuals without any lung ailments. The objective is to employ several transfer learning models (VGG19, Inception V3, Efficient Net V2m, Xception, Mobile Netv3, and Dense net201) to train these images, classify them, and then select the most influential Model for addressing this task. Most models yielded promising results, particularly dense net201, mobile net, and VGG19, which achieved more effective yields with 95.49%, 94.89%, and 93.69% accuracy rates, respectively.