Pneumonia is a life-threatening disease caused by an infection or infection in the human lungs. Early diagnosis of pneumonia is an important part of effective treatment. The recent development of deep learning, which involves many layers to understand hierarchical data representation, has achieved the consequences of the state of many things, especially in the analysis and distribution of human diseases. Therefore, to improve the performance of lung cancer diagnosis, it is necessary to use automatic models based on deep learning models, which can detect images at high X-ray and facilitate lung diagnosis procedures for novices and patients. This paper develops a neural network (CNN) model for diagnosing lung disease using chest X-ray images. The proposed framework has two main stages: image preprocessing stage and feature extraction and image classification stage. The proposed CNN model provides high results with high performance, recall, F1 score, and accuracy of 98%, 98%, 97%, and 99.82%, respectively. According to the results, the proposed CNN model based on lung disease diagnosis achieved similar and accurate results and outperformed other previous deep learning models such as spare part (ResNet 50) and VGG16. It also goes beyond existing models recently mentioned in the literature. Therefore, the significant performance of CNN model-based lung diagnosis on all performance measures can provide better patient care and reduce mortality.