Deep Learning-Based Framework for Lung Disease Diagnosis using Integrated CNN and RNN

The precise identification and categorization of pulmonary ailments are of utmost importance in the timely recognition and strategic management of respiratory disorders. This research paper presents a novel approach for lung disease detection and multi-classification by proposing an integrated model that combines convolutional neural network (CNN) and recurrent neural network (RNN) architectures. The model has been trained using a dataset comprising 10,000 lung images sourced from a repository on Kaggle. The initial step involves conducting binary classification in order to differentiate between lung images that exhibit signs of disease and those that are considered healthy. Following this, the proposed study employs a multi-classification approach to categorize the afflicted cases into four distinct lung disease classifications, specifically pneumonia, COVID-19, bronchitis, and tuberculosis. The performance of the proposed integrated CNN and RNN model is noteworthy in both binary and multi-classification tasks. The binary classification model demonstrates a notable overall accuracy rate of 95.53%, effectively discerning whether an image depicts a lung with a disease or a healthy lung. In the context of multi-classification, the proposed model demonstrates a comprehensive accuracy rate of 95.93% when categorizing various lung disease classes. The model that has been proposed offers a robust and precise methodology for the detection and classification of lung diseases. The aforementioned findings make a valuable contribution to the progress of respiratory healthcare by enhancing the ability to promptly identify and precisely categorize lung diseases, ultimately resulting in enhanced patient outcomes. Future investigations could delve into exploring alternative architectures and fine-tuning hyperparameters to augment the performance of the model. Furthermore, expanding the dataset and integrating a wider range of lung disease categories could yield a more comprehensive comprehension of the model’s capabilities and its potential clinical applications.