Classification of pulmonary diseases from chest radiographs using deep transfer learning

Area of Science:

• Medical Imaging
• Artificial Intelligence in Healthcare
• Pulmonology

Background:

• Pulmonary diseases are a major global cause of disability and death.
• Early diagnosis of lung conditions is crucial for reducing mortality rates.
• Diagnosing pulmonary diseases from chest radiographs is challenging due to complex anatomy and image quality variations.

Purpose of the Study:

• To explore the effectiveness of Convolutional Neural Networks (CNNs) and transfer learning for diagnosing pulmonary diseases.
• To develop and evaluate a deep transfer learning model for improved predictive outcomes in chest radiograph analysis.
• To enhance the automated detection and classification of fifteen different pulmonary diseases.

Main Methods:

• Utilized deep transfer learning techniques combined with Convolutional Neural Networks (CNNs).
• Applied the model to chest radiographs for the analysis of fifteen distinct pulmonary diseases.
• Compared the proposed model's performance against existing state-of-the-art methods.

Main Results:

• The deep transfer learning model achieved high performance metrics.
• Reported an overall specificity of 97.92%, sensitivity of 97.30%, and precision of 97.94%.
• Achieved an Area Under the Curve (AUC) of 97.61%, indicating strong diagnostic capability.

Conclusions:

• The proposed deep transfer learning model shows significant promise for pulmonary disease diagnosis.
• This AI-driven approach can serve as a valuable tool for practitioners in clinical decision-making.
• Automated analysis of chest radiographs using CNNs can lead to more efficient and accurate diagnosis of lung conditions.