A Deep Convolutional Neural Network Model for Lung Disease Detection Using Chest X-Ray Imaging

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Pulmonology

Background:

• Lung diseases like pneumonia and COVID-19 pose significant global health challenges.
• Early and accurate diagnosis is crucial for effective patient management and treatment.
• Medical imaging, particularly chest radiographs, is a cornerstone for lung disease detection due to accessibility and speed.

Purpose of the Study:

• To develop an automated system for detecting multiple lung diseases, including pneumonia and COVID-19, in medical scans.
• To leverage a customized convolutional neural network (CNN) integrated with pretrained models and image enhancement for improved diagnostic accuracy.
• To evaluate the system's performance on a diverse dataset of chest x-ray and CT images.

Main Methods:

• Utilized a dataset of 6400 chest x-ray and CT images categorized into pneumonia, COVID-19, and normal classes.
• Employed data augmentation techniques to address class imbalance within the dataset.
• Developed a deep learning model incorporating a customized CNN, pretrained models, image enhancement, preprocessing, and classification stages.

Main Results:

• The automated system achieved high performance metrics: 96% precision, 95.33% recall, 95.66% F1-score, and 97.24% accuracy.
• Demonstrated superior effectiveness compared to other existing deep learning models for lung disease detection.
• The integrated approach of CNN, pretrained models, and image enhancement proved highly successful.

Conclusions:

• The proposed automated system shows significant promise for the accurate and efficient detection of multiple lung diseases.
• This AI-driven approach can enhance early diagnosis, potentially improving patient outcomes and disease management.
• The study highlights the potential of customized CNNs and image enhancement in medical diagnostics.