Automated Computer-Assisted Diagnosis of Pleural Effusion in Chest X-Rays via Deep Learning

Area of Science:

• Medical Imaging
• Artificial Intelligence in Healthcare
• Pulmonary Medicine

Background:

• Pleural effusion is a common pulmonary condition that can lead to severe complications if untreated.
• Manual interpretation of chest X-ray (CXR) images for pleural effusion detection is time-consuming.
• Automated detection systems are needed to improve efficiency and accuracy in diagnosing pleural effusion.

Purpose of the Study:

• To develop and evaluate an automated system for detecting pleural effusion in CXR images.
• To enhance the efficiency and accuracy of pleural effusion diagnosis using artificial intelligence.

Main Methods:

• Integration of image cropping, image enhancement, and the EfficientNet-B0 deep learning model.
• Image cropping focused on the region from the heart to the costophrenic angle.
• Image enhancement techniques were applied to emphasize pleural effusion features for improved model learning.

Main Results:

• The proposed image enhancement improved model recognition accuracy by approximately 4.33%.
• The automated system achieved a final accuracy of 93.27% for pleural effusion detection.
• This represents a significant improvement of 21.30% compared to previous study results (77.00%).

Conclusions:

• The developed system serves as an effective assistive diagnostic tool for physicians.
• It provides standardized detection results, reduces manual interpretation workload, and enhances pulmonary care efficiency.
• The system demonstrates significant advancements in automated pleural effusion detection.