MCADS: Simultaneous Detection and Analysis of 18 Chest Radiographic Abnormalities Using Multi-Label Deep Learning

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Radiology

Background:

• Chest radiography is vital but time-consuming, facing radiologist shortages.
• Delayed interpretation of chest X-rays can lead to treatment delays.

Purpose of the Study:

• Introduce the Multi-label Chest Abnormality Detection System (MCADS).
• Develop a deep learning platform for automated detection of 18 radiographic abnormalities.
• Address diagnostic challenges in chest X-ray interpretation.

Main Methods:

• Utilized a pre-trained DenseNet121 convolutional neural network (via TorchXRayVision).
• Implemented asynchronous image processing on a central server.
• Employed Gradient-weighted Class Activation Mapping (Grad-CAM) for explainability.
• Evaluated on eight large, public datasets.

Main Results:

• Achieved high area-under-the-curve performance across all 18 conditions.
• Provided accurate, multi-condition analyses in under 30 seconds per image.
• Demonstrated reliability and speed suitable for clinical settings.

Conclusions:

• MCADS accelerates chest X-ray interpretation with fast, reliable, and explainable screening.
• Potential to reduce radiologist workload and diagnostic delays.
• Offers a pathway to improve patient care in data-driven healthcare.