Combining MEA-Net and LAP-Net for Pneumoconiosis Staging Framework

Area of Science:

• Occupational Medicine
• Radiology
• Artificial Intelligence

Background:

• Pneumoconiosis is a significant occupational hazard.
• Current staging relies on subjective interpretation of lung X-rays by experienced physicians.
• Individual clinical judgment introduces variability in diagnosis.

Purpose of the Study:

• To develop an objective and accurate framework for pneumoconiosis staging.
• To improve the diagnostic process by reducing subjectivity.
• To leverage deep learning for enhanced classification of pneumoconiosis stages.

Main Methods:

• A novel deep learning framework was proposed.
• The framework integrates MEA-Net and LAP-Net architectures.
• The method was evaluated for its effectiveness in pneumoconiosis staging.

Main Results:

• The deep learning framework achieved high performance metrics.
• Accuracy reached 95.24%, precision 95.15%, recall 95.15%, specificity 90.58%, F1-score 94.85%, and AUC 98.87% for four-stage classification.
• These results demonstrate robust performance in classifying pneumoconiosis stages.

Conclusions:

• The proposed deep learning framework offers a more accurate method for pneumoconiosis staging.
• This AI-driven approach can assist physicians in diagnosing pneumoconiosis stages.
• The study highlights the potential of AI in improving occupational disease diagnosis.