PNEUMONIA DETECTION ON CHEST X-RAY USING RADIOMIC FEATURES AND CONTRASTIVE LEARNING

Area of Science:

• Medical Imaging Analysis
• Artificial Intelligence in Radiology

Background:

• Chest X-rays are crucial for noninvasive diagnosis, but manual interpretation leads to radiologist burnout and diagnostic delays.
• Radiomics offers quantitative feature extraction from medical images, aiding diagnosis before the deep learning era.
• Deep learning models for chest X-ray diagnosis lack transparency and explainability.

Purpose of the Study:

• To develop a novel framework for pneumonia detection in chest X-rays.
• To enhance the interpretability of AI models in medical imaging.
• To combine radiomics features with contrastive learning for improved diagnostic performance.

Main Methods:

• A novel framework integrating radiomics features with contrastive learning was proposed.
• The model was trained and evaluated on the RSNA Pneumonia Detection Challenge dataset.
• Performance was compared against several state-of-the-art models.

Main Results:

• The proposed model achieved superior results compared to existing state-of-the-art methods.
• The model demonstrated a significant improvement in F1-score, exceeding 10%.
• Enhanced model interpretability was achieved.

Conclusions:

• The framework effectively detects pneumonia in chest X-rays.
• Combining radiomics and contrastive learning offers a promising approach for interpretable AI in medical diagnostics.
• This method addresses the explainability gap in deep learning for chest X-ray analysis.