Deep learning applied to automatic disease detection using chest X-rays.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Radiology

Background:

• Chest X-rays (CXRs) are crucial for diagnosing pathologies globally.
• A shortage of trained radiologists necessitates advanced diagnostic tools.
• Deep learning (DL) offers potential for automated CXR analysis.

Purpose of the Study:

• To review the application of deep learning (DL) in analyzing chest X-ray (CXR) images.
• To discuss DL models for detecting common CXR abnormalities.
• To explore the performance and challenges of DL in radiology.

Main Methods:

• Introduction to fundamental DL concepts for CXR analysis.
• Explanation of deep neural network (DNN) structures.
• Discussion of transfer learning and data augmentation techniques.
• Review of recent literature on DL for CXR abnormality detection.

Main Results:

• DL models are increasingly used for detecting lung nodules, pneumonia, tuberculosis, and pneumothorax.
• Multi-class classification approaches are employed for diagnosing multiple diseases simultaneously.
• Performance of DL models is compared against human expert interpretation.

Conclusions:

• DL demonstrates significant potential for enhancing CXR interpretation and disease detection.
• Challenges remain in implementing DL models and integrating them with radiologists.
• Future research should focus on refining DL algorithms and their clinical utility.