Research on chest radiography recognition model based on deep learning.

Area of Science:

• Medical Imaging Analysis
• Artificial Intelligence in Healthcare
• Radiology Reporting

Background:

• Increasing demand for automated chest X-ray analysis due to medical informatization and global epidemics.
• Existing deep learning methods for medical image reporting lack reliability due to image complexity.
• Need for automated systems that generate accurate and doctor-like radiological reports.

Purpose of the Study:

• To propose an improved Recurrent Learning Network (RLN) model for automated chest X-ray report generation.
• To enhance the reliability and accuracy of AI-generated medical reports.
• To develop a system capable of producing sentence-by-sentence impressions and findings in a standard medical tone.

Main Methods:

• Development of an improved Recurrent Learning Network (RLN) model.
• Integration of Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) networks.
• Incorporation of a multi-head attention mechanism to refine report generation.

Main Results:

• The proposed RLN model effectively generates high-level impressions and detailed descriptive findings.
• The model successfully imitates a doctor's standard tone in generated reports.
• Experimental validation on the Open-i dataset demonstrated the algorithm's effectiveness in generating colloquial medical reports.

Conclusions:

• The improved RLN model offers a reliable solution for automated chest X-ray report generation.
• The model's ability to produce detailed, accurate, and tonally appropriate reports addresses current limitations.
• This advancement contributes to efficient and dependable medical reporting in radiology.