Improving Patient Safety in the X-ray Inspection Process with EfficientNet-Based Medical Assistance System.

Area of Science:

• Medical Imaging Analysis
• Artificial Intelligence in Healthcare
• Patient Safety Protocols

Background:

• Patient safety is critical in healthcare, with Artificial Intelligence (AI) offering potential improvements.
• Existing medical imaging processes, particularly X-ray examinations, require enhanced safety measures.
• Deep learning advancements provide opportunities to refine AI applications in clinical settings.

Purpose of the Study:

• To improve the medical operation process during X-ray examinations for enhanced patient safety.
• To develop and evaluate novel neural network architectures for pre-X-ray image classification.
• To ensure consistency between AI-driven classification and physician orders, minimizing discrepancies.

Main Methods:

• Utilized EfficientNet for classifying 49 categories of pre-X-ray images.
• Introduced two novel neural network architectures to further improve classification accuracy.
• Compared classification results against doctor's orders for validation.
• Compiled a dataset of over 12,000 pre-X-ray images across 49 categories from Taichung Veterans General Hospital.

Main Results:

• Achieved a significant improvement in classification accuracy for pre-X-ray images.
• Demonstrated an accuracy enhancement exceeding 4% compared to previous studies.
• Validated the effectiveness of the proposed novel neural network models.

Conclusions:

• The developed AI models significantly enhance the accuracy of pre-X-ray image classification.
• The proposed methods contribute to improved patient safety by increasing consistency in medical operations.
• This research highlights the potential of advanced deep learning techniques in optimizing healthcare diagnostics.