Automatic Detection and Classification of Rib Fractures on Thoracic CT Using Convolutional Neural Network: Accuracy...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Radiology

Background:

• Rib fractures are common injuries requiring accurate diagnosis.
• Current diagnostic methods can be time-consuming and prone to error.
• Automated analysis of CT images can potentially improve diagnostic accuracy and efficiency.

Purpose of the Study:

• To evaluate a convolutional neural network (CNN) model for automatic detection and classification of rib fractures from CT images.
• To assess the model's ability to generate structured reports.
• To compare the diagnostic performance of the CNN model with that of experienced radiologists.

Main Methods:

• A CNN model was developed and trained on CT images from 1079 patients.
• The model classified fractures into fresh, healing, and old categories.
• Performance was evaluated using precision, recall, and F1-score, and compared against radiologists' efficiency.

Main Results:

• The CNN model demonstrated high detection efficiency for fresh and healing fractures (F1-scores 0.849 and 0.856) compared to old fractures (F1-score 0.770).
• The model showed robustness across multiple validation sets with varying image parameters.
• AI-assisted diagnosis significantly improved radiologists' precision (80.3% to 91.1%) and sensitivity (62.4% to 86.3%), reducing diagnosis time by an average of 73.9 seconds.

Conclusions:

• The developed CNN model effectively detects and classifies rib fractures from CT images.
• The model assists radiologists in enhancing diagnostic efficiency and reducing workload.
• This AI tool shows promise for improving the interpretation of rib fractures in clinical practice.