Artificial Intelligence in Imaging-Based Diagnosis of Rib Fractures: Enhancing Clinical Decision-Making and Psychological Confidence in Sports Injury Management

RESUMO

Moamen Abdelfadil Ismail*, Raja mohsin S alibrahim, Sarah Mohammed R Alqurashi, Ruba Mahmoud Almuallim, Yazeed abdulaziz Alhazzani Gp, Fay Naif Abdullah Alanazi, Maryam Ahmed Mahmoud Balkhair, Ziyad Abdulaziz Alajlan, Ghadi Said Alahmadi, Ruba Dhaifallah Alotaibi, Watan Abdulla Alsahlawi, Faisal Faihan Alotaibi, Wessal Abdullah Alwardi, Mohammed Abdulhadi Faris Alghamdi, Rakan Nahedh H Almutairi

Background: Artificial Intelligence (AI) has emerged as a transformative tool in medical diagnostics, particularly in radiology, where it aids in the accurate and timely detection of skeletal injuries such as rib fractures. Rib fractures, often challenging to diagnose due to their subtle presentation on imaging, are commonly assessed using X-rays and CT scans. While X-rays are widely accessible, their sensitivity is limited, whereas CT scans offer higher resolution but are time-consuming to interpret. AI, leveraging convolutional neural networks (CNNs), presents a promising solution to enhance diagnostic accuracy and efficiency in detecting rib fractures.

Methods: This retrospective diagnostic accuracy study evaluated an AI system's performance in identifying rib fractures using chest X-rays and CT scans from 200 patients, including 100 with both imaging modalities and 100 with CT alone. The AI system, trained on a dataset of over 50,000 annotated images, provided binary classifications, fracture localizations, and confidence scores. Two board-certified radiologists independently reviewed the images, with CT serving as the gold standard. Diagnostic metrics, including sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and overall accuracy, were calculated and compared between AI and radiologist interpretations.

Results: The AI system demonstrated high diagnostic accuracy on CT scans, with a sensitivity of 94.9%, specificity of 91.2%, and overall accuracy of 93.3%. For X-rays, the performance was moderate, with a sensitivity of 76.4% and specificity of 83.3%. Radiologists slightly outperformed AI on X-rays, achieving 81.6% sensitivity and 87.4% specificity. The AI system excelled in detecting displaced fractures (96.7%) and posterior fractures (94.4%) but showed lower sensitivity for non-displaced fractures (87.3%).

Conclusion: AI exhibits strong diagnostic capabilities for rib fracture detection, particularly on CT imaging, where its performance rivals or exceeds that of radiologists. While its accuracy on X-rays is lower, AI remains a valuable supportive tool in clinical workflows. The integration of AI into radiology practices can enhance diagnostic efficiency, reduce missed injuries, and improve patient outcomes, especially in high-volume or resource-limited settings. Further advancements in algorithm robustness and dataset diversity are needed to optimize AI performance across all fracture types and imaging modalities.