Artificial Intelligence Detection of Cervical Spine Fractures Using Convolutional Neural Network Models

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Radiology

Background:

• Cervical spine fractures require accurate and timely diagnosis.
• Radiographic interpretation can be challenging, potentially leading to missed fractures.
• Computer-assisted diagnosis holds promise for improving diagnostic accuracy.

Purpose of the Study:

• To develop and evaluate a convolutional neural network (CNN) model for automated detection of cervical spine fractures.
• To assess the performance of deep learning in classifying cervical spine fractures from radiographic X-ray images.
• To explore the potential of AI in assisting radiologists with cervical spine fracture diagnosis.

Main Methods:

• A classification model was developed using CNNs on 500 lateral cervical spine X-ray images (250 normal, 250 fracture).
• The dataset was divided into 70% for training and 30% for testing.
• Konstanz Information Miner (KNIME) was utilized for data annotation, preprocessing, model training, and evaluation.

Main Results:

• The CNN model demonstrated high sensitivity (0.886 for fractures, 0.957 for normal cases) and specificity (0.957 for fractures, 0.886 for normal cases).
• Precision values were 0.954 for fractures and 0.893 for normal cases, indicating a low false positive rate.
• The overall accuracy reached 92.14%, with strong performance indicated by the area under the receiver operating characteristic curve.

Conclusions:

• Deep learning models can be effectively applied for computer-assisted diagnosis of cervical spine fractures.
• The developed CNN technique shows potential to assist radiologists in screening and diagnosing cervical spine fractures.
• This AI-driven approach may enhance diagnostic efficiency and patient outcomes in radiology.