Deep Learning Algorithm Trained on Oblique Cervical Radiographs to Predict Outcomes of Transforaminal Epidural...

Area of Science:

• Radiology
• Artificial Intelligence
• Pain Management

Background:

• Cervical radicular pain due to cervical foraminal stenosis is a common condition.
• Transforaminal epidural steroid injection (TFESI) is a frequently used treatment.
• Predicting TFESI treatment outcomes remains challenging.

Purpose of the Study:

• To develop and evaluate a convolutional neural network (CNN) model for predicting TFESI treatment outcomes.
• To assess the utility of cervical radiographs in predicting treatment success for cervical radicular pain.

Main Methods:

• Retrospective study of 293 patients undergoing cervical TFESI for cervical radicular pain.
• Input data: Square-shaped cervical radiographs including the target foramen and surrounding structures.
• Outcome definition: Favorable outcome defined as ≥50% Numeric Rating Scale (NRS) score reduction at 2 months post-injection.

Main Results:

• The developed CNN model achieved an area under the curve (AUC) of 0.823 for predicting treatment outcomes.
• The model demonstrated significant predictive capability for TFESI success.

Conclusions:

• A CNN model trained on oblique cervical radiographs can effectively predict treatment outcomes for cervical TFESI.
• This deep learning approach shows promise for clinical application in managing cervical radicular pain.
• Further improvements in predictive accuracy could enhance widespread clinical adoption.