A Study on Predicting the Efficacy of Posterior Lumbar Interbody Fusion Surgery Using a Deep Learning Radiomics Model

Area of Science:

• Spine surgery
• Medical imaging
• Machine learning

Background:

• Posterior lumbar interbody fusion (PLIF) is a common surgical procedure for degenerative lumbar diseases.
• Predicting the efficacy of PLIF surgery is crucial for patient outcomes.
• Current prediction methods may not fully integrate diverse data sources.

Purpose of the Study:

• To develop and evaluate a combined model for predicting PLIF surgery efficacy.
• To integrate clinical data, radiomics features, and deep learning (DL) models.
• To assess the performance of the combined model against individual models.

Main Methods:

• Retrospective review of 461 patients undergoing PLIF for degenerative lumbar diseases.
• Development of separate clinical, radiomics, and DL models.
• Integration of these models into a combined predictive model.
• Feature selection using the least absolute shrinkage and selection operator (LASSO) method.
• Performance evaluation using receiver operating characteristic (ROC) curves and area under the ROC curve (AUC).

Main Results:

• Patient age, body weight, and preoperative intervertebral distance were identified as risk factors.
• The radiomics model with an expanded mask achieved an AUC of 0.749 on the test set.
• The DL model showed strong performance with an AUC of 0.803 on the test set.
• The combined model achieved the highest AUC of 0.866 on the test set, outperforming individual models.

Conclusions:

• The combined clinical, radiomics, and deep learning model effectively predicts postoperative efficacy of PLIF surgery.
• This integrated approach demonstrates good clinical applicability for improving surgical outcome prediction.
• The model offers a promising tool for personalized treatment strategies in lumbar fusion surgery.