Prognosis of cervical myelopathy based on diffusion tensor imaging with artificial intelligence methods.

Area of Science:

• Neuroimaging
• Medical Artificial Intelligence
• Spinal Cord Disorders

Background:

• Diffusion Tensor Imaging (DTI) is a promising tool for predicting cervical myelopathy (CM) outcomes.
• Manual analysis of DTI features for CM prognosis is complex and time-intensive.

Purpose of the Study:

• To evaluate the efficacy of artificial intelligence (AI) methods in analyzing DTI for CM prognosis.
• To compare the performance of conventional AI models and deep learning for DTI analysis in CM.

Main Methods:

• Seventy-five CM patients undergoing surgery were included for DTI imaging.
• DTI features (fractional anisotropy, diffusivities) were extracted; conventional AI (LR, KNN, RBF-SVM) and deep learning models were applied.
• Performance was assessed using 50 repeated 10-fold cross-validations.

Main Results:

• The RBF-SVM model achieved the highest classification accuracy (89.7% ± 1.6%).
• RBF-SVM demonstrated excellent sensitivity (85.0% ± 3.4%) and specificity (92.4% ± 1.9%).
• Conventional AI models outperformed the deep learning model in this DTI analysis task.

Conclusions:

• AI methods are feasible and effective for DTI-based prognosis of cervical myelopathy.
• Machine learning, specifically RBF-SVM, offers a robust approach for automated DTI analysis in CM.
• AI can significantly improve the efficiency and accuracy of DTI interpretation for CM patient management.