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

Abstract

Diffusion tensor imaging (DTI) has been proposed for the prognosis of cervical myelopathy (CM), but the manual analysis of DTI features is complicated and time consuming. This study evaluated the potential of artificial intelligence (AI) methods in the analysis of DTI for the prognosis of CM. Seventy-five patients who underwent surgical treatment for CM were recruited for DTI imaging and were divided into two groups based on their one-year follow-up recovery. The DTI features of fractional anisotropy, axial diffusivity, radial diffusivity, and mean diffusivity were extracted from DTI maps of all cervical levels. Conventional AI models using logistic regression (LR), k-nearest neighbors (KNN), and a radial basis function kernel support vector machine (RBF-SVM) were built using these DTI features. In addition, a deep learning model was applied to the DTI maps. Their performances were compared using 50 repeated 10-fold cross-validations. The accuracy of the classifications reached 74.2% ± 1.6% for LR, 85.6% ± 1.4% for KNN, 89.7% ± 1.6% for RBF-SVM, and 59.2% ± 3.8% for the deep leaning model. The RBF-SVM algorithm achieved the best accuracy, with sensitivity and specificity of 85.0% ± 3.4% and 92.4% ± 1.9% respectively. This finding indicates that AI methods are feasible and effective for DTI analysis for the prognosis of CM. KEYWORDS artificial intelligence (AI), cervical myelopathy (CM), diffusion tensor imaging (DTI), prognosis 1 | INTRODUCTION Cervical myelopathy (CM) is a common spinal cord dysfunction that affects millions of people worldwide. 1 Currently, surgical intervention is considered to be the most immediate way to provide relief from the spinal cord compression and to promote neurological recovery when coupled with active post-operative neural rehabilitation. 2,3 An accurate prognosis for the surgery would provide useful assistance to surgeons, helping them to decide on the most appropriate treatment option, manage patient expectations, and plan postoperative rehabilitation. 4 Several prognosticators of surgical outcomes have already been proposed. 5,6 However, the value of these prognosticators remains controversial, and researchers continue to search for more effective prognostic methods.Diffusion tensor imaging (DTI) has been demonstrated to have prognostic value for CM. [7][8][9] In CM, demyelination is believed to be the main pathological change in the spinal cord, and this demyelination changes the organization of nerve fiber bundles, resulting in changes to DTI-related signals. Based on this principle, researchers have used DTI to study CM. Also, it has been reported that DTI has great potential for distinguishing CM patients from healthy subjects and for predicting the exact cervical levels causing symptoms. [10][11][12] The pathological status of the spinal cord,

Learning-based fully automated prediction of lumbar disc degeneration progression with specified clinical parameters and preliminary validation

“…Machine learning for utilizing longitudinal big data to establish predictive models can be a potential solution [9][10][11]. Convolutional Neural Network (CNN) has achieved a remarkable performance in MRI analysis tasks including pathology classification [12][13][14][15], landmark detection [16,17], and segmentation [18][19][20][21].…”

Cortical gray matter microstructural alterations in patients with type 2 diabetes mellitus

“…We trained four frequently used machine learning models for the classification task, namely KNN, LRC SVM with linear kernel, and SVM with RBF kernel, and the SVM model with linear kernel exhibited the best performance among the train models and the KNN performed worst. Because using a low dimensional feature space in classification tasks can improve the classification efficiency and mitigate the overfitting problem (Jin et al., 2019 ), we used PCA for dimensionality reduction. After PCA transformation, the SVM with linear kernel achieved accuracy and AUC of 74% and 0.83, respectively in distinguishing T2DM patients from HCs.…”

DTI in der Diagnostik der zervikalen Myelopathie

“…Auch künstliche Intelligenz (artificial intelligence, AI) könnte die Anwendbarkeit der DTI zukünftig verbessern. So zeigten sowohl Studien von Jin et al [41] als auch von Wang et al [42], dass Machine Learning-und Deep Learning-Algorithmen bei der Auswertung von DTI-Parametern bei DCM helfen und eine hohe Sensitivität und Spezifität zur Unterscheidung von Myelopathie-Patienten und gesunden Probanden erreichen können.…”