Related Experiment Video
Updated: Jun 1, 2025
14:27
Identification of Disease-related Spatial Covariance Patterns using Neuroimaging Data
Published on: June 26, 2013
15.6K
Characterizing brain network alterations in cervical spondylotic myelopathy using static and dynamic functional
Jiyuan Yao1, Bingyong Xie1, Haoyu Ni1
1Department of Orthopedics, The First Affiliated Hospital of Anhui Medical University, Hefei, PR China; Department of Spine Surgery, The First Affiliated Hospital of Anhui Medical University, Hefei, PR China.
Summary
Brain connectivity changes in cervical spondylotic myelopathy (CSM) patients were identified using static and dynamic functional network connectivity (sFNC and dFNC). Machine learning models effectively classified CSM, showing potential as diagnostic biomarkers.
Area of Science:
- Neuroscience
- Medical Imaging
- Computational Biology
Background:
- Cervical spondylotic myelopathy (CSM) causes neurological impairment with poorly understood neural mechanisms.
- Brain network connectivity, including static (sFNC) and dynamic (dFNC) functional connectivity, offers potential insights into CSM pathophysiology.
- Investigating brain-wide connectivity alterations in CSM patients is crucial for understanding disease mechanisms.
Purpose of the Study:
- To investigate brain-wide connectivity alterations in CSM patients using sFNC and dFNC.
- To explore the potential of sFNC and dFNC as biomarkers for CSM classification and progression.
- To apply machine learning approaches to analyze functional connectivity networks (FCNs) in CSM.
Main Methods:
- 191 participants (108 CSM patients, 83 healthy controls) underwent resting-state fMRI.
- Functional connectivity networks (FCNs) were analyzed for sFNC and dFNC features.
- K-means clustering identified dFNC states; machine learning models (SVM, DT, LDA, LR, RF) were trained for classification.
Main Results:
- CSM patients showed altered sFNC, including enhanced connectivity in specific networks and weakened connectivity elsewhere.
- dFNC analysis revealed four distinct states, with CSM patients exhibiting altered connectivity in State 1 and State 3.
- Machine learning models achieved high classification performance: SVM (sFNC) reached AUC 0.92, accuracy 85.86%; dFNC State 3 model achieved AUC 0.91, accuracy 84.97%.
Conclusions:
- Significant sFNC and dFNC alterations in CSM patients suggest underlying neural mechanisms.
- FNC features, particularly with SVM, show strong potential as neuroimaging biomarkers for CSM diagnosis and monitoring.
- Future research should focus on longitudinal studies and multimodal neuroimaging to validate these findings.
