Machine Learning-Based Framework for Differential Diagnosis Between Vascular Dementia and Alzheimer's Disease Using...

Area of Science:

• Neuroimaging
• Machine Learning
• Biomarkers

Background:

• Vascular dementia (VaD) and Alzheimer's disease (AD) present similar dementia syndromes.
• Accurate differential diagnosis is crucial for appropriate patient management.
• Current diagnostic methods may not always distinguish between VaD and AD effectively.

Purpose of the Study:

• To evaluate multi-parameter structural MRI features as biomarkers for differentiating VaD from AD.
• To develop and assess machine learning models for computer-aided diagnosis of VaD vs. AD.

Main Methods:

• Recruited 93 patients (58 AD, 35 VaD) with brain MRI scans.
• Extracted multi-parameter volumetric measurements using automated brain tissue segmentation.
• Employed LASSO for feature selection and SVM for classification, comparing various algorithms.

Main Results:

• The SVM with RBF kernel achieved high diagnostic performance.
• Sensitivity: 82.65%, Specificity: 87.17%, Accuracy: 84.35% (AUC = 0.861).
• Demonstrated significant differences in structural MRI biomarkers between VaD and AD.

Conclusions:

• Structural MRI features combined with machine learning show promise for distinguishing VaD from AD.
• The proposed computer-aided diagnosis method can support clinical decision-making.
• This approach offers a potential tool for improving the accuracy of dementia diagnosis.