Related Experiment Video
Updated: Oct 4, 2025
Abbiategrasso Brain Bank Protocol for Collecting, Processing and Characterizing Aging Brains
Published on: June 3, 2020
Intelligent Algorithm-Based Quantitative Electroencephalography in Evaluating Cerebral Small Vessel Disease
Hengya Zhu1, Jingjing Qiu1, Xiaoyan Sun1
1Department of Neurology, Huzhou Center Hospital, Affiliated Center Hospital of Huzhou University, No. 1558 Sanhuan North Road, Huzhou, 313000 Zhejiang, China.
Artificial intelligence using Visual Geometry Group-16 (VGG) and quantitative electroencephalography (QEEG) effectively differentiates cognitive impairment in cerebral small vessel disease (CSVD). This AI model shows high accuracy in assessing patients with vascular dementia (VD) and vascular cognitive impairment with no dementia (VCIND).
Area of Science:
- Neurology and Artificial Intelligence
- Medical Imaging and Signal Processing
Background:
- Cerebral small vessel disease (CSVD) frequently leads to cognitive impairment, ranging from mild deficits to vascular dementia (VD).
- Accurate differentiation between vascular cognitive impairment with no dementia (VCIND) and VD is crucial for appropriate patient management.
- Quantitative electroencephalography (QEEG) offers a non-invasive method to assess brain function, but its interpretation can be complex.
Purpose of the Study:
- To evaluate the clinical application value of an artificial intelligence (AI) model, specifically Visual Geometry Group-16 (VGG-16), combined with QEEG for analyzing cognitive impairment in CSVD patients.
- To compare the diagnostic accuracy of a standard VGG-16 model with an improved version (nVGG) in classifying CSVD patients.
- To assess the utility of AI-driven QEEG analysis in distinguishing between VD and VCIND.
Main Methods:
- Seventy-two patients with CSVD and cognitive impairment were recruited and categorized into VD (34 cases) and VCIND (38 cases) groups based on DSM-5 criteria.
- Clinical information, neuropsychological test results, and QEEG data were collected and analyzed using intelligent algorithms over 2 hours.
- A VGG-16 model and a modified VGG (nVGG) model were employed to analyze QEEG patterns for classification accuracy.
Main Results:
- The standard VGG-16 model achieved an accuracy rate of 84.27% with a Kappa value of 0.7, while the improved nVGG model demonstrated higher accuracy at 88.76% with a Kappa value of 0.78.
- AI-driven QEEG analysis revealed statistically significant differences (P < 0.05) in QEEG characteristics between the VCIND and VD groups, including the severity of abnormalities and presence of background changes.
- Specific QEEG findings, such as the number of abnormal waves and background changes, were significantly different between the VCIND and VD groups, aiding in classification.
Conclusions:
- The AI model based on VGG-16 combined with QEEG provides a valuable tool for assessing cognitive impairment in CSVD.
- The improved nVGG algorithm demonstrates enhanced accuracy in differentiating between cognitive impairment levels in CSVD patients.
- AI-powered QEEG analysis shows significant clinical utility in the assessment and differentiation of vascular dementia and milder forms of cognitive impairment in CSVD.
