An Efficient Combination among sMRI, CSF, Cognitive Score, and APOE ε4 Biomarkers for Classification of AD and MCI
1Department. of Information and Communication Engineering, Chosun University, 309 Pilmun-Daero, Dong-Gu, Gwangju 61452, Republic of Korea.
Computational Intelligence and Neuroscience
|June 23, 2020
Summary
This study developed a new algorithm combining brain imaging, cerebrospinal fluid, and genetic data to accurately diagnose Alzheimer's disease (AD) and mild cognitive impairment (MCI). The method achieved high accuracy in distinguishing between different stages of cognitive decline.
Area of Science:
- Neuroscience
- Medical Imaging
- Biomarker Discovery
Background:
- Alzheimer's disease (AD) is a leading cause of dementia, necessitating early and accurate diagnosis of AD and mild cognitive impairment (MCI).
- Current diagnostic approaches often struggle with the dynamic nature of AD and the unclear relationships between various indicators over time.
- Multimodal analyses for AD/MCI diagnosis are increasing, but optimal combinations of biomarkers, especially using anatomical MRI measures, remain underexplored.
Purpose of the Study:
- To investigate differences in brain atrophy patterns between individuals with AD, MCI, and healthy controls (HCs).
- To develop and evaluate a novel algorithm for classifying AD and MCI using a combination of multiple biomarkers.
- To explore the full potential of anatomical MRI measures in conjunction with other biomarkers for improved AD detection.
Main Methods:
- Utilized structural MRI (sMRI) for cortical thickness, surface area, and gray matter volume; cerebrospinal fluid (CSF) for protein quantification; cognitive scores; and APOE ε4 allele status.
- Employed filter- and wrapper-based feature selection methods.
- Applied an extreme learning machine (ELM) based approach with 10-fold cross-validation for classification, comparing performance against Support Vector Machine (SVM) classifiers using ADNI datasets.
Main Results:
- The proposed algorithm achieved high classification accuracies: 97.31% (AD vs. HC), 91.72% (MCI vs. HC), 87.91% (MCI vs. AD), and 83.38% (MCIs vs. MCIc).
- Combining multiple biomarkers, evaluated using receiver operating characteristic (ROC) curves and area under the curve (AUC), significantly improved classification performance.
- The algorithm effectively classified the challenging MCIs vs. MCIc task, demonstrating its potential for clinical application.
Conclusions:
- A combination of specific biomarkers, processed through the proposed feature selection and ELM algorithm, demonstrates high efficacy in classifying AD and MCI stages.
- The developed method offers a promising approach to increase the accuracy of AD classification in clinical practice.
- This multimodal biomarker approach, particularly leveraging anatomical MRI, expands the possibilities for early and accurate detection of neurodegenerative conditions.
