ORIGINAL RESEARCH article
Front. Aging Neurosci.
Sec. Alzheimer's Disease and Related Dementias
Distinguishing Early from Late Mild Cognitive Impairment: A Multi-level Analysis of Regional Morphometry and KLS-Derived Network Topology
Peng Yan 1
Xinyu Du 1
Siyu Yang 2
1. The First Affiliated Hospital With Nanjing Medical University, Nanjing, China
2. The Second Affiliated Hospital of Nanjing University of Chinese Medicine, Nanjing, China
Select one of your emails
You have multiple emails registered with Frontiers:
Notify me on publication
Please enter your email address:
If you already have an account, please login
You don't have a Frontiers account ? You can register here
Abstract
Introduction: Distinguishing between early Mild Cognitive Impairment (EMCI) and late Mild Cognitive Impairment (LMCI) is crucial for clinical trials, but objective biomarkers are lacking. We therefore examined regional morphometry and network topology across cognitively normal (CN), EMCI, and LMCI groups to address this gap. We also evaluated whether combining these features could effectively classify Mild Cognitive Impairment (MCI) subtypes. Methods: We analyzed T1-weighted magnetic resonance imaging (MRI) data from 208 Alzheimer's Disease Neuroimaging Initiative (ADNI) participants (67 CN, 83 EMCI, 58 LMCI). We used both voxel- and surface-based morphometry to measure local atrophy and combined this with graph analysis of individual structural covariance networks (SCNs). We also performed correlation and machine learning analyses. Results: We found that cortical thickness (CT) in EMCI was not significantly different from CN, but it was significantly reduced in the LMCI group. The right hippocampus and the left thalamus, however, showed a significant difference between CN and EMCI. In the Kullback–Leibler (KL) divergence-based similarity (KLS) network analysis, the EMCI group showed a greater randomization when compared to the LMCI group, while LMCI was accompanied by elevated nodal centrality in the left hippocampus and orbital frontal region. Correlation analysis confirmed this was a maladaptive phenomenon, as higher centrality was linked to poorer cognitive performance. Finally, a classifier combining these structural and network features successfully differentiated the MCI subtypes. Conclusions: Our findings suggest that differences in Gray matter volume (GMV) may be more easily observed in the EMCI group. We identified a corresponding non-linear pattern of network topology, characterized by randomization in the EMCI group than in the LMCI. These multi-faceted biomarkers enabled the accurate machine-learning-based differentiation of MCI subtypes, offering a powerful framework for improving patient stratification in clinical trials.
Summary
Keywords
Individual Structural Covariance Network5, Kullback-Leibler Similarity4, Machine Learning3, mild cognitive impairment1, Structural Magnetic Resonance Imaging2
Received
22 October 2025
Accepted
20 February 2026
Copyright
© 2026 Yan, Du and Yang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
*Correspondence: Siyu Yang
Disclaimer
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.