Improved classification of Alzheimer’s disease and mild cognitive impairment through dynamic functional network analysis

Nicolas RubidoAberdeen Biomedical Imaging CentreUniversity of Aberdeen, Venia BatziouHealth Data ScienceSwansea University Medical SchoolSwansea University, Marwan FuadHealth Data ScienceSwansea University Medical SchoolSwansea University, Vesna Vuksanović Health Data ScienceSwansea University Medic · 2025

Read Improved classification of Alzheimer’s disease and mild cognitive impairment through dynamic functional network analysis by Nicolas RubidoAberdeen Biomedical Imaging CentreUniversity of Aberdeen, Venia BatziouHealth Data ScienceSwansea University Medical SchoolSwansea University, Marwan FuadHealth Data ScienceSwansea University Medical SchoolSwansea University, Vesna Vuksanović Health Data ScienceSwansea University Medic (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2505-03458.txt.

Introduction

Nicolas Rubido and co-authors sort Alzheimer disease from healthy aging using moving brain networks, in a 2025 paper. Brain scans can map which regions act together over time. Most studies freeze that map into one still picture. This team also built time-varying maps, which catch networks as they shift.

Data came from a large Alzheimer cohort with patients, mildly impaired people, and healthy controls, matched by age and sex. Still and moving maps were built from resting scans using a standard brain atlas. Healthy and mildly impaired people looked alike at the region level in both views. Alzheimer patients stood apart more clearly. Some links differed steadily, while others flickered near the amygdala and related areas.

Catalog record

CREATORNicolas RubidoAberdeen Biomedical Imaging CentreUniversity of Aberdeen, Venia BatziouHealth Data ScienceSwansea University Medical SchoolSwansea University, Marwan FuadHealth Data ScienceSwansea University Medical SchoolSwansea University, Vesna Vuksanović Health Data ScienceSwansea University Medic
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE48 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2505-03458

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Improved classification of Alzheimer’s disease and mild cognitive impairment through dynamic functional network analysis Nicolas RubidoAberdeen Biomedical Imaging CentreUniversity of Aberdeen, Venia BatziouHealth Data ScienceSwansea University Medical SchoolSwansea University, Marwan FuadHealth Data ScienceSwansea University Medical SchoolSwansea University, Vesna Vuksanović Health Data ScienceSwansea University Medical SchoolSwansea University Abstract Brain network analysis using functional MRI has advanced our understanding of cortical activity and its disruption in neurodegenerative disorders underlying dementia. Recently, research has focused on dynamic (time-varying) brain networks that capture both spatial and temporal patterns of regional cortical co-activity. However, this approach remains relatively unexplored across the Alzheimer’s disease (AD) spectrum. In this study, we analysed age- and sex-matched static and dynamic functional brain networks derived from resting-state fMRI data in 315 individuals with AD, mild cognitive impairment (MCI), and cognitively normal healthy controls (HC) from the ADNI-3 cohort. Functional networks were constructed using the Juelich brain atlas, with static connectivity estimated from full time series and dynamic connectivity derived using a sliding-window approach. Group differences were assessed at both the link and node levels using non-parametric statistics and bootstrap resampling. While HC and MCI show similar static and dynamic patterns at the node level, clearer differences emerge in AD. We identified stable (stationary) differences in functional connectivity between white matter regions and parietal and somatosensory cortices, whereas temporally varying differences were consistently observed in connections involving the amygdala and hippocampal formation. In addition, node centrality analysis suggested that white matter connectivity differences are predominantly local in nature. Our results highlight shared and unique functional connectivity patterns in both static and dynamic functional networks, emphasising the importance of incorporating dynamic information into brain network analyses of the Alzheimer’s spectrum. Furthermore, we trained a Random Forest model on regional BOLD time series informed by dynamic and static network metrics, achieving robust classification of MCI, AD, and HC groups and demonstrating the diagnostic potential of time-varying connectivity. A Preprint Introduction Brain networks derived from functional magnetic resonance imaging (fMRI) have improved our understanding of cortical functional connectivity, i.e., patterns of co-activity between remote cortical processes Sporns (2010) ; Fornito et al. (2016) ; Bijsterbosch and Beckmann (2017) . Mapping functional networks from neuroimaging data, has been useful in providing information about altered cortical activity in brain disorders Mattson (2000) ; Zecca et al. (2004) ; Pievani et al. (2011) ; Stam (2014) . In Alzheimer’s disease (AD), functional networks differ from the patterns of connectivity observed in healthy individuals Eguiluz et al. (2005) ; Damoiseaux et al. (2006) ; Honey et al. (2009) ; Bullmore and Sporns (2009) ; Bullmore and Sporns (2012) , and the magnitude of these differences often correlates with the underlying abnormal accumulation of misfolded tau protein and deposits of amyloid-beta, or white matter damage Borne et al. (2024) . Differences also correlate with the severity of clinical symptoms, assessed using cognitive and behavioral metrics Stam et al. (2007) ; Binnewijzend et al. (2014) ; Luo et al. (2017) ; Dai et al. (2019) ; Lei et al. (2019) . Network-based studies further demonstrate that while brain disorders like AD exert a global effect on the brain, their impact is heterogeneous, with certain regions, particularly the central (hub) areas of the temporal, parietal, and frontal regions associated with higher-order cognitive functions, being more severely affected He et al. (2008) ; Stam et al. (2009) . Alzheimer’s disease is increasingly recognised as a disconnection syndrome, disrupting brain networks and leading to cognitive and behavioral decline Delbeuck et al. (2003) . Network changes are observed in Mild Cognitive Impairment (MCI) Zhang et al. (2011) ; Gilligan et al. (2019) , a cognitive disorder that may progress to AD. However, the impact of MCI and AD is not always proportional to their cognitive severity. Both MCI and AD affect local de Haan et al. (2012) ; Heringa et al. (2014) and small-world network metrics Tijms et al. (2013) ; Seo et al. (2013) through changes in clustering (i.e., segregation) Brier et al. (2014) and global efficiency (i.e., integration) Lo et al. (2010) ; Yao et al. (2010) ; Reijmer et al. (2013) . Betweenness centrality Liu et al. (2012) ; Wang et al. (2020) , modular organization Chen et al. (2013) ; Contreras et al. (2019) ; Vuksanović et al. (2019) and reduced functional connectivity within the default mode network Greicius et al. (2004) ; Petrella et al. (2011) ; Agosta et al. (2012) ; Dillen et al. (2017) , have also been consistent findings throughout studies. Recently, dynamic brain network analysis, which considers temporal fluctuations in the resting-state fMRI signal, has revealed patterns of activity that are usually averaged out by conventional functional network analysis. These patterns of activity, termed dynamic functional networks (dFNs), reveal transient (metastable) states, whose flexible links (reconfiguration) support cognitive processing Vuksanović and Hövel (2014) ; Alderson et al. (2020) . Analysis of dynamic functional networks from functional brain recordings, i.e., those acquired over a period of time, has shown potential to reveal clinically relevant information Filippi et al. (2019) ; Moguilner et al. (2021) . For example, He et al. (2018) demonstrated decreased flexibility of interactions in peripheral regions of the language network in patients with temporal lobe epilepsy, due to reduced metastability, which was associated with decreased verbal fluency. Recent studies on dFC in schizophrenia have shown different dwell times in two states, particularly along FC within the default mode (DMN) and language networks Weber et al. (2020) ; Rabany et al. (2019) . These findings highlight the importance of integrating temporal information into brain network analysis. However, in addition to a few recent studies Rubido et al. (2024) ; Córdova-Palomera et al. (2017) ; Quevenco et al. (2017) ; Núñez et al. (2021) ; Sendi et al. (2021) , this approach remains largely unexplored in the context of functional network disruptions in the Alzheimer’s spectrum. In addition, comparative analyses of static and dynamic FNs in AD are lacking. In this study, our aim was to investigate the potential of fMRI-derived cortical functional networks to assess disrupted connectivity in AD. Using comparative analysis of dynamic and static functional networks, we studied whether the temporal characteristics of functional connectivity are more effective than static ones in identifying reliable functional networks’ features in individuals with AD compared to normal ageing and MCI. In addition, we examined sensitivity and specificity of these features to identify AD and MCI individuals from healthy elderly. We hypothesised that dFNs in AD behave differently compared to HC and MCI, reflecting disrupted functional connectivity patterns associated with AD. Methods Participants and Cohorts Data used in this study were obtained from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) data base. We downloaded demographic, clinical and MRI data from N=315 individuals participating in the study. The main inclusion criterion was that the fMRI datasets were acquired using the same resting-state protocols (see Sec. Data Analysis for details). The 315 participants were selected from three groups: Healthy Controls (HC), Mild cognitive Impairment (MCI), and Alzheimer’s Disease (AD) groups. Demographic and clinical characteristics of the participant averaged across each group were shown in Table 1 . Table 1: Participants’ demographic and clinical characteristics. Abbreviations: M-male, F-Female, s.d.-standard deviation, HC-healthy controls, AD-Alzheimer’s Disease, MCI-Mild Cognitive Impairment, ⋆ significant difference

Read the complete text