Improved classification of Alzheimer's disease and mild cognitive impairment through dynamic functional network analysis
New method analyzes 315 brain scans, revealing distinct temporal patterns in Alzheimer's patients.
A research team led by Nicolas Rubido, Venia Batziou, and Vesna Vuksanovic has published a new study demonstrating improved classification of Alzheimer's Disease (AD) and Mild Cognitive Impairment (MCI) through dynamic functional network analysis of fMRI data. The study analyzed brain scans from 315 age- and sex-matched individuals from the ADNI-3 protocol, comparing Alzheimer's patients, those with MCI, and cognitively normal healthy elderly participants. Unlike traditional static network analysis, their approach used sliding-window correlations to capture how brain connections change over time, revealing patterns invisible to conventional methods.
While healthy elderly and MCI groups showed similar static and dynamic networks, significant differences emerged when comparing them to Alzheimer's participants. The analysis identified stable (stationary) differences in functional connections between white matter regions and the parietal lobe/somatosensory cortices. More importantly, it revealed metastable (temporal) network differences consistently found between the amygdala and hippocampal formation—brain regions crucial for memory and emotion. The researchers employed the Juelich brain atlas for network nodes and used non-parametric statistics to assess between-group differences at both the link and node centrality levels.
The findings highlight that dynamic network analysis provides unique insights beyond what static analysis can offer. The team's node centrality analysis further showed that white matter connectivity patterns are local in nature. This research represents a significant methodological advancement, demonstrating that including temporal information in brain network analysis can improve our understanding of neurodegenerative progression along the Alzheimer's spectrum. The approach could lead to more accurate diagnostic tools and better tracking of disease progression over time.
- Analyzed fMRI data from 315 individuals using dynamic (time-varying) network analysis with sliding-window correlations
- Found stable differences in white matter-parietal connections and temporal differences in amygdala-hippocampal networks in Alzheimer's patients
- Demonstrated that dynamic analysis provides classification improvements over static methods for Alzheimer's and Mild Cognitive Impairment
Why It Matters
Could lead to earlier, more accurate Alzheimer's diagnosis by detecting subtle temporal brain network changes before cognitive symptoms appear.