Individualized brain network attribution shows modest subject reliability in adni
How Reliable Is Individualized Functional-Network Attribution in Resting-State fMRI Cognitive Prediction in ADNI?
Computer Vision and Pattern RecognitionArtificial Intelligence
Summary
This paper studies how reliable a method is for identifying which brain networks influence predictions about cognition using resting-state brain scans in Alzheimer's patients. The authors found that while the method shows very high reliability within the same scanning session, its reliability decreases when scans are repeated on different days. There is some individual-specific information, but much of the signal reflects patterns shared across the whole population. They suggest reporting reliability across repeated scans and including control tests to avoid overstating individual differences.
What this means in practice
- •For clinical neuroimaging teams: Assess the consistency of brain network markers for individual patients over multiple scan sessions to improve diagnostic confidence.
- •For medical data scientists: Use scan-rescan reliability measures and permutation-based controls when developing personalized biomarkers from resting-state fMRI data.
Tested on one dataset.
Authors
Sheethal Bhat, Mahfuzur Rahman Chowdhury, Imam Ahmed Shaik, Rose Dawn Bharath, Siming Bayer, Andreas Maier
Abstract
Individualized attribution estimates which functional networks drive a model prediction for a given subject and is often proposed as an interpretable resting-state fMRI (rs-fMRI) biomarker, but its test-retest reliability is rarely quantified. We evaluate per-subject network-ablation attribution for a frozen resting-state functional connectivity (FC) predictor (Schaefer-200 FC predicting MMSE/ADAS-Cog13) trained on N=875 subjects from the Alzheimer's Disease Neuroimaging Initiative (ADNI), using, and (iii) a subject-level permutation null. Split-half reliability is near-ceiling (Spearman p=0.948/0.950) but drops on scan-rescan (p=0.609/0.616). Under the mismatched-subject null, correlations remain substantial (p~0.46-0.47), indicating strong population-shared structure, yet matched pairs exceed this null (z~6.6-6.9, p<=10-4). Thus, in ADNI, individualized network attribution contains a modest but non-zero person-specific component atop a strong population-shared pattern. We recommend reporting scan-rescan reliability together with a permutation-null control, rather than within-session proxies alone, to avoid overstating individual specificity.