Classifying Mild Cognitive Impairment from Normal Cognition: fMRI Complexity Matches Tau PET Performance

Area of Science:

• Neuroimaging
• Biomarker Discovery
• Alzheimer's Disease Research

Background:

• Tau protein accumulation is a hallmark of Alzheimer's disease (AD), leading to neuronal loss and cognitive decline.
• Tau-PET imaging is a direct biomarker but is costly, involves radiation, and lacks widespread accessibility.
• Resting-state functional MRI (rs-fMRI) complexity is explored as a non-invasive surrogate for early neuronal dysfunction in AD.

Purpose of the Study:

• To compare the efficacy of fMRI complexity (sample and multiscale entropy) versus tau-PET in classifying cognitively normal (CN) individuals from those with mild cognitive impairment (MCI) or AD.
• To identify and compare critical brain network regions-of-interest (ROIs) for classification using both fMRI complexity and tau-PET, revealing neuroanatomical correlates of AD.
• To establish fMRI complexity as a potential alternative biomarker for early AD detection.

Main Methods:

• A cross-sectional study utilizing data from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database.
• Whole-brain complexity maps derived from rs-fMRI and tau-PET standardized uptake value ratio (SUVR) maps were generated.
• 3D convolutional neural network (CNN) classification with five-fold cross-validation and leave-one-network-out analysis were employed for classification and ROI identification.

Main Results:

• fMRI complexity achieved classification accuracy comparable to tau-PET, with superior F1-score (0.64 vs. 0.61) and area under the curve (AUC; 0.73 vs. 0.67).
• Salience and dorsal attention networks were most influential for fMRI complexity-based classification.
• The dorsal attention network was the primary contributor to tau-PET-based classification.

Conclusions:

• fMRI complexity demonstrates comparable performance to tau-PET in identifying cognitive impairment associated with AD.
• This suggests fMRI complexity as a viable, radiation-free imaging biomarker for earlier detection and broader clinical application in AD.
• Partially distinct critical ROIs identified by each modality highlight complementary neuroanatomical insights into AD-related changes.