The analysis of fMRI complexity as a biomarker of cognitive decline compared to tau‐PET

Background Tau-PET detects the neurofibrillary tangle pathology and is considered a biomarker of neuronal injury cognitive decline. However, PET imaging has radiation exposure risks costly. Recently, brain entropy mapping based on rs-fMRI emerged as functionality that indicates complexity BOLD signals. Various studies have implied reduced values correlate significantly with decline in Alzheimer’s disease. This study analyzes effect fMRI AD compared to tau-PET. Method We obtained dataset for this from Disease Neuroimaging Initiative (ADNI). The consisted 147 subjects (age = 72.5±7.5, M/F 60/87, CN/MCI/AD 95/45/7) tau-PET (tracer: 18F-AV1451) two scans. FMRI images underwent standard processing generate Multi-Scale Entropy (MSE) maps. calculated parametric SUVR cerebellar reference regions images. To determine ability MSE risk factor AD, convolutional neural networks were used classify individuals into CN or MCI/AD groups. Each classifier was trained separately These models had same network structures hyperparameters. performed 5-fold cross-validation evaluate each model. Result using fMRI-MSE achieved an accuracy equivalent (0.7823 vs. 0.7687). also Area Under ROC curve (AUC) sensitivity specificity observed model AUC 0.7022, resulted 0.6889. Conclusion classification performance showed use could potentially be alternative Further multisite, larger datasets are required validate our preliminary results provide more conclusive evidence.