Application Of Structural-functional Magnetic Resonance Imaging Fusion Analysis In The Study Of Subjective Cognitive Decline

Subjective cognitive decline(SCD)is an initial stage of Alzheimer’s disease(AD)and it happens prior to mild cognitive impairment(MCI).Identifying biomarkers of SCD is of great significance for the early diagnosis and prevention of MCI and AD.Clinically,brain imaging techniques,such as multimodal magnetic resonance imaging(MRI),can help doctors diagnose pre-AD diseases like SCD and MCI.Different modalities of MRI,such as structural MRI(s MRI)and functional MRI(f MRI),can provide complementary information about brain anatomy or functional activity from different aspects.Previous MRI studies of SCD and MCI are mainly focused on single modality MRI or they processed different modalities separately.However,these single modality MRI-based researches ignored the joint covariant information between different modalities.The joint information is important because it can reflect the complex and multivariate connections between brain structure and function in SCD and MCI.In order to fully capture the joint information between multimodal MRI and to find latent biomarkers of SCD and MCI for clinical diagnosis,we established a novel classification framework based on multimodal MRI(s MRI and f MRI)fusion,which can extract interpretable multimodal MRI features.This framework is mainly composed of the following three steps.The first step is to preprocess s MRI and resting-state f MRI(rsf MRI)and to extract gray matter(GM)and amplitude of low frequency fluctuations(ALFF)as structural and functional features.Then,a multimodal canonical correlation analysis-joint independent component analysis(m CCA-j ICA)model was employed to conduct the fusion analysis on the features to determine the GM-ALFF covariant features in SCD and MCI.Third,the features extracted by m CCA-j ICA was input into classifiers for diagnose SCD and MCI.This novel s MRI-f MRI fusion method was applied to a dataset including s MRI and f MRI from 214 subjects(healthy control [HC]: 66,SCD: 55,MCI: 93).Results showed that alterations of GM and ALFF can be found in many brain regions of SCD and MCI patients.Importantly,through the proposed s MRI-f MRI fusion analysis,we obtained structural-functional covariant pattern(middle frontal gyrus/ middle temporal gyrus-lingual)in the alterations between SCD and MCI,and the covariant network(middle temporal gyrus-insula,middle frontal gyrus/ middle occipital gyrus-superior temporal gyrus,and middle frontal gyrus/ middle temporal gyrus-lingual)between HC and MCI.Classification results showed that the classification accuracies based on the above findings uncovered by s MRI-f MRI fusion analysis were significantly higher than the accuracies based on single-modality MRI.This study provides a new methodological support for identifying structuralfunctional neural markers of SCD and MCI.So,this work is an important step towards a better understanding of the neuropathological mechanism of SCD and MCI,and has the potential to provide new neural biomarkers for early diagnosis and treatment of pre-AD diseases.