A spatially constrained independent component analysis jointly informed by structural and functional connectivity

Abstract

AbstractThere are a growing number of neuroimaging studies motivating joint structural and functional brain connectivity models. Connectivity features of different modalities provide insight into brain functional organization by leveraging complementary information, especially in brain disorders as schizophrenia. To this end, we proposed a multimodal ICA model to utilize information from both structural and functional brain connectivity as well as guided by spatial maps to estimate intrinsic connectivity networks (ICNs). Structural connectivity is estimated through whole-brain tractography on diffusion-weighted MRI (dMRI), while functional connectivity information is derived from resting-state functional MRI (rs-fMRI). The proposed structural-functional connectivity and spatially constrained ICA (sfCICA) model optimizes ICNs at the individual level using a multi-objective framework. We evaluated our method using synthetic data and a real dataset (including dMRI and rs-fMRI images from 300 schizophrenia patients and controls). The results demonstrated that our method enhances the functional coupling between ICs with higher structural connectivity in both synthetic and real data. Additionally, the resulting component maps showed improved modularity and enhanced network distinction, particularly in the patients group. Statistical analysis on the functional connectivity networks revealed more significant group differences when comparing the estimated structural-functional connectivity and spatially constrained ICNs with single modality ICNs. In summary, compared to an fMRI only method, the proposed joint approach for estimating ICNs showed multiple benefits from being jointly informed by structural and functional connectivity information. These findings suggest advantages in simultaneously learning from structural and functional connectivity information in brain network studies, effectively enhancing connectivity estimates based on structural connectivity.