Adaptive independent vector analysis for multi-subject complex-valued fMRI data

Abstract

3 Highlights:An adaptive IVA algorithm was proposed for multi-subject complex-valued fMRI data.An MGGD-based nonlinear function was exploited to match varying SCV distributions.The MGGD shape parameter was estimated using maximum likelihood estimation.Subspace de-noising, post-IVA phase de-noising, and noncircularity were utilized.Our method detected more contiguous activations than magnitude-only methods. 4 Abstract BackgroundComplex-valued fMRI data can provide additional insights beyond magnitude-only data. However, independent vector analysis (IVA), which has exhibited great potential for group analysis of magnitudeonly fMRI data, has rarely been applied to complex-valued fMRI data. The main challenges in this application include the extremely noisy nature and large variability of the source component vector (SCV) distribution. New MethodTo address these challenges, we propose an adaptive fixed-point IVA algorithm for analyzing multiplesubject complex-valued fMRI data. We exploited a multivariate generalized Gaussian distribution (MGGD)-based nonlinear function to match varying SCV distributions in which the MGGD shape parameter was estimated using maximum likelihood estimation. To achieve our de-noising goal, we updated the MGGD-based nonlinearity in the dominant SCV subspace, and employed a post-IVA denoising strategy based on phase information in the IVA estimates. We also incorporated the pseudocovariance matrix of fMRI data into the algorithm to emphasize the noncircularity of complex-valued fMRI sources. ResultsResults from simulated and experimental fMRI data demonstrated the efficacy of our method. Comparison with Existing Method(s)Our approach exhibited significant improvements over typical complex-valued IVA algorithms, especially during higher noise levels and larger spatial and temporal changes. As expected, the proposed complex-valued IVA algorithm detected more contiguous and reasonable activations than the magnitudeonly method for task-related (393%) and default mode (301%) spatial maps. ConclusionsThe proposed approach is suitable for decomposing multi-subject complex-valued fMRI data, and has great potential for capturing additional subject variability.

Shift-Invariant Canonical Polyadic Decomposition of Complex-Valued Multi-Subject fMRI Data With a Phase Sparsity Constraint

“…More precisely, the proposed algorithm detects 73.9% more contiguous voxels in RPMA+SMA (932 vs. 536) and 746.5% more contiguous voxels in LPMA (838 vs. 99) than ICA-sCPD. In summary, our proposed method extracts 178.7% more contiguous task-related activations than ICA-sCPD, consistent with the previous finding that phase data provides additional brain information [15], [18], [20], [35], [39].…”

“…Our previous studies have verified that the complexvalued fMRI data analysis can detect more contiguous and reasonable activations than magnitude-only fMRI data analysis [15], [18], [20], [35]. By using complex-valued ICA and IVA algorithms, the complex-valued analyses extracted 139% more voxels for ICA and 393% more voxels for IVA than the magnitude-only analyses for the task fMRI data [15], [35].…”

“…In this study, we use complex-valued fMRI data sets from 10 subjects performing a finger-tapping motor task while receiving auditory instructions [15], [35]. All participants signed IRB-approved informed consent at the University of New Mexico.…”

“…The model order of each algorithm was obtained by first estimating the model order for each single subject of fMRI data using the minimum description length (MDL) criterion with a subsampling scheme [40] and then averaging over 10 subjects. The final model order was set to 50, which is consistent with our previous study [35].…”

Model order effects on ICA of resting-state complex-valued fMRI data: Application to schizophrenia

“…Finally, we used across-subject averaging to generate group components. Group ICA and independent vector analysis (Garrity et al, 2007;Calhoun and Adalı, 2012;Gopal et al, 2016;Chen et al, 2017a;Kuang et al, 2017c) Step 6: Denoise = { ( )} ( = 1, ⋯ , ) based on the phase value of each voxel:…”