R-fMRI reconstruction from k–t undersampled data using a subject-invariant dictionary model and VB-EM with nested minorization

•Faster R-fMRI imaging allows higher spatial resolution and more reliable functional connectivity analysis.•We propose a novel undersampling scheme in k-space and time (k-t) both to provide the necessary speedup.•We propose a novel dictionary-based model on the signal, which is robust, spatially regularized, and subject invariant by leveraging an equivalence-class structure on the dictionary.•We propose a novel Bayesian inference framework based on variational Bayesian expectation maximization with nested minorization (VB-EM-NM), which allows us to estimate per-voxel uncertainty in the reconstruction.•Empirical evaluation of (i) R-fMRI reconstructions from simulated data and (ii) functional-network estimates from reconstructions of brain R-fMRI demonstrate that our framework improves over the state of the art, and, additionally, enables significantly higher spatial resolution. [Display omitted] Higher spatial resolution in resting-state functional magnetic resonance imaging (R-fMRI) can give reliable information about the functional networks in the cerebral cortex. Typical methods can achieve higher spatial or temporal resolution by speeding up scans using either (i) complex pulse-sequence designs or (ii) k-space undersampling coupled with priors on the signal. We propose to undersample the R-fMRI acquisition in k-space and time to speedup scans in order to improve spatial resolution. We propose a novel model-based R-fMRI reconstruction framework using a robust, subject-invariant, spatially regularized dictionary prior on the signal. Furthermore, we propose a novel inference framework based on variational Bayesian expectation maximization with nested minorization (VB-EM-NM). Our inference framework allows us to provide an estimate of uncertainty of the reconstruction, unlike typical reconstruction methods. Empirical evaluation of (i) simulated R-fMRI reconstruction and (ii) functional-network estimates from brain R-fMRI reconstructions demonstrate that our framework improves over the state of the art, and, additionally, enables significantly higher spatial resolution.