Bayesian longitudinal tensor response regression for modeling neuroplasticity.

A major interest in longitudinal neuroimaging studies involves investigating voxel-level neuroplasticity due to treatment and other factors across visits. However, traditional voxel-wise methods are beset with several pitfalls, which can compromise the accuracy of these approaches. We propose a novel Bayesian tensor response regression approach for longitudinal imaging data, which pools information across spatially distributed voxels to infer significant changes while adjusting for covariates. The proposed method, which is implemented using Markov chain Monte Carlo (MCMC) sampling, utilizes low-rank decomposition to reduce dimensionality and preserve spatial configurations of voxels when estimating coefficients. It also enables feature selection via joint credible regions which respect the shape of the posterior distributions for more accurate inference. In addition to group level inferences, the method is able to infer individual-level neuroplasticity, allowing for examination of personalized disease or recovery trajectories. The advantages of the proposed approach in terms of prediction and feature selection over voxel-wise regression are highlighted via extensive simulation studies. Subsequently, we apply the approach to a longitudinal Aphasia dataset consisting of task functional MRI images from a group of subjects who were administered either a control intervention or intention treatment at baseline and were followed up over subsequent visits. Our analysis revealed that while the control therapy showed long-term increases in brain activity, the intention treatment produced predominantly short-term changes, both of which were concentrated in distinct localized regions. In contrast, the voxel-wise regression failed to detect any significant neuroplasticity after multiplicity adjustments, which is biologically implausible and implies lack of power.

A method to mitigate spatio-temporally varying task-correlated motion artifacts from overt-speech fMRI paradigms in aphasia.

Quantifying accurate functional magnetic resonance imaging (fMRI) activation maps can be dampened by spatio-temporally varying task-correlated motion (TCM) artifacts in certain task paradigms (e.g., overt speech). Such real-world tasks are relevant to characterize longitudinal brain reorganization poststroke, and removal of TCM artifacts is vital for improved clinical interpretation and translation. In this study, we developed a novel independent component analysis (ICA)-based approach to denoise spatio-temporally varying TCM artifacts in 14 persons with aphasia who participated in an overt language fMRI paradigm. We compared the new methodology with other existing approaches such as "standard" volume registration, nonselective motion correction ICA packages (i.e., AROMA), and combining the novel approach with AROMA. Results show that the proposed methodology outperforms other approaches in removing TCM-related false positive activity (i.e., improved detectability power) with high spatial specificity. The proposed method was also effective in maintaining a balance between removal of TCM-related trial-by-trial variability and signal retention. Finally, we show that the TCM artifact is related to clinical metrics, such as speech fluency and aphasia severity, and the implication of TCM denoising on such relationship is also discussed. Overall, our work suggests that routine bulkhead motion based denoising packages cannot effectively account for spatio-temporally varying TCM. Further, the proposed TCM denoising approach requires a one-time front-end effort to hand label and train the classifiers that can be cost-effectively utilized to denoise large clinical data sets.

The effect of time since stroke, gender, age, and lesion size on thalamus volume in chronic stroke: a pilot study.

Recent stroke studies have shown that the ipsi-lesional thalamus longitudinally and significantly decreases after stroke in the acute and subacute stages. However, additional considerations in the chronic stages of stroke require exploration including time since stroke, gender, intracortical volume, aging, and lesion volume to better characterize thalamic differences after cortical infarct. This cross-sectional retrospective study quantified the ipsilesional and contralesional thalamus volume from 69 chronic stroke subjects' anatomical MRI data (age 35-92) and related the thalamus volume to time since stroke, gender, intracortical volume, age, and lesion volume. The ipsi-lesional thalamus volume was significantly smaller than the contra-lesional thalamus volume (t(68) = 13.89, p < 0.0001). In the ipsilesional thalamus, significant effect for intracortical volume (t(68) = 2.76, p = 0.008), age (t(68) = 2.47, p = 0.02), lesion volume (t(68) = - 3.54, p = 0.0008), and age*time since stroke (t(68) = 2.46, p = 0.02) were identified. In the contralesional thalamus, significant effect for intracortical volume (t(68) = 3.2, p = 0.002) and age (t = - 3.17, p = 0.002) were identified. Clinical factors age and intracortical volume influence both ipsi- and contralesional thalamus volume and lesion volume influences the ipsilesional thalamus. Due to the cross-sectional nature of this study, additional research is warranted to understand differences in the neural circuitry and subsequent influence on volumetrics after stroke.