How reliable are our current techniques for processing spinal cord fMRI data? In a recently published paper, Dr. Mark Hoggarth and Max Wang explore how person-to-person variability in the contouring of fMRI spinal cord masks affects downstream analyses. Using masks drawn by 8 raters of varying experience to inform co-registration to standard template space, they identify spatial differences in the accuracy of registration related to underlying image contrast and rater experience. At the group-level, spatial differences were present in activation maps although no systematic effect on overall activation level was identified. This work characterizes how variability in manually contoured masks propagates to results in spinal cord fMRI, demonstrating that standardization of processing pipelines and improving image acquisition should be prioritized.

This work was performed in collaboration with Dr. Kenneth Weber at the Stanford Systems Neuroscience and Pain Lab.

You can find the full access paper here!

 

A recent paper from the lab, led by Dr. Rachael Stickland, identifies a positive correlation between baseline cerebral blood flow (CBF) and Blood Oxygenation Level Dependent cerebrovascular reactivity (BOLD-CVR), measured with MRI. This positive correlation is seen between and within subjects and is strengthened by the addition of a breathing task to a resting-state acquisition and by optimizing for hemodynamic lag effects in CVR modeling. This work demonstrates how two analytical factors affect the observed relationship between baseline CBF and BOLD-CVR in healthy populations; the relationship becomes even more relevant to understand when interpreting results in
populations with altered vascular baselines or impaired cerebrovascular reserve.

This work was performed in collaboration with Dr. César Caballero-Gaudes’ lab at the Basque Center on Cognition, Brain and Language in Spain.

Check out the paper here!

 

Check out our conference paper that was presented at the IEEE Engineering in Medicine and Biology Conference and published online in December 2021! This paper details our method, developed by PhD candidate Kimberly Hemmerling, for visualization of complex spinal cord fMRI data as heatmaps. This work was inspired by a visualization technique designed for brain fMRI data (Power et al., 2017). Our heatmaps can be visualized alongside motion or physiological traces that may be used to identify coincidence of signal variation. The technique may be easily integrated into a preprocessing pipeline to visualize the effect of a preprocessing step, identify spatially or temporally varying artifacts, or to assess general data quality.Check out the work here, and our GitHub repository (BrightLab-ANVIL/spinalcordplot), which contains the code and example data necessary to generate the plots.

Check out our latest publication in Frontiers Physiology, titled “Cerebrovascular Reactivity Mapping Without Gas Challenges: A Methodological Guide”. This work was lead by the excellent Joana Pinto, now a postdoc in Oxford, and covers breathing task methodology and resting state approaches for measuring CVR. This article is part of a special Frontiers “research topic” on Imaging CVR: Physiology, Physics, and Therapy.

You can find the open access article here.

Announcing the availability of two new publications in collaboration with the SPiN lab, lead by Cesar Caballero-Gaudes at the Basque Center on Cognition, Brain and Language in Spain! Both focus on the preprocessing and analysis of breath-hold fMRI data to measure cerebrovascular reactivity (CVR) and hemodynamic lag:

BCBL PhD student Stefano Moia and ANVIL postdoc Rachael Stickland co-authored an IEEE paper describing our lagged Generalized Linear Model approach for estimating the temporal delay between systemic carbon dioxide changes and the local BOLD fMRI response, mapping CVR and lag at the voxel level. The approach simultaneously fits other fMRI confounds, such as head motion, to better isolate the physiologic effects of interest. Download the pdf here!

Using the BCBL EuskalIBUR dataset, which uniquely studies breath-hold CVR across 10 repeated weekly MRI sessions, a new preprint in BioRxiv demonstrates the benefits of multi-echo data and compares multi-echo ICA denoising strategies for reliably mapping CVR and lag. Download the pdf here!

CVR maps for an example subject, scanned weekly for 10 weeks. Each row reflects a different use of multi-echo fMRI data, either combining information across the echo times or using the relationship across echo times and Independent Component Analysis (ICA) to differentiate the CVR effects from highly collinear noise confounds.