BOLD response is more than just magnitude: improving detection sensitivity through capturing hemodynamic profiles

Abstract

Typical FMRI analyses assume a canonical hemodynamic response function (HRF) with a focus on the overshoot peak height, while other morphological aspects are largely ignored. Thus, in most reported analyses, the overall effect is reduced from a curve to a single scalar. Here, we adopt a data-driven approach to HRF estimation at the whole-brain voxel level, without assuming a profile at the individual level. Then, we estimate the BOLD response in its entirety with a smoothness constraint at the population level to improve predictive accuracy and inferential efficiency. Instead of using just the scalar that represents the effect magnitude, we assess the whole HRF shape, which reveals additional information that may prove relevant for many aspects of a study, as well as for cross-study reproducibility. Through a fast event-related FMRI dataset, we demonstrate the extent of under-fitting and information loss that occurs when adopting the canonical approach. We also address the following questions: 1) How much does the HRF shape vary across regions, conditions, and clinical groups? 2) Does an agnostic approach improve sensitivity to detect an effect compared to an assumed HRF? 3) Can examining HRF shape help validate the presence of an effect complementing statistical evidence? 4) Could the HRF shape provide evidence for whole-brain BOLD response during a simple task?

Whole-brain, gray, and white matter time-locked functional signal changes with simple tasks and model-free analysis

“…Here, we used a spectral analysis that made no assumptions about the detailed shape of the BOLD signal changes. In line with our results, recent work that estimates the BOLD response directly from the signal ( 28 ), permitting flexibility of hemodynamic profile and shapes, revealed hemodynamic activity in white matter that was not found when using the canonical gray matter hemodynamic response function. Similar approaches that relax the overly strict response models may be necessary to improve detection sensitivity in white matter.…”

“…For example, previous studies have reported white matter BOLD signal responses to tasks and at rest ( 13 – 16 ), relationships between white matter signals and the gray matter regions to which they connect ( 17 – 19 ), alterations of white matter BOLD signal in disease/disorder ( 20 – 25 ), and finally, robust, reproducible network properties of white matter BOLD signals ( 18 , 26 ). Importantly, recent reports have confirmed that white matter responses to stimuli are similar to, but different from, those in gray matter ( 13 , 14 , 27 , 28 ), generally indicating a slower response, smaller percent signal change, and variation across regions ( 29 ). Together, the explicit removal of white matter BOLD signal as a nuisance regressor ( 30 ), inaccuracies in modeling white matter BOLD response functions ( 28 ), and decreased detection sensitivity due to low SNR, has made white matter a blind spot in the fMRI literature ( 30 ).…”

“…Importantly, recent reports have confirmed that white matter responses to stimuli are similar to, but different from, those in gray matter ( 13 , 14 , 27 , 28 ), generally indicating a slower response, smaller percent signal change, and variation across regions ( 29 ). Together, the explicit removal of white matter BOLD signal as a nuisance regressor ( 30 ), inaccuracies in modeling white matter BOLD response functions ( 28 ), and decreased detection sensitivity due to low SNR, has made white matter a blind spot in the fMRI literature ( 30 ).…”

Estimation of fMRI responses related to epileptic discharges using Bayesian hierarchical modeling

“…This concept has been proposed by Spencer et al (2020), but only lobe‐level correlations were considered due to current computation power limitations; (4) although the adapted four HRF process considered the variability of HRF shapes across brain regions and balanced fitting accuracy and generalizability, there are more flexible approaches that do not require predefined HRFs. For instance, the HRF deconvolution approach with regularization on the smoothness of the HRF can directly estimate the HRF from data (Chen et al, 2023; Cherkaoui et al, 2021; LeVan et al, 2010; Lu et al, 2006, 2007). This might be more effective for our application, considering that the variability of HRF shapes related to epilepsy discharges might not be fully encompassed by the four HRFs; (5) when comparing the presurgical evaluation performance with our previous study that employed a voxel‐wise approach involving a mixed‐effects model for within‐ and between‐run variability of the BOLD responses, an improvement was observed primarily in specificity.…”

Blink-related arousal network surges are shaped by cortical vigilance states

“…The canonical HRF convolution assumes standard time-to-peak values de ned by the HRF and does not consider the brain region nor the state of vigilance. However, the time-to-peak and the shape of the HRF BOLD response might differ between brain regions and for different subjects (Handwerker et al, 2004;Rangaprakash et al, 2018) with substantial HRF shape variability (Chen et al, 2023). Vigilance could also impact HRF responses due to excitability changes.…”