Nonlinear Deconvolution by Sampling Biophysically Plausible Hemodynamic Models

Abstract

Non-invasive methods to measure brain activity are important to understand cognitive processes in the human brain. A prominent example is functional magnetic resonance imaging (fMRI), which is a noisy measurement of a delayed signal that depends non-linearly on the neuronal activity through the neurovascular coupling. These characteristics make the inference of neuronal activity from fMRI a difficult but important step in fMRI studies that require information at the neuronal level. In this article, we address this inference problem using a Bayesian approach where we model the latent neural activity as a stochastic process and assume that the observed BOLD signal results from a realistic physiological (Balloon) model. We apply a recently developed smoothing method called APIS to efficiently sample the posterior given single event fMRI time series. To infer neuronal signals with high likelihood for multiple time series efficiently, a modification of the original algorithm is introduced. We demonstrate that our adaptive procedure is able to compensate the lacking of inputs in the model to infer the neuronal activity and that it outperforms dramatically the standard bootstrap particle filter-smoother in this setting. This makes the proposed procedure specially attractive to deconvolve resting state fMRI data. To validate the method, we evaluate the quality of the signals inferred using the timing information contained in them. APIS obtains reliable event timing estimates based on fMRI data gathered during a reaction time experiment with short stimuli. Hence, we show for the first time that one can obtain accurate absolute timing of neuronal activity by reconstructing the latent neural signal.

Hemodynamic Deconvolution Demystified: Sparsity-Driven Regularization at Work

“…Most approaches assume a linear time-invariant model for the hemodynamic response that is inverted by means of variational (regularized) least-squares estimators (5,6,(28)(29)(30)(31)(32)(33)(34)(35)(36), logistic functions (37)(38)(39), probabilistic mixture models (40), convolutional autoencoders (41), or nonparametric homomorphic filtering (42). Alternatively, several methods have also been proposed to invert nonlinear models of the neuronal and hemodynamic coupling (43)(44)(45)(46)(47)(48)(49).…”

Hemodynamic Deconvolution Demystified: Sparsity-Driven Regularization at Work

“…Most approaches assume a linear time-invariant model for the hemodynamic response that is inverted by means of variational (regularized) least squares estimators (Glover 1999;Gitelman et al 2003;Gaudes et al 2010Gaudes et al , 2012Gaudes et al , 2013Caballero-Gaudes et al 2019; Hernandez-Garcia and Ulfarsson 2011; Cherkaoui et al 2019;Costantini et al 2021;Hütel et al 2021), logistic functions (Bush and Cisler 2013;Bush et al 2015;Loula et al 2018), probabilistic mixture models (Pidnebesna et al 2019), convolutional autoencoders (Hütel et al 2018) or nonparametric homomorphic filtering (Sreenivasan et al 2015). Alternatively, several methods have also been proposed to invert non-linear models of the neuronal and hemodynamic coupling (Riera et al 2004;Friston et al 2008;Havlicek et al 2011;Aslan et al 2016;Madi and Karameh 2017;Ruiz-Euler et al 2018).…”