Data-driven HRF estimation for encoding and decoding models

• Published in: NeuroImage (Orlando, Fla.) Vol. 104; pp. 209 - 220
• Main Authors: Pedregosa, Fabian, Eickenberg, Michael, Ciuciu, Philippe, Thirion, Bertrand, Gramfort, Alexandre
• Format: Journal Article
• Language: English
• Published: United States Elsevier Inc 01.01.2015; Elsevier Limited; Elsevier
• Subjects: BOLD; Brain - physiology; Brain Mapping - methods; Computer Science; Decoding; Derivatives; Economic models; Encoding; Estimates; Finite impulse response (FIR); Functional MRI (fMRI); Hemodynamic response function (HRF); Humans; Image Processing, Computer-Assisted; Machine Learning; Magnetic Resonance Imaging - methods; Methods; Models, Neurological; Neurovascular Coupling; Optimization; Regression Analysis; Studies; Visual Perception - physiology; Machine learning; Hemodynamic response function (HRF); Finite impulse response (FIR); BOLD; Encoding; Functional MRI (fMRI); Decoding; Optimization; Finite Impulse Response (FIR); Hemodynamic Response Function (HRF); Machine Learning
• ISSN: 1053-8119; 1095-9572
• DOI: 10.1016/j.neuroimage.2014.09.060

Despite the common usage of a canonical, data-independent, hemodynamic response function (HRF), it is known that the shape of the HRF varies across brain regions and subjects. This suggests that a data-driven estimation of this function could lead to more statistical power when modeling BOLD fMRI data. However, unconstrained estimation of the HRF can yield highly unstable results when the number of free parameters is large. We develop a method for the joint estimation of activation and HRF by means of a rank constraint, forcing the estimated HRF to be equal across events or experimental conditions, yet permitting it to differ across voxels. Model estimation leads to an optimization problem that we propose to solve with an efficient quasi-Newton method, exploiting fast gradient computations. This model, called GLM with Rank-1 constraint (R1-GLM), can be extended to the setting of GLM with separate designs which has been shown to improve decoding accuracy in brain activity decoding experiments. We compare 10 different HRF modeling methods in terms of encoding and decoding scores on two different datasets. Our results show that the R1-GLM model outperforms competing methods in both encoding and decoding settings, positioning it as an attractive method both from the points of view of accuracy and computational efficiency. [Display omitted] •R1-GLM method allows to jointly estimate activation patterns and HRF.•Rank constraint reduces variance in voxelwise fitting and is solved efficiently.•R1-GLM extends to parametric models of HRF and GLM with separate designs.•R1-GLM outperforms competing methods in the fit of the BOLD response (encoding).•R1-GLM also improves decoding accuracy in brain reading settings.