Research data supporting "Fine-scale computations for adaptive processing in the human brain"
MetadataShow full item record
Zamboni, E., Kemper, V., Goncalves, N., Jia, K., Karlaftis, V., Bell, S., Giorgio, J., et al. (2020). Research data supporting "Fine-scale computations for adaptive processing in the human brain" [Dataset]. https://doi.org/10.17863/CAM.60330
Ultra-high (7T) functional magnetic resonance imaging (fMRI) was used to determine the contribution of feedforward and feedback brain mechanisms to adaptive processing, i.e. a rapid form of plasticity critical for efficient processing of sensory information. Specifically, we acquired sub-millimetre (0.8mm isotropic) BOLD fMRI data while healthy participants were presented with blocks of gratings at the same or different orientations and engaged in an attentional demanding task at fixation. We then contrasted the fMRI BOLD responses for the two conditions (adaptation, i.e. repeated orientation, and non-adaptation, i.e. different orientations) at different cortical depths in different regions of interest: early visual cortex (V1), extrastriate areas (V2, V3, V4), and posterior parietal (IPS1, IPS2). This allowed us to inform our understanding of the circuit involved in adaptive processing: increased fMRI adaptation (i.e. decreased BOLD response) was observed for superficial and middle rather than deeper layers across the visual cortex. Moreover, functional connectivity analysis across cortical depths of visual and posterior parietal areas showed increased functional connectivity for adaptation condition across visual cortex: increased feedforward connectivity between V1-V2, V1-V3, and V1-V4 superficial-middle layers, increased feedback connectivity for adaptation condition between deeper layers of V2 and deeper layers of V1, as well as deeper layers of V1 and IPS1. Several pre-processing steps were performed to accurately account for potential biases and distortions in the fMRI images that could confound the BOLD signal. In particular, methods for distortion corrections using images acquired with inverted phase encoding were applied to reduce blurring and distortions due to non-zero off-resonance field; alignment between functional and structural images was validated using mcheck tools provided in this repository (i.e. computing the spatial correlation between images); borders between white matter and grey matter, as well as grey matter and cerebrospinal fluid in each participant's brain segmentation were inspected and manually adjusted. In order to account for superficial bias in the BOLD signal (i.e. increased BOLD signal towards the grey matter - pial surface), several steps were applied: (a) voxels with low temporal signal to noise ratio and high t-values in a stimulus vs fixation GLM contrast were removed from each region of interest; (b) across cortical depths, the same amount of voxels was kept to avoid biases in the signal; (c) BOLD signal was z-scored to control for differences in signal levels across cortical depths while preserving signal differences across conditions. The data presented in the dataset here are therefore the result of these steps, in particular: normalised fMRI responses for each condition (adaptation, non-adaptation) were averaged across the stimulus presentation (excluding responses; 32-34s after stimulus onset), blocks, and runs for each condition. For visual cortex ROIs we focussed on the 4 to 18s after stimulus onset, a time window capturing the peak of the haemodynamic response. These values are reported for each participant, each cortical depth (deeper ,middle, superficial) and each ROI (V1, V2, V3, V4, IPS1, and IPS2). We computed functional connectivity within visual cortex and between visual and posterior parietal cortex. We used ICA-based and Finite Impulse Response (FIR) functions to denoise and deconvolve the fMRI time course data per cortical depth, controlling for noise and potential task-timing confounds. We then conducted Pearson correlations between the fMIR eigenvariate time courses across cortical depths. Feedforward functional connectivity is computed between superficial and middle layers, feedback connectivity is computed between deeper layers. The data presented in the dataset correspond to the difference in Fisher's z-transformed R values between adaptation and non-adaptation, i.e. positive value corresponds to increased functional connectivity for adaptation compared to non-adaptation, for the pathway (feedforward / feedback) of interest. Please see also the file 'Description of uploaded data' for a detailed description of the dataset.
excel, matlab, BrainVoyager
adaptive processing, Laminar fMRI
Publication Reference: https://doi.org/10.7554/eLife.57637https://www.repository.cam.ac.uk/handle/1810/312756
This record's DOI: https://doi.org/10.17863/CAM.60330
Attribution-NonCommercial-ShareAlike 4.0 International
Licence URL: http://creativecommons.org/licenses/by-nc-sa/4.0/