Dataset for article "Goal-directed shaping of cortical waves"

Institution: Université Paris-Saclay, CNRS, Institut des Neurosciences Paris-Saclay (NeuroPSI), 91400,Saclay, France. Isabelle Ferezou and Luc Estebanez are co-last authors. Correspondence: [e-mail], [e-mail] BioRxiv DOI: https://doi.org/10.1101/2024.09.10.612344 Abstract At the surface of the cerebral cortex, the dynamics of brain activity at the mesoscopic scale are characterized by complex patterns of synchronized neuronal activity, including waves. These dynamical structures in the cortical activity have been shown to impact the processing of sensory information, but can they be actively shaped in a goal directed manner? To address this question, we designed a fast widefield optical brain-machine interface for mice, that can detect and reinforce individual patterns of activity in function of their specific displacement at the surface of the somatosensory cortex. Among mice that learned the task, half solved it by consistently generating traveling waves of activity which became progressively more stereotyped. These conditioned patterns of activity resulted from a reshaping of the cortical activity generated during limb movements via processes that included a sharp pre-movement cortical suppression that emerged with learning. Our work demonstrates that cortical mesoscale patterns of activity—and in particular traveling cortical waves—can be subject to operant control. It provides evidence for the plasticity and functional relevance of large-scale cortical dynamics, and establishes a new paradigm for manipulating mesoscale brain activity. Dataset Cortical pattern dataset The dataset consists of a set of .mat files acquired on multiple consecutive training days, in 16 mice. To access the data in the .mat files, Matlab should include in its path the class definitions found in the Class_description.zip file. For each session contained in the .mat file names with unique identifier YEAR-MONTH-DAY-HOUR_MINUTE_SECOND of the corresponding experiment, there are several fields, and in particular including: Waves: the listing of all patterns that took place at the surface of the cortex during the recording. RewardableWaves: patterns that filled the requirement for reward. the wave structure includes the time of occurence, path at the surface of the cortex, and strength of the wave. Reference: a construct that includes a reference blue and green image, the mask that delineates the cortical window, and the ROIs that are used in the operant conditionning, enabling to identify the condition that is trained on a given session. Mesoscale calcium imaging dataset In addition to this systematic dataset, the full calcium image time series acquired on the first and best performance session in Mouse 12 have been also provided. The cortical pattern trajectories are derived from the analysis of these raw images. Code for the extraction of patterns from cortical imaging The matlab script provided in compressed folder MesoBrain_code extracts cortical patterns from time series of mesoscale cortical imaging with calcium reporter GCaMP6f. It is compatible with modern versions of the matlab platform (later than Matlab 2020), and depends on the "Signal Processing" toolbox. This code is also available in GitHub, at https://github.com/adogadov/MesoBrain Inputs: dFF0: 3D matrix (2D space x time) that contain a time stack of cortex calcium imaging fluorescence data (dF/F). Mask: Binary image mask that defines the contour of the cortical window. Outputs: Waves: an object containing individual wave instances detected on the time stack, that encompass a description of the individual trajectories (x,y), onset time (t0), and strength of the fluorescence at the local maximum. Demo: Example data (mask and time stacks) are provided in the Demos folder. Additional data is available in compressed folder DATA.