Unraveling whole-brain activity from light-sheet microscopy recording

© 2019 EPFL

© 2019 EPFL

Researchers from MIPLAB (Prof. Dimitri Van De Ville) have developed a new approach to deconvolute signals generated by calcium indicators in whole-brain light-sheet microscopy of Zebrafish larvae. The methodology overcomes current limitations in terms of temporal resolution and signal-to-noise ratio, which allowed them to retrieve neural activation that matches the locomotor behavior of the Zebrafish larvae.

Recent technological advances in light-sheet microscopy make it possible to perform whole-brain functional imaging in small animals at the cellular level with the use of calcium indicators. The outstanding spatial extent and resolution of this type of data open unique opportunities for understanding the complex organization of neuronal circuits across the brain. However, the analysis of this data remains challenging because the observed variations in fluorescence are, in fact, noisy and indirect measures of the neuronal activity. Moreover, measuring over large field-of-view negatively impacts temporal resolution and signal-to-noise ratio, which further impedes conventional spike inference. Here we argue that meaningful information can be extracted from large-scale functional imaging data by deconvolving the calcium response and by assuming of model of sustained neuronal activity instead of individual spikes. Specifically, we characterize the calcium response by a linear system of which the inverse is a differential operator, which is then included in sparsity-promoting regularization. In experimental data from zebrafish larvæ, the algorithm correctly retrieves neural activation that matches locomotor behavior unknown to the method, but also large-scale patterns of functional neural networks that fire coherently in a spontaneous way. 
FIGURE: A screen shot of activity maps from the whole brain of a zebrafish larvæ. Strong activity in specific groups of neurons is correlated with the locomotor behaviour (left and right turns). Copyright: MIPLAB
ACKNOWLEDGMENTS: Authors are grateful to Misha Ahrens, PhD, and Yu Mu at Janelia Research Campus for providing whole-brain imaging data of behaving larval zebrafish and sharing their code for pre-processing and Engert Lab for the open-source zebrafish brain atlas.

Carl ZEISS AG is funding EPFL scientists to develop signal processing tools for unraveling whole-brain activity from light-sheet microscopy recordings. These new techniques are made available as open-source codes to help neuroscientists working on functional imaging of neuronal populations with large field-of-view.


Younes Farouj, Işık Fikret Karahanoğlu & Dimitri Van de Ville. Deconvolution of Sustained Neural Activity from Large-Scale Calcium Imaging Data. IEEE Transactions on Medical Imaging (in press).

Author: Niels Lion