Brainways: An AI-based Tool for Automated Registration, Quantification and Generation of Brain-wide Activity Networks Based on Fluorescence in Coronal Slices

Abstract

A central current trend in neuroscience involves the identification of brain-wide neural circuits associated with complex behavior. A major challenge for this approach involves the laborious process for registration and quantification of fluorescence on histological brain slices, as well as the difficulty of deriving functional insight from the complex resulting datasets. As a solution, we developed Brainways, a simple-to-use AI-based open-source software for the identification of neural networks involved in a specific behavior, from digital images to network analysis. Brainways offers automatic registration of coronal slices to any 3D brain atlas, and provides quantification of fluorescent markers (e.g. activity marker, tracer) per region, as well as statistical comparisons with visual mapping of contrasts between conditions. A built-in partial least squares task analysis provides the neural patterns associated with a specific contrast, as well as network graph analysis representing functional connectivity. Trained on atlases for rats and mice, Brainways currently provides above 80% atlas registration accuracy and allows the user to easily adjust the outputs for better fit. Below, a case study validation of Brainways is demonstrated on a previously published data set describing the neural correlates of empathic helping behavior in rats. The original results were successfully replicated and expanded upon, due to the exponentially larger sample size that covered over a 100 times more brain tissue compared to the original manual sampling. Brainways thus provides a fast, accurate solution for quantification of large-scale projects and facilitates novel neurobiological insights about the structural and functional neural networks involved in complex behavior. Brainways has a highly accessible GUI and is functionality exposed through a Python-based API, which can be enhanced for different applications.