SpecSeg: cross spectral power-based segmentation of neurons and neurites in chronic calcium imaging datasets

Imaging calcium signals in neurons of awake, behaving mice using two-photon microscopy makes it possible to study how learning alters coding in large neural populations. Such experiments produce extremely large datasets requiring powerful automated methods to extract responses from hundreds of neurons. Here we present a new open source toolbox for 1) segmentation of regions of interest (ROIs) representing neuronal structures, 2) inspection and manual splitting, addition or rejection of ROIs, 3) neuropil correction and signal extraction and 4) matching of ROIs in sequential recordings. Our toolbox is built around a novel method for the identification of ROIs based on temporal cross-correlations of low frequency components derived by Fourier analysis, of each pixel with its neighbors. The approach is insightful for the end user and allows for the ROI detection around neurons or irregular structures such as dendrites and axons by adjusting simple morphological constraints. This new pipeline thus provides an efficient and user-friendly approach for analyzing calcium responses in neuronal structures imaged over prolonged periods of time.