Parameters

The successful running of CNMF-E requires several parameter selections. We specially designed our algorithms to make sure the final performance of the analysis is less affected by these parameters. In practice, these parameters can be easily chosen from a broad range. Here we provided a practical guidance on the parameter selections.

gSig & gSiz

gSig: the Gaussian width of the selected filtering kernel

gSiz: the cut-off width of the filtering kernel.

The spatial filtering step is a very important procedure for running CNMF-E. Besides removing the background components, it also performs as a template matching for highlighting pixels within cell bodies. Thus the kernel usually approximates the spatial footprints of cell bodies in the FOV.

Th value of gSiz is roughly the diameter of a cell body and gSig is close to gSiz/4. The values can be customized to different datasets.

ssub and tsub

bg_neuron_ratio