@epnev got it! A couple more questions: 1/ How is the temporal data calculated over each neuron (rather than each pixel)? Is it a sum / mean of each constituent pixel's fluorescence over time? Or is it a weighted mean? (many of my components are donut shaped, ie. spikes occur along the rim, so a simple average would attenuate the df/f values.)
2/ How is the background estimated? Is this estimation subtracted from each component's temporal data?
@angusfong It is roughly a weighted mean with the values of the spatial components after all the other components and background have been removed. The background is computed as a low rank components on the residual image after all the neuron activity has been computed. Please see the paper for details.