decay_time parameter interpretation

[camilledonoho]

Hi,

In the notebooks the parameter to specify GCaMP dynamics is the value "decay time" in seconds. Does this imply that this value is the "half-decay time" referenced in the GCaMP literature? Or is it something different?

Also, how much of a direct impact does decay time have on segmentation? Is this parameter more important for later deconvolution or is it also used in segmentation?

Thanks!

[pgunn]

Hello,
decay_time is used to help constrain the range of frames used in detecting components (see caiman/components_evaluation.py:estimate_components_quality_auto()

[kushalkolar]

it almost certainly has to be half decay time, it's an exponential decay so the full decay time is technically inifinity

it has nothing to do with "segmentation" which is just greedy initialization for the spatial seeds, it's essential for getting meaningful denoising and deconvolution