Smooth slow dynamics baseline estimation causes larger transients at later time points

[slceto]

When using the "smooth slow dynamics" option for baseline estimation, I can see that the amplitude of calcium transients increases as time goes on across pretty much all cells (see image below). When I look at my raw traces, it's clear that the baseline fluorescence is gradually falling over the course of the ten minute video (I'm guessing it's a bit of photobleaching), which is probably causing an artifact where the same amplitude of raw transient now looks bigger in the dF/F trace because the baseline is lower. Is there any way to tweak the code to account for this?

Thanks!
(BTW if anyone would like to use the script I made to only call the increasing portion of the calcium transient significant, I posted it here: https://github.com/slceto/CalFDR)