Processing ROI data appropriately

[alessandrofelder]

The z position of each imaging plane is used as it's identifier, and is read both from ROI.dat and Zplane_Pockels_values.dat. For the former, the data is read as float16 by pandas, but as the default float64 for the latter. This wasn't a problem previously, as all z values were multiples of 0.5, and usually integers, which can be accurately represented with 16 bits.

More modern data (e.g. LabViewData2020\HG_30_exp01\200107_13_13_33 FunctAcq) has z values that don't follow this pattern anymore, which causes differences between the 16 and 64 bit data that break our code (We use the z-position from ROI.dat as a key in a dictionary for a zplane from Zplane_Pockels_values.dat, so those values need to match).

I will fix this for now by reading as float64 in both cases, but we should have a discussion about what precision we need for this and other columns in both .dat files.

[ageorgou]

Will confirm with the group, but from memory:

• x and y start/stop are integers (though written as decimals in file) since they're pixel "numbers"
• z start/stop used to have a resolution of 0.5, but now that's increased to 0.1, so we do need the higher precision

[alessandrofelder]

That's what I remember too. Sameer was going to send us an updated sheet that included types for all the columns of ROI.dat.