3D bright field and fluorescence image registration

I’m trying to register a 3D bright field stack to a MIP of fluorescent channels for adherent cells (Nile-Red stain of lipid droplets). images were taken on a spinning disk confocal. Visual inspection suggests there is some shift between BF planes in the stack, as well as a shift between the BF images and the fluorescent ones. I’m looking for a registration algorithm to solve both this issues. I know for fluorescence that is easy but BF registration is more complex. I saw many algorithms but don’t know how they compare in terms of performance for BF and BF to fluorescence. manual inspection of the registration result isn’t possible due to the volume of images needed to be processed.
Any help would be greatly appreciated,
Barak.AssayPlate_PerkinElmer_CellCarrier96 Ultra_C03_T0001F001L01A01Z01C03.tif (7.6 MB) AssayPlate_PerkinElmer_CellCarrier96 Ultra_C03_T0001F001L01A01Z02C03.tif (7.6 MB) AssayPlate_PerkinElmer_CellCarrier96 Ultra_C03_T0001F001L01A01Z03C03.tif (7.6 MB) AssayPlate_PerkinElmer_CellCarrier96 Ultra_C03_T0001F001L01A02Z01C01.tif (7.6 MB) AssayPlate_PerkinElmer_CellCarrier96 Ultra_C03_T0001F001L01A02Z01C02.tif (7.6 MB)

*Have applied this to other brightfield/IF combinations, but not the specific use case you describe.

I haven’t taken a look at the images, but assuming the brightfield images are mono, I recommend inverting them and using BF as if it were a fluorescent channel. It may help to set some sort of threshold before/after the inversion if the background is similar enough across all images.