ImageJ - Avoid / get rid of out-of-focus halo after maximum intensity projection (solved)

I have acquired some immunofluorescence images of a HeLa cell line stained against the processing body (P-body) marker DCP1A. As I have acquired the images at a high magnification (100x) all of the P-bodies are not in focus at a single z-plane. I have therefore acquired z-stacks during image acquisition to capture as many P-bodies as possible in focus. My goal is to quantify the number, pixel intensity and area of P-bodies.

My problem is that when I make a maximum intensity projection of my image stacks in ImageJ, out-of-focus light creates a halo effect around P-bodies in the final image. The halo subtly increases the size of the segmented P-body area when thresholding (see images below).

Single slice:
single_slice

Maximum intensity projection:
maximum_intensity_projection

Line profile:

Including fewer slices in the projection would help, but result in some P-bodies not entering focus and therefore not be detected in the segmentation step. Is there a method to get rid of or completely avoid the out-of-focus halo when doing z-stack projections? I am not a fan of the average intensity projection as it affects the pixel intensity quite profoundly. Any ideas?

Thanks!

/citizen_043

I haven’t tried this much with IF images, but you might be interested in an Extended Depth of Focus functionality. I usually play around with that in Zen and brightfield images, but if you can find an implementation, it might be worth a try.

Alternatively, there are a few deconvolution modules for FIJI that might specifically be tuned to your problem, but would require some digging into and image processing time.

1 Like

This might help if you opt for the decon.

1 Like

Hello Research_Associate,

Thank you so much for your reply! I tried to Google Extended Depth of Focus and found this plugin for ImageJ (http://bigwww.epfl.ch/demo/edf/). From a quick try out it appears to work really well!

I am planning to learn to do deconvolution, but from what I can understand the best way to do it is to experimentally determine the PSF of the microscope with fluorescent beads? Anyway, I think Extended Depth of Field will suffice for now.

Thanks!

/citizen_043

1 Like

Yes, always best, but if you have a good, well tuned system, you can frequently get good improvement with estimated point spread functions based on information about the system. Most software or plugins will have a way to estimate the PSF based on a number of parameters like the objective, NA, wavelength, etc.