Fourier ring correlation plugin for beads

Dear all,

I am using the Fourier ring correlation plugin from the BIOP site. I have two confocal images from two APDs separated by a 50/50 beam splitter. The signals are from the scattering of gold beads. After running the plugin, I obtained a result like this. There is a huge amount of high-frequency components and a “resolution” cannot be determined. Does anyone have an idea why this happens? Sorry I cannot upload the original tiff images because of the restriction of the forum.

Kind regards,

Hello Vincent,
There are many reasons for this, mostly because of the surface of the object being imaged which can be dealt with.
Respond to this and you should be able to then upload an image of what is going on.

Hello Bob,

Thank you for your response. Could you have a look at my images.gold_beads_10nm_561nm_apd1.tif (1.4 MB)
gold_beads_10nm_561nm_apd2.tif (1.4 MB)


Sure Vincent,
Be back with you shortly,

Hello again,
I am not familiar with the BIOP site or plugins, but here are two enhances of your images.Vincent ca59475d71b0c5736ab0810660bc1b18a94ad6b9_A.tif (1.4 MB)
Vincent_7695c9cbb29c9b99551d4a8282332f4d69057466_A.tif (1.4 MB)
I also used Process > Image Calculator > Difference to see if there is any, there is a very small difference so I’m not quite sure of what you are looking for.
Therefore if you need further help please clarify to me what it is that you are looking for and I’ll give it another go.

Hello Bob,

thank you again for your reply. What do you mean by “enhances”?

The two images are the same field of view but from different detection channels separated by a beam splitter, so they are “with independent noise realizations” (Tortarolo, G., M. Castello, A. Diaspro, S. Koho and G. Vicidomini (2018). “Evaluating image resolution in stimulated emission depletion microscopy.” Optica 5(1): 32-35.). This is needed to calculate Fourier ring correlation to determine the resolution. Of course I can get the resolution by fitting the spot with Gaussian and take the FWHM, but I am trying to do the same with FRC. I expect a correlation curve that drops close to zero, not like the one in the original post.

Hi Vincent,

the FRC measures the similarity between two images, which supposedly are different only for their noise content. In this respect it is an indirect measure of the signal-to-noise ratio in your final image, and the resolution estimate you obtain is an indication of the smallest feature in it that can be reliably resolved above the noise. In your images there is almost no noise in the background, and the noise over the spots is also very small. Basically the FRC curve tells you that you don’t have any limitation from the noise in resolving the spots, and therefore your resolution is actually limited only by the PSF, which can be estimated by gaussian fit as you already mentioned.


Hello again Vincent,
Since you were looking for noise and not intensity I simply balanced and increased the dynamic range for the images for better detection. By balanced I mean that the maximum intensity of each image are equal and range are equal.
If I understand Giovanni correctly, the comparison I made to compare the images gives a visual indication of the noise level differences which were very small.