Question and answer session with Dr Ricardo Henriques
What's the difference between NanoJ-SRRF and SRRF-Stream? Are they based on the same algorithm?
Yes, both NanoJ-SRRF and SRRF-Stream use the same mathematical basis, particularly on the calculation of the radiality transform. NanoJ-SRRF does have a few more analytical features, particularly for temporal analysis, that have been reduced in SRRF-Stream in favour of considerably accelerating the analysis time.
What fluorophores work best with SRRF? How does the choice of fluorophores impact the spatial resolution?
Fluorophores that are bright and blink are the best. Our all time favourite dye is Alexa647, and genetically encoded fluorophore Skylan. The brighter and the more they oscillate their intensities, they higher the resolution you get. However, SRRF will work with any fluorophore, just with less resolution improvement for fluorophores with lower intensities and minimal intensity fluctuations.
What microscopes work best with SRRF?
Optical sectioning helps a lot. SRRF image quality and resolution benefits from reduction of fluorophore density within the field of view, and as suchreducing the volume in Z helps by reducing the contributions of out-of-focus fluorophores perturbing the image. Also, SRRF works better on cameras than on laser scanning devices as you get better temporal sampling on a camera.
I would say SRRF works best in TIRF, then Spinning Disk, then on widefield if you’re imaging thin samples, then on Laser Scanning Confocals. I would expect that SRRF would work extremely well on a lattice light-sheet microscope (which we will be soon trying it on). We have seen that, especially for live samples, SRRF performs well with LED illumination.
Is SRRF compatible with both EM-CCDs and sCMOS cameras?
It works best on EM-CCDs as sCMOS have a pixel-wise non-homogenous noise and gain. However, we are currently working on improving SRRF for sCMOS. We’ve seen good SRRF data on sCMOS if you have a high signal-to-noise ratio.
How should data be acquired for SRRF? What are the best conditions?
The rule of thumb is acquiring 100 frames at 100 frames-per-second to generate a SRRF image. However, it’s impossible to define perfect acquisitionparameters as they are very much sample- and microscope-dependent.
Currently we use our other algorithm NanoJ-SQUIRREL to evaluate the evolution of the quality of SRRF images over different acquisition settings. This allows us to find the optimal parameters to image a sample.
What artefacts can be expected when using SRRF?
As with any other super-resolution method, there are artefacts that can appear on SRRF images. NanoJ-SQUIRREL mentioned above can help identify these problems. The major artefact we expect is that motion from the sample will prevent SRRF from getting sufficient temporal information to accurately represent some structures. So, a fast moving cell will yield poor SRRF images.
How easily can you do two-colour (or three!) live microscopy using the SRRF system?
Quite easily, you will find some examples on our website or Andor’s website. This is simple for fixed samples, however for live samples the main challenge will be that most SRRF images are acquired at around 1 frame-per-second, and when doing multicolour this may mean a significant delay between channels.
Is the Z resolution determined by the microscope/acquisition system? Is there any advantage to using TIRF or airy scan microscopes with this system?
Yes. SRRF doesn’t increase resolution in Z nor does it provide “virtual” optical sectioning capacity. This means that you need an optical method to constrain the resolution in Z. TIRF in this aspect is optimal and confocals will also do a good job.
Any reason why SRRF couldn't be combined with light sheet microscopy for even lower illumination dose?
Not really, although the low-NA objectives may constrain the quality of SRRF images. We are internally testing SRRF on a diSPIM and will soon test it on a Lattice Light-Sheet Microscope.
What are the limitations of SRRF - for example in terms of fluorophore density?
SRRF does not have an upper limit of fluorophore density, however as expected resolution will be worst at high densities.