I am currently trying to determine the cost/benefit analysis for getting a fast spinning disk confocal. My major question is why should it matter that a confocal, for example CSU-X1 can do 2,000 frames per second and a CSU-W1 can do 200 frames per second, if my camera can only do 100 fps. What am I missing?
This does not even consider other components that could slow this process down. For example is it not more important to consider other things like: dual camera simultaneous imaging, triggering, or sectioning, rather than putting too much focus on what spinning disk speed you are using?
Many modern cameras can go well beyond 100fps, though it usually requires restricting the field of view. (That 100fps you are mentioning was probably full frame). So the camera isnât always the limitation. It comes down more to your sample. If you donât ever need more than 200fps to capture what youâre looking at, then it doesnât matter. But if youâre looking at crazy fast things, then that extra 1800fps in the X1 might be useful. Itâs a big âmightâ though, cause in addition to restricting your field of view on the camera, actually achieving 1000fps also assumes that you can get sufficient SNR with a 1ms exposure. That in turn depends on the maximum laser power in your system and the intrinsic brightness of your sample. (Of course, issues of photobleaching and toxicity resulting from the increased laser power aside).
(As a side note here, remember also that getting the very highest frame rates theoretically achievable by a spinning disk requires pretty good synchrony between the disk speed and exposure time if you are to avoid disk artifacts from incomplete frame coverage).
So those numbers about spinning disk frame rate that you were quoting usually refer to the time it takes to make one complete sweep of the field of view (for instance, 1/12 of a rotation on an X1).
I would call those other issues (triggering, simultaneous multichannel, etc) independent of the question of speed⌠and only you can decide whether they are âmore importantâ to you than the maximum frame rate your system can achieve (again, based on the demands of your experiment and your biological question).
edit: since you mention triggering, itâs worth pointing out that on any of these system, hitting something like 500+ FPS probably assumes that youâre just streaming the camera nonstop, without turning the laser on and off⌠so triggering is not an issue for speed in that case. Though a carefully triggered system would certainly have benefits for minimizing light dose at lower frame rates (so again⌠I see that as an independent issue)
We actually run our CSU-X1s below maximum speed most of the time and have only rarely needed to turn up to full speed. In most cases, something else limits the frame rate (exposure time, z motor movement, filter switching, etc). As Talley said, what matters most is your sample and any frame rate requirements imposed by your experiment.
That said, in my opinion speed isnât the most important difference between the X1 and the W1. I havenât used a W1 much, but my understanding is that the field of view is bigger and thereâs a wider variety of disk options. Even if you get a disk with the same pinhole size as the X1, the pinhole spacing on the W1 disk is larger, and we have some anecdotal evidence that the increased spacing makes quite a big difference for samples with a lot of out of focus light. Also, because the X1 is a much older model than the W1, it may be more difficult to get replacement parts/service - definitely something to ask about in the purchasing process. Again, take this with a grain of salt - I donât have a W1!
And of course, itâs very difficult (impossible?) to completely predict performance. The best way to tell which one is going to work best for you is to demo each model with your samples.
Speed of the disk relative to the camera matters due to aliasing.
Imagine you have a pinhole disk with 1 set of pinholes. If it spins at 200 fps, then it traces out pinholes over the entire FOV 200 times per second, or once every 0.5 ms. If you use a camera exposure time of 1 ms, then you get 2 full cycles of the disk over your sample. If instead you had an exposure time of 1.25 ms, then youâd get 2.5 cycles. Half of your specimen would get a 2x-sized dose of light, and the other half a 3x-sized dose. The image would have inhomogeneous intensity across the FOV. In a stack, each image would have this inhomogeneity but the line between the two areas might move. This manifests as banding or flickering in the image. Is an example of the âwagon wheelâ effect.
This paper describes the issue specific to an EMCCD implementation:
something similar would happen with sCMOS cameras together with the rolling shutter.
More patterns per disk, faster disk rotation, and longer exposure times all help even this out. For example, if part of the FOV sees 1000 transits if the pinhole pattern and the rest sees 1001, itâd be tough to tell. But ultimately the best way to avoid this is to synchronize the exposure with the disk rotation period.