Hi all,
We’ve recently updated our Leica inverted epi scope. It has a motorized stage and turrets. We’ll be running LAS X with modules for automated plate imaging (96 well, autofocus), and calcium imaging.
The plate-based image analysis will be done on other computers, but I suspect some portion of the calcium imaging will be done on this machine.
We want to build a custom workstation. What components should we optimize? What matters in terms of performance in image aquisition as opposed to analysis?
Sounds fun! Some comments from my experience and then a question for you.
We basically only do high-throughput tiling and multi-well imaging, so pretty similar to what you are describing. We work with a Nikon/Leica/Zeiss (pharma auctions are dangerous ;-)) and we have some general guidelines that seem system independent.
a. Input/Output cards: A dedicated USB 3 (or faster) card is more reliable than the built-in USB ports that might be on the motherboard. We’ve had a lot of trouble with built-in ports. If your camera supports a faster transfer protocol, you can consider that as well if needed. Finally, consider how your team will move data from the system. USB/thunderbolt SSDs are great for moving data, but make sure you have enough/right kind of ports for this. My students were spending hours transferring data until I forcibly upgraded all of them to USB SSD drives and made sure they understood the difference between USB 2 and USB 3 ports on the computer.
b. Drives: We run the OS/scope software off a separate hard drive from the capture drive. We use a basic 256-512 SSD for the OS/LASX/Elements/etc, and then we use a 1TB NVMe SSD for acquisition specifically. This prevents streaming capture issues or hold-ups when large stitches or datasets are being captured.
c. RAM: More seems to be better, but not always relevant. More RAM can help with buffers for some things, and allow other apps to be running. It’s also really important for spot-checks of the data if we want to load things in FIJI. It’s sometimes difficult to work within LASX’s gui for inspecting data mid-run.
d. GPU: Honestly unsure what LASX does with the GPU. Our Nikon makes use of a Quadro card, but it’s not that powerful. If you are going to do deconvolution-on-the-fly then the GPU definitely more important, but you can upgrade this later as well.
Question: do you have a good solution for multicolor imaging with LASX to allow all images in one color to be captured first and then the next? I’m trying to minimize filter turret switches, but I can’t figure this out and it seems like we’re forced to constantly switch colors at every individual position first.
Thank you very much for this. I’m following this very closely! Can’t help feel that vendors aren’t providing cost effective future proof solutions.
In Nikon Elements there is a section in the multi-dimensional window to dictate the order of acquisition (I.e z, then lambda 2, or Z1 lambda 123, same for tiles)
Thank you for answering my question. This is very helpful! Unfortunately, I don’t have an answer for the Leica LASX, but have you tried reaching out to them directly?