Hello all! I’m in the market for a slide scanner and would love to hear peoples experiences with different models. I know a lot about the Zeiss AxioScan, less about the others. What do you like, what do you find frustrating, etc. More colors is always better. Many thanks!
So we are in the middle of a similar dilemma. We had an olympus dotslide - in Scotland Olympus representation is a little on the sparse side, but I can’t comment on elsewhere. Although I felt the Dotslide we had gave good quality images. It was buggy and we did need technical support from time to time. This was replaced by a Hammamatsu Nanozoomer. What’s good about it is it’s dead easy to use and can hold 180 slides. What’s not so great is fluorescence, it has a triple filter cube, and singles could be added. The control software for fluorescence is not my favourite. We’re looking for other alternatives at the moment specifically for Fluorescence. I like what Nikon does although A Biostation is quite pricey for what we need. Throughout the university in Edinburgh we have 5 AxioScans - which I think is telling. We have moved over to QPath for Analysis and that has been awesome so far.
This is great information, thank you!!
If you are planning to do fluorescence the Zeiss AxioScan is a really nice system and it rarely let us down throughout the 4 years that we have it.
I agree that the Nanozoomer is a nice system as well, in particular for widefield but fluorescence is not good and they have an odd data format that can not be read by BioFormats and thus OMERO which is a nogo for us.
Also important to consider is the loading regime: does the system load single slides or does it use a tray? Usually trays (like on the Axioscan) are better because sticky slides with excessive amounts of mounting medium will not get stuck…
Unless there is some version dependence, the Nanozoomer’s NDPIS files (NDPIS+# NDPI channel files) should be readable by BioFormats, unless the system itself is set up incorrectly (our was for a while). It requires wavelength information to be set. On the other hand, I also don’t like the control software for our Nanozoomer, as it is incredibly limited. You select exposure on a 1, 2, 4, 8 scale? Really?
I’m not sure if OMERO has further problems unrelated to BioFormats as I don’t have any experience with it.
On a somewhat similar note, the quality of the 3DHistech scanners was quite good, and possibly the price? But the resulting MRXS format is terrible, and only supported (partially) by OpenSlide and another couple of libraries that are less used than BioFormats. Of course, if I am, in turn, wrong about that I would love to know!
Ah, Edinburgh, cool! Not surprised you have started using QuPath
Do you know what filter cubes you are using for the AxioScans? I was helping with some severe bleedthrough problems at another nearby institution using the Quad cubes… they ended up swapping to singles despite the slowdown in scan speed. I am not sure how consistent the bleedthrough problem actually is across many fluorophores though.
Maybe I could get a better idea through FPbase if we are ever shopping…
I have seen and used an Olympus VS120 multiple times, it is excellent. There are options to the build, uses an LED, I think it has an 8 position filter wheel, depending on the rest of the build you can have many wavelengths. Can get it with fluorescence only, or fluorescence and a great color camera. Can get it with a 6 position stage or an autoloader. I’ve heard rumor that the VS120 is being replaced by a new instrument - this could get you a good deal!
Just saw something really interesting from Nikon. BioPipeline-slides, part of their High Content Analysis systems products.Check out the website and brochures. https://www.microscope.healthcare.nikon.com/products/high-content-imaging
The choice will depend a lot on what your priorities are, such as throughput (how many slides you can load/ which format), scanning speed (if you’ll do many whole slides might be an important factor, and how many times you’ll need to change the filter cube may become critical for speed), spectral channels, autofocusing reliability, automation (such as auto-detection of the tissue sections etc.), flexibility for non-standard applications (like high-plex) vs. streamlined standard slide scanning and budget.
Perkin Elmer has some powerful high-throughput systems like Polaris, which are good but expensive.
Olympus VS120 is decent, but there is little you can change with what it does, and autofocusing has not always been reliable for me.
Rarecyte and GE also have some systems that you can look into.
Thorlabs started to offer a TIDE slide scanning system that is expected to be really fast in scanning ( https://www.thorlabs.com/newgrouppage9.cfm?objectgroup_ID=6570 ).
What is your opinion on the brightfield imaging on the 3DHistech? We had them demo it for two days, but we never through to ask about the format.
The quality was good, the format is… kind of terrible and not at all open source friendly.
Things to be aware of:
The files are a tiny header file, with the data stored in a folder of the same name, so you can’t just grab a single file for any given image, it is a file+folder.
Hopefully they have improved export functions in CaseViewer, but when I last checked the only options were a single giant TIFF. The overall software and file design seems intended to railroad you into using their analysis software, which is fine if you want to use their analysis software. I would be VERY sure that you can do what you need there, though, if you intend to go that way.
That said, OpenSlide is the one option for opening the MRXS file type (BioFormats is a no go), and as far as I know it is no longer being updated, so any further changes to the mrxs format may break even that option. Openslide already does not support Zstacks as Zstacks did not exist when the last version of Openslide was released. Probably not an issue if you are only interested in brightfield analysis.
I never had any problem with the speed or quality of focus with the hardware, but I did NOT like the adaptive exposure time/contrasting per tile, which impacted any downstream image analysis (I could see that the apparent optical density was changing per tile, impacting all of my thresholds and measurements).
There may be some other workarounds for exporting, as seen here, but I have not tested these. You may want to take any sample images you have gotten through the process to see how they fare, depending on your downstream analysis.
Further note, the JPEGXR encoding wasn’t read by QuPath or a couple of other analysis programs last I checked, but that could be adjusted at the scanner.
Extra link expanding on that last bit, and I added a link above regarding BioFormats.
Wow, that’s really fast response, Thanks! and very valuable opinions! We are going to think twice before making a decision now.
I very strongly recommend the BioFormats link, as they make some very good points that you might bring up with the vendor. If they are willing to address those issues, or enough people complain…
The hardware always seemed pretty good. Shrug.
We will certainly talk to them.