If you follow this blog, you’ll be aware that I paused my monthly live teaching sessions at the start of the pandemic, around 18 months ago, and that I haven’t held any since. Various life events and work commitments got in the way of arranging an online substitute. However, I’m gradually moving in that direction, and a couple of weeks ago held a mini-teaching session using scanned slides which the attendees could view online beforehand.
How did that work? I’ll talk about the scanning part in this post. This is probably old hat to many pathologists, but it’s been an interesting experience for me.
Context: Our department is looking into digital pathology with high-throughput whole slide imaging and storage and remote reporting, but we don’t currently have a system in place. Meanwhile, I’ve been getting used to a manual slide scanning application from Micropix that works with my microscope camera (also from Micropix). Essentially, the software takes the video feed coming through the camera and stitches it together to make a whole slide image. Pretty neat!
Here are my thoughts after using the software for a few weeks.
It’s straightforward to set up and calibrate. It’s easy to use, although being a manual system, the operator (ie, me) is responsible for ensuring the section stays in focus. It’s happy to scan at whatever magnification you have objective lenses for. Unsurprisingly, higher power takes longer (more passes). I wouldn’t want to go to x40 or x60 unless it was a really small specimen. I found myself using x20 for specimens such as skin biopsies and corneas. For globes, I used x10 or even x4.
Why so low power for globes? Firstly, they’re large specimens for us (say 25 mm), and the couple I tried to scan at x20 took me 45 minutes each (generating 500-600 MB files). I’m too impatient for that. Secondly, I use the scans for MDT demonstrations and teaching purposes. Most of the time, I want to demonstrate a (relatively) large lesion, not a subtle high-power cytological feature, so higher power doesn’t feel necessary. Thirdly, most globes have a big white space in the middle (ie the vitreous cavity). The software gets lost if you take it into empty space, and it finds it easier to stitch together lower power images.
What I LOVED was the ability to demonstrate a whole eye in the one image. Here are three images for comparison.
Standard photomicrograph. The lowest power objective lens I have on my microscope is a x1.25 (and I believe this is lower power than many other pathologists use). Here’s an image of a globe.
Well, part of a globe. We’re at the back of the eye, with the optic nerve, some retina, a large choroidal melanoma and a nodule of extrascleral extension. We can see only a part of the tumour, and we can’t see the anterior part of the eye at all. Of course I could take multiple images, but it’s not very convenient.
Quick and dirty solution. Take an actual photograph of the slide!
This rather blurred image shows an exenteration specimen with a tumour lying behind the globe. It gets the point across, I suppose. But boy, it’s ugly!
And finally, manual slide imaging. Here’s a screen shot of the scanned slide that I’ve sized to fit my screen.
It looks a bit pixelated, but that disappears when I zoom in. I’ll need to play around more with viewing software. But for MDT and teaching purposes, displaying the whole eye at once makes it much easier to correlate clinical and microscopic findings.
There are of course limitations. I didn’t get on well with (aspiration) cytology slides: too much space between cells. Impression cytology didn’t work well either. There, 3D sheets of corneal epithelium are stuck to a cellulose (or similar) membrane which is then mounted on to a slide and stained. With the constant changing of focus between fields, the software didn’t recognise which images belonged together.
A final interesting feature is making composites of more than one slide. For example, I could scan an H&E of a cornea and its corresponding PAS-stained slide. (Check one of my previous posts for matching H&E and PAS-stained corneas). Judging where to start subsequent scans proved tricky, but I’m sure it’ll come with time.
in summary, this certainly isn’t a competitor for a high-throughput system, but it’s a highly convenient and flexible alternative to taking multiple still images.