Expected Max Resolution at 40x in Validation/Test WSIs

Hi all,

We're extracting 8k tiles at 40x from WSIs (resizing to 4k at 20x due to the unavailability of 20x). What’s the highest expected resolution at 40x for validation/test WSIs? This knowledge will help us determine the maximum number of tiles that could come from a WSI, and hence to ensure the proper running of our algorithm in the given computational constraints at validation and test time.

Thanks!

Dear participant,

Thank you for your question. The test and validation image format, spacing, and resolution will fully resemble the training images. The 20x/40x notation can be confusing as it is scanner-specific, see the info below, thus we opt to use microns per pixel notation instead. The training, validation, and testing images are provided at a maximum resolution of ~0.25 microns per pixel (what is most widely referred to as 40x). The subsequent resolutions provided in the pyramidal tiff differ by a factor of 4. Thus each training, validation, and testing image has resolutions: 0.25, 1.0, 4.0, 16.0.... The intermediate resolutions (0.5, 2.0, 8.0) were removed from the images, to optimize the data storage/transfer/download. Please reach out in case of any further questions. Kind regards,

Khrystyna

Leica Aperio:

Scans at 20X, 40X or 63X (with oil immersion lens), have resolutions of 0.5, 0.25, 0.17 micron/pixel respectively.

https://med.virginia.edu/biomolecular-analysis-facility/services/shared-instrumentation/aperio-scanscope-slide-scanner/

Zeiss: 20x/0.8NA Plan-Apochromat (a=0.55 mm) 40x/0.95NA Plan-Apochromat (a=0.25 mm) https://www.helmholtz-munich.de/pathology-and-tissue-analytics/zeiss-axio-scanz1/index.html

Phillips: 40x 0.25 mpp https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8344580/

3DHistech P1000 40x 0.12 mpp https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8344580/

KFBIO

20x 0.25 mpp https://en.kfbio.cn/product/newProduct/subProduct/pro400/

Olympus

20x - 0.33mpp https://www.uib.no/en/rg/mic/120747/olympus-vs120-s6-slide-scanner

Akoya 40x = 0.25 μm/pixel https://www.akoyabio.com/phenoimager/instruments/

The resizing does not change the magnification. In fact, how do you know that 0.25 um/pixel is 40x? You don't... On many Aperio scanners, 20x is 0.25 um.px. It's also 0.5 um/px; there is no 40x, except in marketing brochures.