Datasets:

ArXiv:
License:

MPP=1000, set to MPP=0.5 (20x) manually?

#8
by lely475 - opened

Dear HISTAI team,

When opening the CRC_b1 & CRC_b2 slides with openslide the MPP properties are set to 1000 um/px:

slide = openslide.OpenSlide(fp)
mpp_x = int(slide.properties[openslide.PROPERTY_NAME_MPP_X]) # = 1000 um/px
mpp_y = int(slide.properties[openslide.PROPERTY_NAME_MPP_Y]) # = 1000 um/px

This gets confirmed by the tiff resolution setting:

tiff.ResolutionUnit': 'centimeter'
'tiff.XResolution': '10'
10 pixels per centimeter → 1 pixel = 0.1 cm = 1000 µm

But 1000 um/px logically does not make sense, as it is way to high and in the dataset description it is mentioned that WSIs were scanned at 20x, except if noted differently in the filename.

Most of the slides are stained with Hematoxylin and Eosin (H&E) and scanned at 20X magnification. If a slide differs in magnification from 20X, this information is embedded in the slide filename, as shown above.

So now I am wondering, whether it is correct to manually set the MPP=0.5 for 20x (or 0.25 MPP for 40x)?
Usually scanners do not scan at exactly 0.5 MPP, but can slightly vary (e.g. 0.4954), so I want to confirm that a manual setting to 0.5 is indeed the correct MPP for your slides?

Thank you already.
Best,
Lydia

HistAI org

Hi Lydia,
Yeah, 0.25 / 0.5 is a good approximation in this case.

Alex

Hi Alex,
Thank you for the feedback. Do I understand correctly that the "true" MPP varies slightly from 0.25/ 0.5, but is not available with the slides at the moment? So 0.25/ 0.5 is the closest approximation we can make?
Best,
Lydia

HistAI org

Exactly.

Alex

Thank you for clarifying!

lely475 changed discussion status to closed

Sign up or log in to comment