Input Format

[a.n.d...@gmail.com]

Hi,

Is the submission supposed to accept the entire whole slide image as input, or can relevant patches from each whole slide image be expected?

[Eduardo Conde-Sousa]

Hi 

There is no problem to use just the "informative" part WSI.

Please let me know in the README file which kind of preprocessing should I apply (e.g. remove all the "white" background a keep just a rectangle of X square microns centered on the tissue).

all the best

Eduardo

******************************************************************************************

Eduardo Conde-Sousa

INEB – Instituto Nacional de Engenharia Biomédica

i3S – Instituto de Investigação e Inovação em Saúde, Universidade do Porto
Rua Alfredo Allen, 208 | 4200-135 Porto, Portugal

Telf: +351 220408815
www.bimage.ineb.up.pt

http://www.ineb.up.pt/

https://www.i3s.up.pt/scientific-platform?v=91
www.i3s.up.pt

******************************************************************************************

<a.n.d...@gmail.com> escreveu no dia quinta, 23/01/2020 à(s) 03:25:

Hi,

Is the submission supposed to accept the entire whole slide image as input, or can relevant patches from each whole slide image be expected?

--
You received this message because you are subscribed to the Google Groups "HEROHE Grand-Challenge" group.
To unsubscribe from this group and stop receiving emails from it, send an email to herohe-grand-chal...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/herohe-grand-challenge/2eab6b0e-b59a-4dd7-b145-f922294f3b26%40googlegroups.com.

[Eduardo Conde-Sousa]

Hi,

Just one more thing, please be precise in describing the region of interest, otherwise we may consider different regions. The same is valid for the area of the patch. E.g. in case 19 of the train dataset you have a tissue that covers almost all the area while in case 18 you have 4 or 5 smaller "blocks". In the cases where you have multiple blocks, how should they be treated, as one image containing all or as 4 or 5 different images (and in this case how it will be the classification of the WSI if the individuals differ?)?

As you can see, the room for ambiguity is huge. So if you can describe the region without ambiguity that would be just fine. Otherwise my recommendation is for you to apply an automatic pre-filter that removes unwanted regions.

******************************************************************************************

Eduardo Conde-Sousa

INEB – Instituto Nacional de Engenharia Biomédica

i3S – Instituto de Investigação e Inovação em Saúde, Universidade do Porto
Rua Alfredo Allen, 208 | 4200-135 Porto, Portugal

Telf: +351 220408815
www.bimage.ineb.up.pt

http://www.ineb.up.pt/

https://www.i3s.up.pt/scientific-platform?v=91
www.i3s.up.pt

******************************************************************************************

[Andreas Turzynski]

Just a practical comment on this topic: 3D Histotech‘s CNV is not able to convert much more than 50 square mm of the given WSI into a tif at full resolution.

I don‘t think the goal of the challenge was to hack the mirax format (or others) to make it possible to process WSIs in one step. This is homework for the manufacturers of the scanners. Those manufacturers suppling free, fast and reliable solutions for instantly slicing the WSIs to feed CNNs will finally survive.

All the best (...to the manufacturers)

Andreas

[Eduardo Conde-Sousa]

Hi Andreas

First of all. I'm not part of the manufactures company. The choice of this software and format was made independently of myself or this challenge.

I know that MIRAX is not the ideal for anyone who want to code over those images and I also have to deal with it daily basis.

Second, I agree with you that the goal is far from being "hack the MIRAX format". That's why from the very beginning I always recommended all the participants to convert to tif with slide converter (although some of you used different software, namely quPATH). I just didn’t do it from the beginning because I think that the decision on how the conversion should be made can influence the results and therefore it should be up to each participant. But if you use Slide Converter you can convert all the WSI in one step (you can also batch an entire folder) and from that moment on, you will only need to deal with tif files.

I can't understand your statement "3D Histotech‘s CNV is not able to convert much more than 50 square mm of the given WSI into a tif at full resolution". First of all, I'm assuming that when you wrote  3D Histotech‘s CNV your intention was to wrote " 3D Histotech‘s Case Viewer" (I don't know what CNV is). Case Viewer is not a converter, is just a viewer. The converter is Slide Converter, a separate program that is installed with case viewer. With slide converter your sentence is simply not truth.

All the best

Eduardo

******************************************************************************************

Eduardo Conde-Sousa

INEB – Instituto Nacional de Engenharia Biomédica

i3S – Instituto de Investigação e Inovação em Saúde, Universidade do Porto
Rua Alfredo Allen, 208 | 4200-135 Porto, Portugal

Telf: +351 220408815
www.bimage.ineb.up.pt

http://www.ineb.up.pt/

https://www.i3s.up.pt/scientific-platform?v=91
www.i3s.up.pt

******************************************************************************************

--

You received this message because you are subscribed to the Google Groups "HEROHE Grand-Challenge" group.
To unsubscribe from this group and stop receiving emails from it, send an email to herohe-grand-chal...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/herohe-grand-challenge/34b757c6-b1bf-4405-b672-4f54a86d2c2f%40googlegroups.com.