Dear organizers,
This is my first time working with WSIs data and I need to confirm a few things:
1. The provided data (TIFF files) are with **digital** magnification, am I correct?
2. The state "The segmentation mask for Task 2 should be at 20X magnification.", will that be 20X **digital** magnification or 20X **optical** magnification?
3. For the disease model "NEP25" described [here](https://sites.google.com/view/kpis2024/data?authuser=0), training data has **20X optical** magnification, **40X digital** magnification but validation data has **40X optical** magnification, **40X digital** magnification. Could you check it?
4. In [here](https://sites.google.com/view/kpis2024/evaluation?authuser=0), the description of "Output structure" for tasks 1 & 2, should it be "output_dir" instead of "input_dir"?
5. Do all images in the **test set** for Task 1 (patch-level) have the same filename format as in the validation and training data (e.g., `12-174_40_13312_1024_img.jpg`)?
Created by Capybara capybara @capybara
Hi,
We just figured this out, sorry for the confusion. The output mask should be at 40X **digital** magnification. We have edited the evaluation section and the previous answer. With the validation data, the segmentation of WSI with 80X digital magnification should reduce the resolution by a factor of 2. The segmentation of WSI with 40X digital magnification should kept the same.
Best,
KPIs organization committee @huoy1 @Reyn4Bo
Hi, sorry for asking multiple times!
Looking at the metadata of the disease model `NEP25` [here](https://sites.google.com/view/kpis2024/data?authuser=0), I still doubt that its optical magnification in the validation data is 20x (not 40x). Moreover, when I checked the validation code in docker `hrlblab333/kpis:validation_slide`, it down-sampled all disease models except for `NEP25` (see the code below)
```python
if get_X20_wsi:
'read wsi and get 20X and patches'
for img in images:
if 'NEP25' in img:
lv = 1
else:
lv = 2
now_tiff = tifffile.imread(img,key=0)
tiff_X20 = ndi.zoom(now_tiff, (1/lv, 1/lv, 1), order=1)
```
Could you confirm again if the optical magnification for `NEP25` in the validation data is actually 40x?
Hi @Reyn4Bo
Thanks for the clarification, everything is clear now!
Just a minor problem,
> 4. We have fixed this issue, it should be output_dir
I still see the text `input_dir` under the "Output structure" for tasks 1 & 2 on the website: https://sites.google.com/view/kpis2024/evaluation (but of course we know that this is `output_dir`) Hi! @TigerTeam
We have fixed the typo issue. The file structure at https://sites.google.com/view/kpis2024/evaluation should be the output file structure. For the input file structure, please refer to [Github](https://github.com/hrlblab/KPIs2024).
Best,
KPIs organization committee Hi! @capybara
Sorry for the late reply.
Here's the response to the queries:
1. The provided data are with digital magnification.
2. The segmentation mask for Task 2 should be at 40X **digital** magnification. (The segmentation of WSI with 80X digital magnification should reduce the resolution by a factor of 2. The segmentation of WSI with 40X digital magnification should kept the same.)
3. The information is correct. The testing data will distributed in the same way as the validation data. (40X optical magnification, 40X digital magnification)
4. We have fixed this issue, it should be output_dir
5. Yes, they shall have the same filename format and data organization.
Best,
KPIs organization committee @huoy1
Can you or someone please respond when you have time?
Thanks Dear Organizers
Along the same lines will the input directory pattern be same as provided in the validation format as provided below? or from the https://sites.google.com/view/kpis2024/evaluation input directory format won't have img, mask subdirectories?
input_dir
??? 56NX
??? case1
??? img
-patch1_mask.jpg
??? mask
patch2_mask.jpg
??? case2
??? case3
??? DN
??? NEP25
??? normal
Drop files to upload
Confirming output magnification and validation data page is loading…