pingjunchen/livercancerseg

PAIP2019: Liver Cancer Segmentation

• Principal Data Scientist at UT MD Anderson Cancer Center
• Looking to collaborate on computational pathology

     

livercancerseg's Issues

Regions

Hi, first of all thanks for your work!

I am new with this type of data and is difficult for me. I have used https://github.com/wisepaip/paip2020/blob/master/xml2mask.py to take the masks on this year challenge and generated masks said that have for example 3 regions (https://github.com/wisepaip/paip2020/blob/master/xml2mask.py#L46). Using your library and examples for read WSI, as wsi_head.read_region((0, 0), DISPLAY_LEVEL, wsi_head.level_dimensions[DISPLAY_LEVEL]) I can load de wsi at desired level but I dont know what means 0,0 region... When I train my network I should use 0,0 region or? I load the mask as provided https://github.com/paip-2019/challenge/blob/master/utils.py#L9 but I no see nothing related to regions...

Best regards

how to change code to get 0 level patch?

Some questions about storing data.

Dear PingjunChen,
Thank you very much for sharing your work. I'm also taking part in this competition, I look forward to communicating with you more.
I have a question about storing dataa: I store my predictions in 'npy' or 'tif' format, which takes up a lot of storage space, about 2 gigabytes per image. Do you have good suggestions for storing pathological images？Here is my code to store the data.
Thanks again for your sharing and looking forward to your Suggestions.

svs file

Hi Pingjun,

Thank you for sharing this useful source code. However, when I try implementing python gen_patches.py file, it work till process7 and then stop processing and shows me no such a file in the directory. Can you help me with that please. Thanks again Pingjun, your help would be highly appreciated.

The Caffe pretrained PSPNet link seems to not work anymore

Hi,

Thank you for your work and for sharing it!

It seems like the link to the "Caffe pretrained PSPNet" is not working anymore. I am getting a 404 error when trying to view it.

I found a working link to what I understand to be the same pretrained model in the repository https://github.com/hszhao/PSPNet under "Usage", "3. Evaluation:" "pspnet101_cityscapes.caffemodel:". Here is the working (as for now) link: https://drive.google.com/file/d/1-Me01pX1KUsut_7l6uJXkIbaBxxSVgX0/view.

Best regards

I can`t find the code about merging patches into WSI

I run over the code about training. When I want to validate it, I find there is not code about patch and merge

How to open those tif files?

Thanks for sharing code. It's really helpful for a newcomer.

I can't open those viable or whole tif files by openslide, ASAP, opencv or default viewer. But I find that you use 'skimage.io.imread' function and it works.

I wonder why 'skimage.io.imread' function , rather than other tools, can read the tif image. Could you please explain it?