how to calculate Percentage of IHC+ cell in WSI

[WeiLiuY]

Dear DeepLIIF author,

I am trying to use DeepLIIF to do segmentation of IHC + cells on our WSIs.
In the results output folder, i have multiple tiffs plus a results.pickle file.
I am wondering if there is any way to calculate the % of IHC+ from my WSI.
Or should i calculate from the results.pickle?
Many thanks!

Best,
Wei

[WeiLiuY]

Hi, i have figured out by myself. Thanks.

[paulrbuckley-kcl]

Hi, also wondering this. I am trying to use ski-image to extract the cellular data but quite coplex thus far so assuming I am doing something wrong .. Wondering if anyone could provide some assistance please

[jommarin]

Are you using the infer_results_for_wsi function (either directly or through a call from cli.py)?

https://github.com/nadeemlab/DeepLIIF/blob/42c7ce4558d790832edce5e98e541b05fbc8e137/deepliif/models/init.py#L641

If so, I see that the counts for each patch are generated, but they are just discarded, and only the generated images are saved. You could add code here to keep a tally of the positive and negative counts (region_scoring['num_pos'] and region_scoring['num_neg']), and then calculate the final percentage at the end.

I am not sure why it was written this way (without accumulating and saving the counts), but I will try to update the repo in the next few days to correct this oversight.

[paulrbuckley-kcl]

Thanks a lot, yes I am using that function. I will have a fiddle with the code to keep a tally as you suggest thank you. Possibly a different question so can open another issue if necessary, but I’d like to also extract single cell information (ie pve nve cell coordinates, shape etc) from the segmented mask. I’m assuming the pickle file is the best way to do it? wondering if you wouldn’t mind recommending a way of doing this .. currently I was looking at RBG based thresholding but wanted to ensure that was the most appropriate way .. Paul From: Joseph Marino ***@***.***> Date: Wednesday, 5 February 2025 at 21:47 To: nadeemlab/DeepLIIF ***@***.***> Cc: Paul Buckley ***@***.***>, Comment ***@***.***> Subject: Re: [nadeemlab/DeepLIIF] how to calculate Percentage of IHC+ cell in WSI (Issue #19) Are you using the infer_results_for_wsi function (either directly or through a call from cli.py)? https://github.com/nadeemlab/DeepLIIF/blob/42c7ce4558d790832edce5e98e541b05fbc8e137/deepliif/models/init.py#L641<https://github.com/nadeemlab/DeepLIIF/blob/42c7ce4558d790832edce5e98e541b05fbc8e137/deepliif/models/__init__.py#L641> If so, I see that the counts for each patch are generated, but they are just discarded, and only the generated images are saved. You could add code here to keep a tally of the positive and negative counts (region_scoring['num_pos'] and region_scoring['num_neg']), and then calculate the final percentage at the end. I am not sure why it was written this way (without accumulating and saving the counts), but I will try to update the repo in the next few days to correct this oversight. — Reply to this email directly, view it on GitHub<#19 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/A4CGWGD4BRC6CHTA4EL7X732OKBIBAVCNFSM6AAAAABWQXEDX6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDMMZYGA4TIMZXGE>. You are receiving this because you commented.Message ID: ***@***.***>

[sanadeem]

@paulrbuckley-kcl can you please open a separate issue for the following point and we will keep this open to resolve the point you originally raised.

Possibly a different question so can open another issue if necessary, but I’d like to also extract single cell information (ie pve nve cell coordinates, shape etc) from the segmented mask. I’m assuming the pickle file is the best way to do it? wondering if you wouldn’t mind recommending a way of doing this .. currently I was looking at RBG based thresholding but wanted to ensure that was the most appropriate way ..

Paul