(how) can apply my own image?

[duernaT]

I noticed your paper (very nice work!) mention about HistoSegNet can generalize to other datasets without retraining. If I am not misunderstanding, I can load my own image and get the segment information?

If so, can you point out how to input my images?

With many thanks, Duerna

[lyndonchan]

Hello Duerna, yes, HistoSegNet can generalize to other datasets without retraining, assuming that:

• They are histopathology images using stains found in our dataset (e.g. H&E) and not fluorescent scans
• You rescale your images to fit the same microns/pixel resolution stated in our paper and then reshape to the 224x224 pixel resolution the model uses as input
• The organs present in your histopathology images are roughly the same as those mentioned in our paper

If those requirements are met, then you can drop your images into a new sub-folder of the "img" folder and set the "input_name" parameter of HistoSegNetV1 accordingly. You can check out "demo_01_segment_patches.ipynb" for an example.

[duernaT]

Thank you so much.

I have the HE-stain pictures that reshaped to 224x224 pixels. And then drop the images to "img" and set the HistoSegNetV1. However, I have tried several times, it not work with my images.

Here are they:

Can you please kindly point out how can I apply my img to the HistoSegNet?

Best regards, Duerna

[guascy666]

hi，i want to know that after I have got the segmentation result of each patch,how can I merge the patches to the origin size of the WSI