Open Bontempogianpaolo1 opened 1 year ago
@Bontempogianpaolo1
@binli123 shared the patches for TCGA TCGA multiscale patches (2.5x and 10x magnification). He said that the patches were extracted at (2.5x and 10x) and not (5x and 20x) as originally reported in the paper (see Issue#21).
I have not checked those patches yet, but I checked the patches produced for testing from the 6 slides in ./test/input (see https://github.com/binli123/dsmil-wsi#tcga-dataset for instructions for download and patching). I viewed the 6 slides using QPath and found the edge patches (they were easier to find manually) on the slide. They looked more like 10x than 20x.
@Bontempogianpaolo1
@binli123 shared the patches for TCGA TCGA multiscale patches (2.5x and 10x magnification). He said that the patches were extracted at (2.5x and 10x) and not (5x and 20x) as originally reported in the paper (see Issue#21).
I have not checked those patches yet, but I checked the patches produced for testing from the 6 slides in ./test/input (see https://github.com/binli123/dsmil-wsi#tcga-dataset for instructions for download and patching). I viewed the 6 slides using QPath and found the edge patches (they were easier to find manually) on the slide. They looked more like 10x than 20x.
I tried patching the slide at 10x, but I got 4.3 million patches in total, still mismatching the number provided in paper. Could you provide some guidance? Thanks!
Hi,
Could you upload a patch example with its slide name,x and y position for each magnitude? I m trying to replicate your numbers with even a threshold of 30 but I still obtain 30000 patches at 20x. Since it could be the magnitude I need a visual comparison thanks