Lung Cancer Segmentation

The model was trained on ACDC-HP challenge and got 3rd place in post-challenge leaderboard. It was further fine-tuned with data from the TCGA-LUAD and TCGA-LUSC projects.

The input images should be whole-slide images, with formats including 'svs', 'tif' etc. The images should have three channels (RGB) and a pixel spacing of about 0.5μm/pixel.