Lung adenocarcinoma growth pattern annotations for H&E images

This repository contains the training data, H&E tiles and associated annotations, used for the model developed in the paper The Artificial Intelligence-based model ANORAK improves histopathological grading of lung adenocarcinoma publised in Nature Cancer, doi: 10.1038/s43018-023-00694-w, togerther with raw data for reproducing figure 5b.

Table. Interpretation of the digital label for the training data

label
RGB color
class

0
(0, 0, 0)
others, including non-tissue

1
(0, 255, 255)
cribriform

2
(255, 0, 255)
micropapillary

3
(128, 0, 0)
solid

4
(255, 255, 0)
papillary

5
(255, 0, 0)
acinar

6
(0, 0, 255)
lepidic