Image and label patches used to train GLASS-AI

This archive contains the paired image and label patches used to train our machine learning pipeline, Grading of Lung Adenocarcinoma with Simultaneous Segmentation by Artificial Intelligence (GLASS-AI). Image patches were generated from whole slide images of H&E-stained sections using an Aperio ScanScope AT2 Slide Scanner (Leica) at 20x magnification with a 0.5022 microns/pixel resolution. The individual tumors and airways were annotated by an expert human before being divided into 224x224 pixel patches of the H&E image and paired annotation layer. For more details regarding how these data were used to train GLASS-AI, please see our forthcoming manuscript.