EC-PD-L1: A Histopathology Dataset for PD-L1 Status Prediction in Esophageal Cancer

EC-PD-L1 is a histopathology image dataset for predicting PD-L1 status in esophageal cancer from hematoxylin and eosin (H&E)-stained images. The dataset was constructed from digitized pathological slides of esophageal cancer, with PD-L1 labels assigned by expert pathologists through review of paired H&E and corresponding PD-L1 immunohistochemistry (IHC) images. Cases with inadequate staining quality or tissue fragmentation were excluded during dataset construction to improve annotation reliability.
The dataset contains image patches from two categories: PD-L1-negative and PD-L1-positive. According to the current dataset summary, the dataset includes a total of 6,335 image patches, comprising 2,591 PD-L1-negative samples and 3,744 PD-L1-positive samples. This dataset is intended for the development and evaluation of artificial intelligence and computational pathology methods for biomarker prediction from routine H&E histology images.
This dataset may support research in digital pathology, biomarker prediction, medical image classification, and deep learning for cancer diagnosis and treatment decision support. We only uploaded 200 images for each category in the current release. For access to the full dataset, please contact [e-mail] and [e-mail].