Artefact segmentation in digital pathology whole-slide images: pen markings binary masks

This dataset contains 51 binary segmentation masks for pen markings (blue, black, red, and green) observed on publicly available Whole Slide Images (WSIs) from The Cancer Genome Atlas (TCGA). The original WSIs are not included due to licensing constraints, but the corresponding file names are provided for reference in the correspondence file. Each binary mask (PNG format) highlights regions with visible pen markings, whether located on tissue or background. All annotations were manually created by a Master's student in Biomedical Engineering (non-expert in pathology) as part of a master's thesis on digital pathology quality control. The dataset is split into different folders: Train: 40 images Validation: 6 images Test: 5 images The TCGA images were extracted at x2.5 magnification. Use Case:The dataset is suitable for training and evaluating artefact detection models, especially in preprocessing pipelines aiming to improve the quality of computational pathology workflows.