6,335 H&E-stained histopathology image patches for predicting PD-L1 status in esophageal cancer, comprising 2,591 PD-L1-negative and 3,744 PD-L1-positive samples. The dataset was constructed by the MIaMIA Group from digitized pathological slides, with labels assigned by expert pathologists. It was last updated on 2026-04 12.
Use Cases
- Training deep learning models for PD-L1 status classification based on H&E-stained image patches.
- Developing computational pathology methods for biomarker prediction from routine histology images.
- Evaluating AI model performance for cancer diagnosis and treatment decision support based on histopathology data.
- Researching the relationship between H&E morphology and PD-L1 expression in esophageal cancer.
Strengths
- 6,335 labeled image patches provide a defined scale for model development.
- Labels were assigned by expert pathologists through review of paired H&E and PD-L1 IHC images.
- Cases with inadequate staining quality or tissue fragmentation were excluded to improve annotation reliability.
Limitations
- The current public release contains only 200 images per category (400 total), with the full dataset requiring contact.
- Row count for the full dataset is unknown, which may limit suitability assessment.
- Column-level documentation is absent; field semantics must be inferred after download.
Provenance
- Source
- MIaMIA Group, uploaded to figshare.
- Collection Method
- Constructed from digitized pathological slides of esophageal cancer, with PD-L1 labels assigned by expert pathologists.
- Time Range
- null
- Freshness
- Last updated 2026-04-12 12:09:52; freshness should be verified.
- Geography
- null