Published October 27, 2025
| Version v1
Dataset
Open
KidneyAI Dataset
Authors/Creators
- 1. Pathology, Clinical Pharmacology and Safety Sciences, R&D AstraZeneca, Gothenburg, Sweden
-
2.
KTH Royal Institute of Technology
-
3.
Science for Life Laboratory
- 4. Bioscience Technology, Research and Early Development, Cardiovascular, Renal and Metabolism (CVRM), BioPharmaceuticals R&D, AstraZeneca, Gothenburg, Sweden
- 5. Bioscience Renal, Research and Early Development, Cardiovascular, Renal and Metabolism (CVRM), BioPharmaceuticals R&D, AstraZeneca, Gothenburg, Sweden
- 6. Bioscience Renal, Research and Early Development, Cardiovascular, Renal and Metabolism (CVRM), BioPharmaceuticals R&D, AstraZeneca, Cambridge, UK
-
7.
Karolinska Institutet
- 8. Department of Clinical Pathology, Sahlgrenska University Hospital, Region Västra Götaland, Gothenburg, Sweden
- 9. Department of Laboratory Medicine, Institute of Biomedicine, Sahlgrenska Academy, University of Gothenburg, Sweden
- 10. Ribocure Pharmaceuticals AB/Ribo Life Science Co Ltd, Gothenburg, Sweden
Description
Streamlining the Histopathologic Workflow in Diabetic Kidney Disease with Artificial Intelligence
Summary
A curated set of 3,928 PAS-stained mouse glomerulus images extracted from 30 whole slide images (WSIs) in a BTBR ob/ob diabetic nephropathy study. Each glomerulus is center-cropped and annotated independently by three expert kidney pathologists using a 5-class scheme: Normal, Mild, Moderate, Severe, and Excluded (uninformative).
Contents
- Images: 3,928 cropped glomerulus patches (PAS stain), sourced from 30 WSIs.
- Labels: Triplicate expert scores per glomerulus (Experts 1–3) plus metadata for slide/study origin.
- Schema: Class labels {0,1,2,3,Excluded}; optional split files for cross-validation as used in the manuscript.
- File formats:
- Images: PNG
- Annotations and metadata: JSON
Citation
Please cite our paper “Streamlining the Histopathological Workflow in Diabetic Kidney Disease with Artificial Intelligence” when using this dataset.