Published: 09JAN2020 Version: 1.0.0
In daily clinical practice, pathologists need to spend amount time manually searching each glomerulus on the pathology slide images for the further measurement of clinical index, description of tissues, and diagnosis. The project aims of this database was to develop of an artificial intelligence algorithm to auto-segmentation of glomerulus base on the whole slide image (WSI). The algorithm will decrease the clinical burden for pathologists.
The original database contains 4,524 high resolution WSI from two mice’s kidney, one SLE mouse and another mouse without SLE. All 40X scanned images were produced and scanned from whole microscope slide using HAMAMATSU NanoZoomer S360. The whole microscope slides with hematoxylin-eosin staining (H&E) were produced by the specialized staffs from department of Tissue Bank in Chang Gung Memorial Hospital. The size range for each WSI from 1 GB up to 5GB. Some of WSI with annotations for the glomerulus. One WSI may have hundreds of annotations due to the one tissue continue lots of glomerulus.
Range for Image size: (46,000 ~ 80,000) x (57,000 ~ 65,000)
Images type: .ndpi
Mouse’s Kidney (Original file 46,080 x 59,904)
Mouse’s Kidney (Original file x 1.85)
Mouse’s Kidney (Original file x 1.85)
Labeling Process
Send research proposal to cgmhailab@gmail.com (free format)
None
TBA