Table 1.

Summary of publicly available datasets containing LN tissue sections.

Dataset	Summary	Ground truth	Sample size	Dimensionality	Suitable applications
Camelyon16	H&E stained slides of breast cancer sentinel lymph nodes from 2 centers	Patient-level labels: pN stage. Pixel-level: partial tumor annotations	399 WSIs	WSIs scanned at 20x and 40x magnifications	Segmentation, slide-level classification, MIL
Camelyon17	H&E stained slides of breast cancer sentinel lymph nodes from 5 centers	Patient-level pN stage label Pixel-level: partial tumor annotations	1000 WSIs	WSIs scanned at 20x and 40x	Segmentation, classification, MIL
Patch Camelyon	Patches extracted from Camelyon16 dataset	Patch-level label: tumor positive or negative	327,680 patches	96x 95 pixel patches at 10x	Patch classification
SLN-Breast, TCIA	H&E stained slides scanned in one center	Patient-level labels: pN stage label	130 slides from 78 patients	WSIs scanned at 20x	Slide-level classification, MIL
Axillary lymph nodes in breast cancer, AIDA	H&E slides of breast cancer sentinel lymph nodes (2 consecutive slices) from multiple scanners	Patient-level labels: treatment, pN stage, primary tumor. IHC slides for cytokeratin AE1/AE3	396 patients	WSIs scanned at 20x	Tumor segmentation, slide-level classification, MIL
Breast cancer lymph nodes for HEV detection	Immunofluorescently stained HEVs using HEV markers for peripheral node addressin PNAd and vascular marker Claudin-5	Immunofluorescence signal for HEVs	73 patients with invasive, non-invasive breast carcinoma, or cancer-free	Not specified	HEV segmentation and diameter evaluation in LNs
Multi-magnification organ network (MMO-Net) dataset	H&E stained slides of rat mandibular lymph nodes from 3 centers	Manual LN contour annotations	53 WSIs containing lymph nodes, 267 WSIs from other organs	Not specified	LN contour segmentation