MoNuSeg: Multi-Organ Nuclei Segmentation in H&E Stained Tissue Images

MoNuSeg is a medical imaging dataset of H&E stained tissue images captured at 40x magnification, annotated for nuclei segmentation. The images were obtained from the TCGA archive, representing patients with tumors from different organs diagnosed at multiple hospitals. The dataset was created by RationAI and was last updated on HuggingFace in April 2025.

Use Cases

• Train nuclei segmentation models based on annotated tissue images.
• Develop algorithms for tumor assessment (grading, staging) based on H&E stained tissue features.
• Benchmark computer vision models for multi-organ cancer image analysis.
• Study tissue morphology and cell structure across different tumor types.

Strengths

• Images are carefully annotated for nuclei segmentation, as stated in the description.
• Data originates from the authoritative TCGA archive and multiple hospitals.
• Images are captured at a standard 40x magnification for histopathology.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count and dataset size are unknown, which may limit suitability assessment.
• Data may reflect geographic or institutional bias inherent to the source hospitals.

Provenance

Source

TCGA archive, multiple hospitals.

Collection Method

H&E stained tissue images were downloaded and carefully annotated.

Freshness

Last updated 2025-04-20 23:20:47; freshness should be verified.