HoVer-NeXt: A Fast Nuclei Segmentation and Classification Pipeline for Next Generation Histopathology - Datasets
Creators
-
1.
University of Bern
-
2.
Max Delbrück Center
-
3.
Humboldt-Universität zu Berlin
-
4.
Radboud University Medical Center
-
5.
Yale School of Medicine
Description
This repository contains training and validation data for
HoVer-NeXt: A Fast Nuclei Segmentation and Classification Pipeline for Next Generation Histopathology
Accepted for Oral Presentation at MIDL2024: https://openreview.net/pdf?id=3vmB43oqIO
More information and code are available at https://github.com/digitalpathologybern/hover_next_inference
Modified Lizard dataset to include mitosis (lizard_mitosis.zip), mitosis dataset (mitosis_ds.zip) and a holdout eosinophil validation set (eos_eval.zip)
mitosis_ds.zip also contains the hold-out H&E mitosis test set.
The original lizard dataset was createdy by Simon Graham et al. and was shared under CC BY-NC-SA 4.0. The tile-based dataset can be downloaded from https://conic-challenge.grand-challenge.org/Data/ after registering for the challenge. We modify the dataset by including an additional mitosis class, however note that there are a number of mitosis which are still not (correctly annotated).
Files
eos_val.zip
Additional details
Software
- Programming language
- Python
- Development Status
- Active
References
- Graham, S., Jahanifar, M., Azam, A.S., Nimir, M., Tsang, Y., Dodd, K.C., Hero, E., Sahota, H., Tank, A., Benes, K., Wahab, N., Minhas, F.A., Raza, S.E., Eldaly, H., Gopalakrishnan, K., Snead, D.R., & Rajpoot, N.M. (2021). Lizard: A Large-Scale Dataset for Colonic Nuclear Instance Segmentation and Classification. 2021 IEEE/CVF International Conference on Computer Vision Workshops (ICCVW), 684-693.