UD Ureter-Uterine Artery-Nerve Dataset

Citation Author(s):
University of Debrecen
Submitted by:
Norbert Serban
Last updated:
Tue, 07/11/2023 - 01:54
0 ratings - Please login to submit your rating.


The UD Ureter-Uterine Artery-Nerve Dataset contains 586 images captured from 38 laparoscopic surgeries with corresponding masks for binary and multiclass semantic segmentation annotated by gynecological experts from the University of Debrecen. The aim of this dataset is to provide image data for automatic organ segmentation and differentiation during laparoscopic hysterectomy. Our dataset can be used to train semantic segmentation models to provide pixel-wise detection of the three labelled organs: ureter, uterine artery and nerves or train them to distinguish the ureter from the other organs since this is a main concern and challenge when the surgeons perform the laparoscopic hysterectomy.



File structure:

UD Ureter-Uterine Artery-Nerve Dataset
|-- images
|-- mask
|-- multi_mask

The 'images' folder contains the 586 high-resolution RGB images of the dataset.

The 'mask' folder contains the decomposed, binary mask images for each RGB image. The file name of the mask images uses the following rule: {RGB image name}_{ureter | arthery | nerve}.png.

The 'multi_mask' folder has the multi-labelled masks of the RGB images with the following labels: 0- background, 1 - uterine artery, 2 - ureter, 3 - nerve