High resolution annotated HiP-CT training dataset, models and predictions for paper - Multiscale Segmentation using Hierarchical Phase-contrast Tomography and Deep Learning

This dataset is associated with the publication: Multiscale Segmentation using Hierarchical Phase-contrast Tomography and Deep Learning. Corresponding author: Claire L. Walsh [e-mail] Codes available at https://github.com/UCL-MSM-Bio/2025-zhou-hipct-hierarchical-segmentation. The contents in the repository are: highres_training_16bit_cubes.zip: the original high-resolution HiP-CT cubes of size 512x512x512, 16 bits; highres_training_labels.zip: the manual annotations linked to the training cubes; Manual_annotated_high_resolution_data_spreadsheet.xlsx: details of each cube. In the latest version: Manual_annotated_high_resolution_data_spreadsheet.xlsx: the x, y coordinates where the training cube was extracted were added; nnunet_models.zip: the best fold of models from each resolution and samples are provided; predictions.zip: the predictions after post-processing for each resolution; cortex_masks.zip: cortical masks for two kidneys in the manuscript. If you use this dataset, please cite the dataset and the following papers: @article{zhou2025multiscale, title={Multiscale Segmentation using Hierarchical Phase-contrast Tomography and Deep Learning}, author={Zhou, Yang and Aslani, Shahab and Javanmardi, Yousef and Brunet, Joseph and Stansby, David and Carroll, Saskia and Bellier, Alexandre and Ackermann, Maximilian and Tafforeau, Paul and Lee, Peter D and others}, journal={bioRxiv}, pages={2025--05}, year={2025}, publisher={Cold Spring Harbor Laboratory} }@article{walsh2021imaging, title={Imaging intact human organs with local resolution of cellular structures using hierarchical phase-contrast tomography}, author={Walsh, Claire L and Tafforeau, P and Wagner, WL and Jafree, DJ and Bellier, A and Werlein, C and K{\"u}hnel, MP and Boller, E and Walker-Samuel, S and Robertus, JL and others}, journal={Nature methods}, volume={18}, number={12}, pages={1532--1541}, year={2021}, publisher={Nature Publishing Group US New York} }Please also Acknowledge the beamtimes md1252, md1290, and md1389 as sources of the data.