This repository contains the code used to prepare the data in:

The LUMIERE Dataset: Longitudinal Glioblastoma MRI with Expert RANO Evaluation

Please cite the following publication when using this dataset:

Suter, Y., Knecht, U., Valenzuela, W., Notter, M., Hewer, E., Schucht, P., Wiest, R. and Reyes, M., 2022. The LUMIERE dataset: Longitudinal Glioblastoma MRI with expert RANO evaluation. Scientific data, 9(1), p.768. https://doi.org/10.1038/s41597-022-01881-7

The dataset can be downloaded from Figshare. Please note that the data is available for non-commercial use.

Overview

This repository contains technical details and code so other researchers can reproduce and adapt to their specific research questions.

Timepoint discretization

To add ambiguity for stronger anonymization but to retain the necessary time resolution, all timing information is provided relative to the first (pre-operative) MRI study date in weeks. The week was calculated by a floor division by 7 starting with the time span in days.

Automated segmentation

Segmentation was performed with DeepBraTumIA [1] and HD-GLIO-AUTO [2]. The run.py file of HD-GLIO-AUTO was slightly modified to retain the transform matrices to allow back-transformation to the original image space. No other changes were made. The adapted run.py file is available in the Segmentation/HD-GLIO-AUTO folder.

Skull-stripping

HD-BET [3] was used to remove the skull to ensure no conclusion is possible regarding the patient's head shape. Please note that while we provide skull-stripped MRIs, the input to the automated segmentation tools always was the original non-skull-stripped version.

Please note that both DeepBraTumIA and HD-GLIO-AUTO use HD-BET for skull-stripping, but results may differ to different pipelines (e.g., co-registration). We also provide skull-stripped MRIs for cases where not all four MR sequences were available.

Acknowledgement

This project was funded by Swiss Cancer Research (Krebsliga Schweiz), grant KFS-3979-08-2016. The NVIDIA Corporation donated a Titan Xp GPU.

The dataset was curated with great care. Please do not hesitate to contact us if you still find an error or inconsistency.

References

[1] DeepBraTumIA: https://www.nitrc.org/projects/deepbratumia

[2] HD-GLIO-AUTO: Please cite the publications listed on the author's repository https://github.com/NeuroAI-HD/HD-GLIO-AUTO

[3] HD-BET: https://github.com/MIC-DKFZ/HD-BET