The DRR-RATE dataset is built upon the recently released CT-RATE[1] dataset, which comprises 25,692 non-contrast chest CT volumes from 21,304 unique patients. Each study is accompanied by a corresponding radiology text report and binary labels for 18 pathology classes. The dataset has been expanded to 50,188 volumes through the modification of the reconstruction matrix extracted from the raw DICOM study. As the dataset was already anonymized, compliance with the Health Insurance Portability and Accountability Act (HIPAA) was ensured, and the requirement for informed consent was waived. CT-RATE is published under the Creative Commons Attribution- NonCommercial-ShareAlike (CC BY-NC-SA) license. In accordance with this license, we provide appropriate credit to the original creator, do not use the material for commercial purposes, and distribute any derivatives under the same license.
DRR Siddon-Jacobs Ray Tracing
To generate DRRs of the CT-RATE dataset, the binary tool ./getDRRSiddonJacobsRayTracing
[2,3] was utilized. Most parameters were kept at their default settings, with the exception of adjusting the volume by 300 mm along the y-axis to enhance the field of view. A threshold cutoff was also set at -100 Hounsfield units (HU). To create lateral view images, the volume was rotated 90 degrees counterclockwise around the z-axis.
user@machine:~$ ./getDRRSiddonJacobsRayTracing input_volume.nii.gz \
-o output_drr.png \
-threshold -100 \
-t 0 300 0 \
-rz -90 # if LATERAL view
References
[1] I. E. Hamamci, S. Er, F. Almas, A. G. Simsek, S. N. Esirgun, I. Dogan, M. F. Dasdelen,
B. Wittmann, E. Simsar, M. Simsar, et al., “A foundation model utilizing chest ct volumes
and radiology reports for supervised-level zero-shot detection of abnormalities,” arXiv
preprint arXiv:2403.17834, 2024.
[2] J. Wu, “Itk-based implementation of two-projection 2d/3d registration method with an
application in patient setup for external beam radiotherapy,” Virginia Commonwealth
University, 12 2010.
[3] B. Hou, “farrell236/midas-journal-784.”
https://github.com/farrell236/midas-journal-784, 2019.
Citation
@article{drr-rate,
title={Shadow and Light: Digitally Reconstructed Radiographs for Disease Classification},
author={Hou, Benjamin and Zhu, Qingqing and Mathai, Tejas Sudarshan and Jin, Qiao and Lu, Zhiyong and Summers, Ronald M},
journal={arXiv preprint arXiv:2406.03688},
year={2024}
}
- Downloads last month
- 68