...
When sharing any kind of scientific data it is of great utility to provide as much information as possible to aid researchers in their understanding of that data. This is especially true with imaging since there are many important details that are required to perform common types of analysis which are not part of the DICOM header information. As a result TCIA strives to include this missing information on the associated wiki page for each Collection. What follows is a series of recommendations to data providers which will help them provide all of the context necessary to maximize the value of their data contribution to the TCIA research community.
...
We would like to thank the following members from the NCI Quantitative Imaging Network for the creation of their contributions to these recommendations:
- Matthew Oborski, University of Pittsburgh
- Charles Laymon, University of Pittsburgh
- Mark Muzi, University of Washington
- John Sunderland, University of Iowa
Clinical Data
Supporting clinical information such as patient demographics, treatment details, and outcomes can significantly increase the utility of datasets shared on TCIA. The NCI Genomic Data Commons has established a data dictionary viewer and templates which TCIA recommends as a good model for sharing clinical data in a standardized fashion.
General Description Information for Cancer Imaging Studies
...
- Prognostic blood/tissue biomarkers (e.g. histochemical assay of Ki-67, VEGF, HIF1a; promoter methylation status (MGMT); a characteristic chromosomal deletion/amplification; ER, HER2 or PR status).
- Molecular blood/tissue biomarkers that are predicative of therapy response (e.g., PSA blood levels)
- Example kinetic analysis of dynamic data set or parameter output file for the primary tumor and a control/normal region.
...