De-identification of DICOM dates
De-identification of dates for this dataset uses the DICOM Part 3.15 Annex E standard “Retain Longitudinal With Modified Dates Option” which allows dates to be retained as long as they are modified from the original date. TCIA implements this using a technique which de-identifies the dates while preserving the longitudinal relationship between them. Original dates will be first normalized to January 1, 1960 and then offset relative to the date of registration for each patient. This normalized date system was chosen in order to make it obvious that the dates are not real, and to make it easy to quickly determine how much time has passed between the date of registration and the patients' related imaging studies.
For example, if the real date of a patient's registration was 03/27/2018 and the original imaging Study Date was 03/29/2018 then the anonymized TCIA Study Date would become 01/03/1960 (two days after the base date of 1/1/1960).
Insertion of computed "REGISTRATION"/Days offset from registration" value
In addition to modifying the actual date fields in the DICOM header, the "days from registration" values are calculated and stored in the DICOM tag (0012,0052) Longitudinal Temporal Offset from Event with the associated tag (0012,0053) Longitudinal Temporal Event Type set to “REGISTRATION”. Here is an example DICOM header from a scan where the patient's imaging was performed 2 days before the registration, resulting in a negative offset value.
|(0012,0052)||Longitudinal Temporal Offset from Event||-2.0|
|(0012,0053)||Longitudinal Temporal Event Type||REGISTRATION|
If you would like to filter your search results using this information you can leverage the "Clinical Trial Time Points" filter shown below via our data portal at https://nbia.cancerimagingarchive.net/nbia-search/.