Protecting privacy of the patient is of utmost importance. Hence all data available in CHAVI is de-identified. In order to ensure that de-identified data is put in the CHAVI system, a bespoke de-identification system was developed. This de-identification system (CHAVI DDIS) was created to fulfil several objectives:
- De-identification of DICOM images
- Allow de-identification of DICOM RT data objects viz. RT Structure set, RT Dose, RT Plan while retaining the relationship between the objects.
- Allow de-identification of multiple studies while retaining longitudinal temporal integrity - that is, date differences are maintained between studies while dates are de-identified.
- Allow de-identification of linked clinical data while retaining the longitudinal temporal relationships. This clinical data is different from data encoded in the DICOM object and is stored separately in the database.
- Ensure that linkage between the original identifying information and de-identified data is maintained in a secure, local and separate database to allow re-identification of the data if required.
The de-identification system is installed in a separate computer and de-identified objects are transferred to the CHAVI data storage server. CHAVI cannot accept images or clinical data which have not been de-identified using the CHAVI DDIS.
To read more about the CHAVI DDIS please see this paper :
Kundu S, Chakraborty S, Chatterjee S, Das S, Achari RB, Mukhopadhyay J, et al. De-Identification of Radiomics Data Retaining Longitudinal Temporal Information. J Med Syst. 2020 Apr 2;44(5):99. http://dx.doi.org/10.1007/s10916-020-01563-0
For DICOM de-identification, we will be using a system for anonymization of DICOM images and the image databank will be created on a different web based system to store/retrieve the de-identified images. These two systems are isolated from each other. Users of the de-identification system will be only from the host institute while, the web based system (CHAVI-RO) which will be in public domain and will have also external users for accessing image bank data.
This system will de-identify the DICOM images. It accepts a single DICOM image or a DICOM series (.zip format) as an input from the external drive and provides de-identified DICOM images. The process flow of de-identification is shown in Figure 2.
The steps involved are detailed in our publication referenced above.