Automated Standardsbased Anonymization Profile for Image Sharing Using

Automated Standards-based Anonymization Profile for Image Sharing Using RSNA’s Clinical Trial Processor Justin Kirby 1, John Perry 2, Carl Jaffe 1, John Freymann 1 SAIC-Frederick 1, Radiological Society of North America 2, Abstract About DICOM WG 18’s Supplement 142 Overview: Image Submission Workflow Handling Profiles in CTP DICOM Working Group 18 drafted Supplement 142 in an effort to provide expert reviewed guidelines which protect patient confidentiality for personnel who are responsible for the management of DICOM images. The document also seeks to accomplish this task in a way that ensures certain specific information needed for quality control and image analysis is not lost. In the CTP/NBIA system, a CTP program runs at the client site. Images are imported into CTP via a DICOM transfer from a PACS or from a local hard drive. CTP de-identifies the images and exports them to a CTP program running at the NBIA site. Typically this is a transfer across the internet, encrypted in SSL. The CTP server at the NBIA site stores the images and exports many of their DICOM elements to the NBIA database. NBIA administrators then review the data for QC purposes. Below is a visual diagram of this process: CTP now provides support for multiple Supplement 142 de-identification profiles. The Basic Profile is the default, and the Retain UIDs Profile has recently been added. Additional profiles will be added in the future. Profiles are simply selected through the Profile menu in the CTP DICOM Anonymizer Configurator. The guidelines are provided via a coded set of instructions. The full list of codes are copied below: PURPOSE Clinical trial patient images in DICOM format are increasingly being shared by cross disciplinary research teams where some key DICOM tags need to be retained but HIPAA compliant. A versatile, template-based, software tool has been developed which respects HIPAA guidelines while retaining analytically-critical selectable DICOM tags. METHOD AND MATERIALS HIPAA and IRB compliance is a major obstacle to clinical image sharing for central data analysis and research. A key component to achieving compliance is proper deidentification of images which will be used in research. The goal of this activity was to provide easy to use functionality to perform a standards-based safe removal of PHI. A team was assembled to review DICOM Supplement 142 with a focus on generating a template which follows the de-identification recommendations for the “Baseline Profile”. The “Baseline Profile” is the most aggressive and secure. A detailed review was performed to compare each DICOM tag listed in Supplement 142 and its recommended method for de-identification to see if there was a corresponding function available in the CTP script language. If necessary, new functions were created to meet these recommendations. New functionality was added to the RSNA Clinical Trial Processor to manage the standard profile templates and derivatives. The tool incorporates a new audit trail of the de-identification history for provenance assurance. Funded by NCI Contract No. HHSN 261200800001 E. RESULTS The RSNA Clinical Trials Processor software was extended to include the ability to create and manage deidentification profiles. The default profile provided with the application will de-identify images according to Dicom Working Group 18’s Supplement 142 “Basic Profile” guidelines. CONCLUSION The Clinical Trials Processor in combination with the National Biomedical Imaging Archive have been improved to allow standards based de-identification and storage of images. TEMPLATE DESIGN © 2008 www. Poster. Presentations. com CTP Client PACS/HDD HTTP/DICOM Import NCIA System Verify QC Review The Supplement reviews the standard DICOM tags and provides recommendations on how to de-identify the information they contain. To meet a variety of de-identification requirements, the supplement introduces the concept of "profiles". The “Basic Profile” is the most aggressive method. This is the starting point for de-identifying anything that might eventually be visible to the general public, such as in the case of NCI’s hosted NBIA application. Additionally, the supplement provides several profiles which relax certain de-identification requirements, retaining UIDs, retaining device information, etc. These may be used in circumstances where absolute confidentiality is not required, as when sharing images internally or when sharing between sites who have drafted limited data use agreements, etc. Submission De. Identification CTP Database Export HTTP Export Server In addition to the pre-defined profiles based on Supplement 142, User-defined profiles can also be created and saved. With this new functionality, a user of CTP can now load and customize any number of profiles. File Storage The de-identification step at the client site leverages the Supplement 142 Basic Profile. This ensures that images are fully de-identified before they leave the network of the organization where they were created. Implementing the “Basic Profile” in CTP Implementing the Basic Profile required creating a CTP anonymization script to implement the deidentification recommendations for each DICOM element. This in turn required a thorough review of the Supplement 142 document, selecting the appropriate CTP script function for each of the coded recommendations, entering the scripts into CTP’s DICOM Anonymizer Configurator tool, and saving the result as a profile. Future Plans We plan to create additional profiles which leverage the additional Supplement 142 options, including Retain Device Info, Retain Safe Private Tags, Retain Patient Chars, and Retain Long Info. The Supplement 142 profiles and their implementations in CTP provide clinical trial administrators access to the expertise of the DICOM Committee as well as experience of the NCI when configuring their trials. Contact information If you have any questions regarding the content of this poster please contact me to discuss: Justin Kirby T: 301. 496. 9532 F: 301. 480. 4631 E: kirbyju@mail. nih. gov