XW2021 Round Table: XNAT in Translational Research

Time: Day 1, 12:15 – 12:45 pm CDT

Host: Dave Cash, University College London

Panelists: Pam LaMontagne (CIL), Mark White (National Health Service UK), Marcel Koek (Erasmus MC)

Questions and Answers

When possible and as time permitted, questions that were brought up in the Q&A module during each talk were addressed in real time by the presenter. Other responses were entered in the Q&A interface itself. Those written responses are included below.

Questions	Answers
Would be good to have a list of trusts/ boards in NHS in UK with instances of XNAT running. Also more sharing / curation of dicom anonymisation scripts that have been proven to be robust and the processes of how this was achieved. Would help me convince my information governance dept in NHS that running XNAT and using it as a vehicle for shipping data out is a safe thing to do	We agree! The Cancer Imaging Archive is one known example of well-published and documented anonymization techniques. See: https://wiki.cancerimagingarchive.net/display/Public/De-identification+Knowledge+Base
With growing prevalence of malware attacks etc is there a security protocol/procedure you follow for quarantining incoming data from external sources to minimize risk?	At Wash U, we do not allow incoming SCP traffic from outside the network. DICOM SCP is not a secure channel. In general, our dev ops team is very proactive in restricting network access to hosted XNAT sites based on their usage and the data risk.
Most of the discussion so far has been about pulling data from clinic to researchers. I see translation as also going from research towards the patient. E.g. using XNAT to facilitate clinical evaluation of tools, collect data for healthcare economics How do you think this (if deemed relevant) could be facilitated (safety, info compliance, r&d, etc)?	Answered live This is a wide-ranging topic worthy of future discussion. It jibes with some of the work at Radiologics / Flywheel in their hosted XNATs for clinical trials, and could also shape future XNAT ML work. It's worth noting that the TIP project does steer results directly to presurgical planning. See the poster in Spatial Chat for more.
Are there pipeline containers for running anonymisation checking? or plugins to implement these checks easily via the XNAT gui?	From Dave Maffitt: "The tools I think of for anon checking aren’t containerized. They could be but their use is problematic. A dicom validator, like dciodvfy, finds problems in just about every DICOM file so it’s output needs interpretation. WashU uses one tool that was written internally, called tagsniffer, to dump different views of all the tags extant in a dataset. This is primarily used to help write the anonymization script and not so much post-upload QA. It’s output needs a lot of interpretation. The output is extensive and isn’t suited for quick reporting nor has it been containerized. There is also some work being done on automatically looking for text burned-in to pixels but that is difficult becuase not all text in images is PHI. The short answer is none that I know of." From Dan Beasley: "I'm presenting a poster tomorrow on DASHER, including an anonymisation container that might be interesting. It contains many checks etc, face masking. DASHER is for for clinical trial protocols in mind etc"
Has anyone done process or other work using XNAT in Clinical Trials and changes are being made to protocols (for example, eligiblity, or CRF changes) that may impact what's in XNAT?