01 December 2016
Share this article

XNAT: A big step forward for population imaging

Professor Aad van der Lugt is radiologist with a focus on Neuroradiology and Head-Neck radiology at the Erasmus MC. He spends a lot of his time doing research with imaging data. Van der Lugt uses and promotes XNAT, that was implemented and provided by TraIT, part of BBMRI-NL 2.0.

What is your experience with BBMRI-NL?

For years I have been involved in large cohort studies that use imaging data, such as the Rotterdam Study or Generation R study. We call this population imaging: imaging not used for diagnostics, but as a measuring instrument in scientific research. I got to know BBMRI-NL 2.0 because this project aims to build a national infrastructure for biobanking. In the past this infrastructure focused on biomaterials, blood and clinical data. However, the last 10 years and especially under BBMRI-NL 2.0 (including TraIT) there has been growing attention for imaging data and image analysis.

What is XNAT?

XNAT is an open-source database for medical images. Normally imaging data is stored into archive systems of hospitals. To use these data for scientific research asks for a completely different way of storage. Data needs to be anonymous, easily accessible for researchers and shareable. The classical archives of hospitals do not meet these requirements. Within BBRMI-NL 2.0, TraIT implemented an open-source database, called XNAT (which was developed by Washington University, St Louis), and made it available for all Dutch researchers. XNAT within BBMRI 2.0 functions as an overarching multicentre database to which all research centres can be connected.

How does XNAT work?

A request for use can be submitted via the service desk. Within a week you can be up and running. Investigators from research centres can send anonymized imaging data to the XNAT database, and can make the images accessible to other researchers. A big step forward, as in the past images were only available and accessible within the centre where they were stored. With XNAT the data is safely stored and the principal investigator that collected the data remains the owner. Researchers that want to use the XNAT data apply for access from the principal investigator via the XNAT system. Once access has been granted they can download the data from the cloud.

Why do we need XNAT?

There is a growing call upon researchers to save their data for future research and to make it available for others.  XNAT offers the infrastructure to do so. It facilitates the image analysis by automated pipelines and is a structured way to safely store anonymized data and make it available for other researchers. Next to that XNAT has proven to be very useful for multicentre studies. All centres can directly upload their images. Access rights are based on prior agreements and regulations. In the past the data were stored in one of the centres, which could be a barrier to access and a source of frustration among the other participants in the study.

Is the XNAT tool a success?

We see that XNAT is used more and more, and we see that it works efficiently. We are now in the phase where we have images from multiple studies in the database  The number of requests for data are growing and the data is actively used. We also see that research centres further develop and customize  the database. This proves that XNAT fulfils the needs and wishes of researchers.

What makes this tool unique?

XNAT is not only an imaging data infrastructure solution. What makes XNAT unique, is that it is developed and maintained by an active community to which researchers can contribute. It has become popular because it is open-source. This not only means that it is free, but it can be further developed by the users. Scientists do not like instant solutions, they want to think along and build it towards their own wishes and needs. With XNAT they can do so.

What will be the future steps?

TraIT has made this imaging database available. Within BBMRI-NL 2.0 there are several storage solutions for other types of non-imaging data. For the future we need to focus on making more connections and links between the different data infrastructures and use this in research practice.