Taking images of the brain using magnetic resonance imaging, or MRI for short, also captures the face. For a long time, techniques that remove facial features from neuroimages have allowed for open sharing of anonymised neuroimages. But new developments in artificial intelligence and machine learning have made it possible to re-create the removed facial features and re-identify research participants. A recent study by researchers from within and outside the Human Brain Project explores what measures to take to continue to share neuroimages, but in a way that is safe and protects the data and privacy of the people involved.
Today, neuroscientists are faced with the open question of whether to develop better de-facing techniques (that remove more facial features and might damage the scientific utility of the image) or stop sharing neuroimages altogether. Developments in artificial intelligence and machine learning have turned what was previously considered fully anonymised MRI images into what now, at best, can be considered pseudonymised or de-identified images. Both insufficient for open sharing in the eyes of the GDPR. The authors of a paper recently published in NeuroImage: Reports provide a legal analysis of the measures to take to be able to continue sharing MRI images without reducing the scientific utility of the data, and while sufficiently protecting the privacy of everyone involved. Outlining technical and organisational measures to take, such as informed consent, pseudonymisation, access control, encryption and data use agreements. Creating a system that ensures data protection by design and by default.
“We want to provide the neuroimaging community with the opportunity to embrace this change and improve the way neuroimages are shared. Our analysis provides the tools needed to improve the quality of the data and the possibilities for the scientific community to make use of it. Ensuring that neuroimaging data can be shared in a way that retains its scientific value and at the same time protects the privacy of the data subjects or research participants. To increase the societal value derived from the data and encouraging people to keep participating in research. With the aim to increase and protect the overall societal acceptability of neuroscience,” says Damian Eke, researcher at De Montfort University, part of the Human Brain Project’s Data Governance Working Group, and one of the authors.
By Anna Holm
Damian Eke, Ida E.J. Aasebø, Simisola Akintoye, William Knight, Alexandros Karakasidis, Ezequiel Mikulan, Paschal Ochang, George Ogoh, Robert Oostenveld, Andrea Pigorini, Bernd Carsten Stahl, Tonya White, Lyuba Zehl, Pseudonymization of neuroimages and data protection: Increasing access to data while retaining scientific utility, Neuroimage: Reports, 2021 Volume 1, Issue 4, DOI: 10.1016/j.ynirp.2021.100053.