Your data. Your questions.

The main points

• Sharing can make research more useful. Other researchers can check findings and use existing data to answer new questions.
• Removing a name is one part of protecting privacy. Images and information collected alongside them may also need attention.
• The details of your study matter. Ask what will be shared, who will have access and what protections will apply. 1, 2

What information is involved?

A study may collect brain scans or recordings alongside information such as age, health history and questionnaire answers. Researchers use this accompanying information to understand differences between people and interpret their results. Privacy protection therefore needs to consider these pieces together. 1

Illustrative example: an age and a broad location might seem unremarkable separately. If only one person in a study has that combination, someone who already knows the person took part could potentially recognise their record. This is why researchers inspect combinations of details, not just names.

Could someone recognise you from a scan?

Some head scans include the shape of the face. Studies have shown that facial reconstructions from MRI can be matched to photographs. Researchers may remove facial features or use access restrictions, depending on the data and the study's arrangements. 2, 3

Brain measurements can also contain patterns that help researchers recognise the same participant across two recordings. That is not the same as discovering the person's name. Learning an identity usually needs another connection, such as a photograph, an identity-linked record or prior knowledge about the person. 2, 4

These studies describe particular methods and conditions. They do not tell us that every brain scan can identify its participant, or that any one protection removes every possible risk.

What did the metadata study show?

The metaprivBIDS study examined information accompanying selected datasets shared through OpenNeuro. Serious potential vulnerabilities were uncommon, but the researchers found opportunities to improve privacy. They assessed possible routes to identification; they did not try to find out who the participants were. 1

Age and combinations of demographic details needed particular care. Health questionnaire scores contributed less to identification in the datasets studied, but those answers can still be sensitive if connected to a person. A low identification risk does not mean the information is unimportant or needs no protection. 1

How can researchers reduce the risk?

Possible measures include keeping names and contact details separate, reducing unnecessary detail in shared tables, removing facial information where appropriate and limiting who can obtain the data. Some resources are publicly accessible; others require a researcher to apply and accept conditions of use. The right combination depends on the study. 1, 2

Privacy measures can also affect how useful the data are. For example, a broader age range may provide less detail for an analysis. Researchers should explain the protections and their limitations in language participants can understand.

Questions to ask your study team

• What scans, recordings and other information will you collect?
• Will the shared information include my face or detailed demographic information?
• Will any data be publicly accessible, or will other researchers need permission?
• Who will keep the connection between my name and study code?
• What future uses are envisaged, and who will oversee them?
• What choices can you offer about sharing, and what happens if I later change my mind?
• Who should I contact about privacy or data use?

Your participant information sheet and study team are the starting points for answers about your own data. In particular, ask them to explain any limits on changing a decision after data have already been shared or used.

Sources

1. Kibsgaard et al. (2026), Assessing metadata privacy in neuroimaging: published research on information accompanying OpenNeuro datasets.
2. Pernet et al., An expert consensus on identification likelihood in NeuroImaging data: working manuscript consulted 6 November 2026, not yet published. The explanations here are draft summaries, not final consensus recommendations.
3. Steeg et al. (2024), research on recognising participants from MRI-derived faces.
4. Finn et al. (2015), research on matching brain-connectivity patterns across recordings.