Measuring the potential risk of re-identification of imaging research participants from open-source automated face recognition software
Contact us if these videos are not relevant.
Contact us if these videos are not relevant.
View abstract on PubMed
Summary
This summary is machine-generated.Facial recognition software can re-identify research participants from brain imaging data. Even freely available open-source tools achieved 59% accuracy, highlighting privacy risks in brain imaging research.
Area Of Science
- Neuroimaging
- Computer Vision
- Biometrics
Background
- Facial recognition software is widely available and accessible.
- Previous studies show commercial software can identify individuals from brain imaging data.
- Open-source facial recognition tools are now readily available.
Purpose Of The Study
- To assess the accuracy of commercial and open-source facial recognition software in re-identifying research participants from brain imaging data.
- To evaluate the feasibility of re-identification using freely available software packages.
Main Methods
- Tested two commercial and several open-source facial recognition software packages.
- Used a "population to sample" threat model.
- Measured re-identification accuracy by matching facial photographs to MRI-based face reconstructions of 182 participants.
Main Results
- Open-source software achieved up to 59% accuracy in re-identifying participants.
- Commercial software achieved higher accuracies of 92% and 98%.
- Demonstrated feasibility of re-identification using accessible, open-source tools.
Conclusions
- Freely available facial recognition software poses a privacy risk in brain imaging research.
- High re-identification accuracy is achievable even with open-source tools.
- Replacing identifiable face imagery in brain scans is crucial for participant privacy.
Contact us if these videos are not relevant.
Contact us if these videos are not relevant.