Privacy-preserving visualization of brain functional connectivity

Visualizations are fundamental to neuroimaging research, facilitating tasks ranging from exploratory data analysis to the communication and interpretation of findings. Despite their necessity, visualizations can potentially compromise the confidentiality of individual participants. In this paper, we discuss how visualizations may inadvertently lead to privacy leakage and explore methods to mitigate such risks. Our work investigates ways to securely share visualizations that faithfully preserve the patterns supporting the derived insights from data analysis, rather than deriving conclusions from the visualizations themselves. We address privacy-preserving visualization within a differential privacy (DP) framework, focusing on commonly used visualization methods for functional network connectivity. Various perturbation-based strategies are investigated for protecting correlation-related measures, with analyses of their privacy costs and the effects of pre-processing and/or post-processing. To achieve a better balance between privacy and visual utility, this paper introduces novel workflows tailored for connectogram and seed-based connectivity visualizations that maintain the qualitative patterns observed in non-private results. Overall, this work illustrates how DP can be effectively applied to neuroimaging visualization, demonstrating its efficacy as a robust methodology for securing sensitive biomedical data.