The Multimodal and Functional Imaging Laboratory (MAFIL), a part of the Central European Institute of Technology (CEITEC) at Masaryk University, is an advanced neuroimaging research facility. MAFIL primarily focuses on mapping the human brain's structure and function.
To manage this research data, the MAFILDB system was developed to handle the entire data lifecycle, from acquisition and processing to secure sharing with research partners and volunteers. A core goal of the platform is to make the data FAIR while digitizing measurement records and managing highly sensitive personal data. The system's architecture integrates data from various sources, including DICOM images from a PACS, which are accessed via a dedicated PACS Connector. 
This thesis investigates methods for de-identification and exporting experimental medical imaging data stored in the DICOM format for research purposes. The work explores the DICOM specification, including metadata fields and vendor-specific extensions, as well as the communication and storage of data in PACS systems. A configurable pipeline will be designed to support policy-driven anonymization, enabling metadata fields to be preserved, pseudonymized, or removed as required.
The solution will extend an existing PACS Connector with functionality for querying studies, performing anonymization, and exporting the processed data either back to PACS or to external repositories such as NRP via S3. Integrity of anonymized data and DICOM conformance will be validated using available tools. Where relevant, the work will address anonymization of proprietary vendor tags and explore optional MRI defacing, either directly on DICOM data or through a configurable round-trip via NIfTI.