Success stories

Automated analysis of MRI images

Scenario: A medical research facility aims to implement an automated analysis system for MRI brain images, which inherently contain identifiable biometrics. Due to the sensitive nature of the data, processing these images must adhere to stringent data protection regulations. Additionally, the considerable computational power required dictates that the analysis is performed in an external data center.

Challenges:

  • Extracting quantifiable and interpretable information from MRI image data within 10 minutes post-examination to support timely clinical decisions.
  • Ensuring the security of highly sensitive personal data while utilizing external computational resources.
  • Coordinating a complex workflow among multiple systems, including MRI scanners, servers, and PACS (Picture Archiving and Communication System), all while complying with the DICOM (Digital Imaging and Communications in Medicine) standard to safeguard patient privacy.

How Secure HPC Solves This:

  • Data Encryption and Access Control: MRI images are routed to a secure intranet node for pseudonymization and pre-processing. Following this, they are analyzed on a secure HPC partition at GWDG, where access is strictly controlled to protect sensitive information.

  • End-to-End Data Protection: The secure HPC framework ensures that third-party access to sensitive personal data is prevented throughout the entire workflow, meeting high data protection requirements crucial for handling medical information.

  • Rapid Processing with Security: The FastSurferCNN AI model is used for the segmentation of the MRI images. By processing only the necessary layers, the model minimizes memory usage while generating reliable insights that can assist medical professionals without compromising data security.

  • Robust System Architecture: The system is designed to ensure seamless interaction among MRI scanners, analysis servers, and PACS under the DICOM standard, fostering a secure operational environment that prioritizes patient confidentiality.

Impact: The project has been well received by both staff and patients, highlighting its potential to transform medical imaging workflows while ensuring the highest standards of patient confidentiality. Through this implementation, the facility successfully conducts automated analysis of MRI images while maintaining the utmost security for sensitive patient data, thus supporting clinical operations effectively and ethically.