PixelGuard: Advancing healthcare data privacy through AI-driven de-identification system for medical imaging research

This article discusses PixelGuard, an AI-driven de-identification system for medical imaging research developed by Northwestern University and deployed on AWS.

• Addresses critical privacy concerns in medical image research
• Uses over 75 AI models to detect and redact sensitive information
• Supports multiple image formats including DICOM, JPEG, and PNG
• Ensures compliance with privacy regulations like HIPAA and GDPR
• Preserves clinical relevance while removing personally identifiable information
• Utilizes AWS services like Textract, Comprehend Medical, and S3

PixelGuard enables secure medical image sharing by anonymizing patient data while maintaining research utility, ultimately promoting medical research and innovation.