New ClinX framework de-identifies multimodal medical data

Researchers have developed ClinX, a new framework designed to de-identify protected health information (PHI) in multimodal medical datasets. This system combines optical character recognition (OCR) for detecting visible identifiers with a generative restoration module, ClinX-PRISM, to suppress burned-in text. ClinX also employs various text-side sanitization techniques, including regex masking and context-aware methods. Evaluations in medical visual question answering (MedVQA) demonstrate that while OCR alone is insufficient, ClinX's restoration-based approach effectively reduces PHI leakage while preserving clinically relevant visual context. AI

IMPACT Enhances privacy in medical AI by enabling safer use of multimodal datasets for research and development.

RANK_REASON The cluster contains an academic paper detailing a new method and framework for de-identification in medical AI. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CV →

• arXiv
• ClinX
• ClinX-PRISM
• MedVQA
• optical character recognition

• paper
• safety
• product

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →