Medical Image Anonymization Target Selection for Flexible Privacy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems fail to easily set or change anonymization targets in medical images based on their intended use, leading to insufficient privacy measures and inefficient workflow improvements in medical video utilization.
Innovation Solution
An information processing system and method that includes an image recognition unit to detect anonymization candidates, an anonymization target determination unit to determine targets based on user operation, and an image processing unit to generate anonymized images by blurring or mosaic, allowing flexible selection of anonymization targets based on the intended use of the medical images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If automated anonymization is applied to medical images, then privacy protection is improved, but the ability to adapt to different presentation contexts and purposes is worsened
Solution Approach 1:
The system dynamically adjusts anonymization targets based on presentation context through a selection interface that allows users to choose which individuals or objects should be anonymized. This dynamic selection capability enables the same system to adapt to different presentation purposes (e.g., educational vs. research contexts) while maintaining reliable privacy protection for selected targets.
Solution Approach 2:
The system segments the anonymization process into detectable candidates and selectable targets. By first detecting all potential anonymization candidates (faces, bodies, instruments) and then allowing selective determination of actual targets, the system maintains comprehensive privacy coverage while enabling context-specific adaptability.
2Reliability
If comprehensive anonymization of all detected candidates is applied, then privacy protection is improved, but operational flexibility and ease of use are worsened
Solution Approach 1:
The system provides self-service through an automated detection phase that identifies all potential anonymization candidates, reducing the manual workload. Users then simply need to make selection determinations among detected candidates rather than performing complete manual analysis, improving ease of operation while maintaining comprehensive privacy protection.
Solution Approach 2:
The system performs preliminary detection and classification of all potential anonymization candidates before the final selection phase. This preliminary action organizes and pre-processes the data, making the subsequent user selection process more efficient and easier to operate while ensuring all privacy-relevant elements are accounted for.
3Adaptability or versatility
If manual selection of anonymization targets is required, then adaptability to different purposes is improved, but processing time and productivity are worsened
Solution Approach 1:
The system performs preliminary automated detection and classification of all potential anonymization candidates before user selection. This pre-processing organizes the data and presents only the relevant candidates to users, significantly reducing the time required for manual selection while maintaining full adaptability to different presentation purposes.
Solution Approach 2:
The system introduces an intermediary selection interface that mediates between automated detection and final user decisions. This intermediary layer processes and organizes detected candidates, presenting them in a structured manner that accelerates user selection while preserving the flexibility to adapt to different anonymization purposes.
Data Source
AI summary
Provided is an information processing system and an information processing method capable of easily setting or changing an anonymization target to be anonymized according to use of an image captured in a medical facility. Anonymization candidates that can be an anonymization target included in a target image captured in a medical facility are detected, an anonymization target in the target image is determined on the basis of an operation of a user, and an anonymized image obtained by anonymizing an image region of the anonymization target determined by the anonymization target determination unit among the anonymization candidates detected by the image recognition unit is generated for the target image.