Medical Image Text Recognition for Personal Data Removal
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Medical images contain sensitive personal data that can be misused if accessed unauthorizedly, posing risks such as identity theft and discrimination, necessitating effective detection and removal of personal information.
Innovation Solution
A computer-implemented method and system that recognizes personal data in medical images by analyzing text within the images, issues notifications, and optionally removes the identified data using image-to-image translation models like GANs or diffusion models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If medical images are stored and processed to enable automatic analysis and training, then productivity and research capabilities are improved, but the risk of unauthorized access and misuse of personal data increases
Solution Approach 1:
The system performs preliminary detection and classification of personal data in medical images before any analysis or processing occurs. By identifying and flagging images containing personal data in advance, the system enables selective processing while maintaining security, thus resolving the contradiction between productivity and security risks.
Solution Approach 2:
The patent introduces an intermediary classification layer that sits between the medical image storage and analysis systems. This intermediary component detects and categorizes images based on personal data presence, acting as a mediator that allows secure processing while preventing unauthorized access to sensitive information.
2Measurement precision
If text recognition is performed on medical images to detect personal data, then measurement precision of data detection is improved, but device complexity increases
Solution Approach 1:
The system segments the detection process into distinct functional modules: image reception, text region detection, optical character recognition, and personal data classification. This segmentation allows each component to specialize in a specific task, improving overall measurement precision while managing complexity through modular architecture.
Solution Approach 2:
The patent employs a multi-functional processing pipeline where a single system performs multiple operations sequentially - detecting text regions, recognizing characters, classifying personal data types, and determining sensitivity levels. This universal approach consolidates multiple detection functions into one integrated system, improving precision without proportionally increasing complexity.
3Object-affected harmful factors
If personal data is removed from medical images to protect patient privacy, then data security is improved, but loss of information occurs
Solution Approach 1:
The system applies local quality differentiation by treating different types of personal data differently. Instead of uniform removal, the classification output enables selective handling - some data may be redacted while other information is retained or anonymized differently, thus protecting privacy while minimizing information loss based on the specific data characteristics and sensitivity levels.
Data Source
Figure 1~2
Figure 3~4
Figure 5(a)~7(c)
AI summary
Systems, methods, and computer programs disclosed herein relate to the recognition of personal data in medical images.