Medical Image-Text Matching for Automated Training Annotation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Annotating large datasets for machine learning algorithms, such as computer-aided detection (CAD) algorithms, is time-consuming and expensive due to the requirement for specialist tools and medical experts, especially for segmentation annotations at a pixel or voxel level.
Innovation Solution
A method and apparatus that utilize a combination of image and text processing techniques to automatically identify abnormal portions in medical images and entities in text data, using pre-determined models to perform matching processes between the identified abnormal portions and entities, thereby generating matched data for training purposes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual annotation by medical experts is used, then annotation accuracy is improved, but time consumption and cost increase
Solution Approach 1:
The patent introduces an intermediary automated annotation system that processes medical images and generates preliminary annotations, which are then reviewed and refined by medical experts. This intermediary layer reduces the time experts need to spend while maintaining annotation accuracy through a two-stage process combining automated speed and expert precision.
Solution Approach 2:
The system performs preliminary automated annotation before expert review, preparing draft annotations that experts can then verify and correct. This preliminary action by automated algorithms reduces the overall time consumption by handling the initial annotation work, allowing experts to focus only on reviewing and refining rather than creating annotations from scratch.
2Measurement precision
If manual annotation by medical experts is used, then annotation quality is improved, but cost increases
Solution Approach 1:
The automated annotation system serves as an intermediary that handles the initial annotation work, reducing the need for extensive expert involvement. This intermediary process maintains annotation quality by generating accurate preliminary annotations that experts can verify, thereby reducing costs while preserving quality through collaborative processing.
Solution Approach 2:
The system uses automated algorithms to create initial annotation copies that can be reviewed and corrected by experts. These automated copies serve as drafts that reduce the overall annotation work required from experts, thereby reducing costs while maintaining quality through the expert review process.
3Productivity
If automated processing is used, then productivity is improved, but annotation accuracy deteriorates
Solution Approach 1:
The patent merges automated annotation processing with expert review in a hybrid system. The automated component provides high-speed initial annotations, while the expert review component ensures accuracy. This merging of automated productivity with expert precision resolves the contradiction by combining the strengths of both approaches in a unified workflow.
4Reliability
If specialist tools and medical experts are required, then annotation reliability is improved, but device complexity increases
Solution Approach 1:
The automated annotation system acts as an intermediary that simplifies the overall process by handling initial annotation work, reducing the need for complex specialist tools and extensive expert involvement. This intermediary layer maintains reliability through automated consistency while reducing system complexity by automating routine tasks.
Data Source
AI summary
An apparatus comprising processing circuitry configured to: obtain image data representing one or more medical images of a region of interest and processing said image data to identify one or more abnormal portions of the image by applying a first pre-determined model to the obtained image data; obtain medical text data corresponding to the one or more medical images of the region of interest and process said medical text data to identify one or more entities and their associated attributes by applying at least one further pre-determined model to the obtained medical text data; perform a matching process between the one or more identified abnormal portions and the one or more identified entities to obtain matched data comprising groupings of at least one identified abnormal portions and at least one entity, wherein the matching process is based on at least one or more properties of the identified abnormal portions of the image data and at least one or more of the attributes associated with the identified entities.