Medical Image Selection for Preference-Aware Reporting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing medical image processing systems struggle to automatically select images for reports while reflecting a user's preference, leading to increased workload when corrections are needed.
Innovation Solution
A medical image processing apparatus that classifies images into categories, allows user input for designation, and selects images based on designated and non-designated categories, with differentiated display modes and diagnostic usefulness, to reduce the workload while incorporating user preferences.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If AI automatically selects images for reports, then productivity is improved, but reliability deteriorates because the selection may not reflect user preferences
Solution Approach 1:
The system implements feedback by allowing users to provide correction inputs when AI-selected images do not meet their preferences. The correction input generation unit creates additional training data from these corrections, which is then used to retrain the AI model, continuously improving its selection accuracy while maintaining automated efficiency.
Solution Approach 2:
The system enables self-service by allowing users to independently provide correction inputs that automatically generate new training data. This user-provided feedback serves the dual purpose of correcting immediate selection errors and improving future AI performance without requiring manual system reconfiguration.
2Loss of time
If AI automatically selects images for reports, then loss of time is reduced, but loss of information increases because user preferences are not captured
Solution Approach 1:
The system captures user preference information through correction inputs, transforming what would be lost information into valuable training data. Each correction provides insight into user preferences regarding image selection criteria such as light source modes, composition preferences, and diagnostic priorities.
Solution Approach 2:
The system performs preliminary action by proactively collecting user preferences through correction inputs and using them to generate training data before future selection tasks. This prevents information loss by preparing the AI with advance knowledge of user preferences rather than reacting to errors after they occur.
3Reliability
If the system allows user correction input, then reliability is improved, but device complexity increases
Solution Approach 1:
The system merges the correction input functionality with the existing AI training pipeline. Instead of creating a separate complex correction system, the correction inputs are integrated into the training data generation process, allowing the same infrastructure to handle both automated selection and user feedback.
Solution Approach 2:
The correction input mechanism is designed as a self-service system where users independently provide feedback without requiring complex administrative interfaces or manual system reconfiguration. The system automatically processes these inputs to generate training data, reducing the complexity burden on both the user and the system administrators.
Data Source
AI summary
An input receiving unit receives a designation input of a user for a medical image displayed on a first display screen of a display. An image selection unit acquires selected images selected from a plurality of medical images based on a designated image category that is a category assigned to a designated image and a non-designated image category that is a category assigned to a non-designated image. The selected images are displayed in a list on a second display screen of the display.