Medical Display Apparatus Distinguishing AI and User Annotations
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current medical image interpretation systems fail to distinguish between annotations made by AI and those made by radiologists, leading to potential overlooking of lesions and incorrect diagnoses due to indistinguishable annotations and mixed reliability of AI and human annotations.
Innovation Solution
A display apparatus and method that simultaneously display AI-generated and user-specified lesion candidate regions with distinct annotations, allowing radiologists to differentiate between them, thereby preventing overlooked lesions and ensuring accurate diagnoses.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If AI detection results are displayed with the same annotation style as user annotations, then the display is simplified and consistent, but the distinguishability between AI and user annotations is lost
Solution Approach 1:
The patent applies local quality by making the annotation style dependent on the source (AI or user). AI annotations use one visual style while user annotations use another, allowing the display to adapt its characteristics based on the local origin of each annotation rather than using a uniform style for all annotations.
Solution Approach 2:
The patent uses color changes as a visual differentiation mechanism. Different colors are assigned to annotations based on their source (AI-generated versus user-created), enabling radiologists to quickly distinguish between automated detection results and human-verified findings without complicating the overall display interface.
2Productivity
If AI annotations and user annotations are displayed together without differentiation, then the workflow is simplified, but the reliability distinction between AI and human annotations is lost
Solution Approach 1:
The patent employs color coding to visually represent the reliability level of annotations. AI annotations are displayed in one color while user-verified annotations appear in another, allowing radiologists to quickly assess which findings have been validated by human expertise and which require further verification, thus maintaining awareness of reliability differences throughout the workflow.
Solution Approach 2:
The patent changes visual parameters (such as color, line style, or iconography) of annotations based on their source and verification status. This parameter modification enables the system to convey reliability information without adding complexity to the interpretation workflow, as the differentiation is automatically applied and consistently maintained throughout the viewing process.
3Device complexity
If the same annotation style is used for all lesion candidates, then the display is uniform and easy to process, but the ability to distinguish between automated and manual detections is reduced
Solution Approach 1:
The patent implements local quality by applying different visual characteristics to annotations based on their origin. Instead of using a single uniform style for all annotations, the system adapts the visual properties (such as color, border style, or icon) to reflect whether the annotation was generated by AI or created by a user, thereby preserving source information while maintaining a relatively simple display framework.
Data Source
AI summary
A display apparatus includes: a hardware processor configured to cause a display part to simultaneously display first display information indicating a first lesion candidate region obtained by computer processing on medical information and second display information indicating a second lesion candidate region specified by a user on the basis of the medical information, wherein, in a case where the first display information and the second display information are displayed at the same time, the hardware processor distinguishably displays the first display information and the second display information.