AI-Guided Endoscopy Support for Complex Stomach Navigation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing endoscopic systems struggle to efficiently navigate and observe specific parts of the stomach, such as the posterior wall of the anglar region and the lesser curvature of the cardia region, due to their complexity and the risk of overlooking lesions during examination.
Innovation Solution
An endoscopy support apparatus equipped with a processor and memory that utilizes a learned model to infer the name, visual field direction, and axis direction of stomach parts from annotated endoscopic images, guiding the operator to navigate and observe target areas more effectively.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If manual observation and navigation of stomach parts is performed during endoscopy, then the operator can examine the stomach, but it is difficult to efficiently navigate and observe specific complex parts such as the posterior wall of the anglar region and the lesser curvature of the cardia region
Solution Approach 1:
The patent introduces an artificial intelligence model as an intermediary between the endoscopic images and the operator. The AI model automatically infers the names of stomach parts, visual field directions, and axis directions from images, converting complex anatomical navigation into automated information processing. This mediator handles the complexity of stomach anatomy, providing clear directional guidance to the operator without requiring manual interpretation of complex spatial relationships.
Solution Approach 2:
The patent replaces manual mechanical navigation and visual assessment with an automated AI-based system. Instead of relying on the operator's manual observation and spatial reasoning skills, the system uses image processing and machine learning algorithms to automatically determine part names and directions. This substitution of mechanical/manual operations with automated intelligence simplifies the navigation process for complex anatomical structures.
2Reliability
If comprehensive observation of multiple examination parts is performed to prevent overlooking lesions, then diagnostic accuracy improves, but the time and complexity of examination increases
Solution Approach 1:
The patent enables the examination system to serve itself by automatically inferring part names and directions without requiring continuous manual intervention. The AI model processes images autonomously, automatically identifying examination parts and their spatial relationships. This self-service capability allows comprehensive observation of multiple parts to be performed efficiently, maintaining high diagnostic accuracy while reducing the time burden on operators.
Solution Approach 2:
The patent changes the parameters of the examination process by introducing automated inference of part names, visual field directions, and axis directions. Instead of manual assessment of these parameters, the system uses AI to automatically determine them from images. This parameter transformation from manual to automated processing enables more comprehensive observation within the same time frame, improving reliability without proportionally increasing examination time.
3Extent of automation
If automated recording of endoscopic images is implemented based on color threshold detection, then recording automation is achieved, but the system cannot identify specific target parts or their directions
Solution Approach 1:
The patent performs preliminary action by automatically inferring part names, visual field directions, and axis directions before recording decisions are made. The AI model processes images in advance to extract comprehensive spatial and contextual information, including the identity and location of target parts. This preliminary information extraction enables automated recording systems to not only detect when to record but also to understand what is being recorded and where, preventing loss of target part location information.
Solution Approach 2:
The patent adds another dimension to automated recording by incorporating spatial orientation information (visual field direction and axis direction) alongside part identification. Instead of simple color-based triggering, the system analyzes images in multiple dimensions including anatomical context and spatial relationships. This multi-dimensional approach ensures that automated recording captures not just the presence of target parts but also their precise locations and orientations, eliminating information loss.
4Measurement precision
If AI-based inference of part names and directions is implemented, then navigation accuracy improves, but the system complexity and computational requirements increase
Solution Approach 1:
The patent implements a universal AI model that performs multiple functions simultaneously: inferring part names, determining visual field directions, and identifying axis directions from a single image input. This multi-functional approach consolidates what would otherwise require separate processing systems into one unified model, improving measurement precision while managing system complexity through functional integration rather than multiplication of separate components.
Data Source
AI summary
An endoscopy support apparatus includes a memory and a processor. The processor is configured to access the memory that stores: a learned model learned by a learning data set in which a name of a part, a visual field direction, and an axis direction are annotated to each of a plurality of endoscopic image; and a name of at least one target part and a positional relationship of a plurality of parts, and the processor inputs a picked-up image into the learned model, to thereby infer the name of the part, the visual field direction, and the axis direction, in the picked-up image, and outputs a direction of the target part in the picked-up image, based on the positional relationship of the plurality of parts, and the name of the part, the visual field direction, and the axis direction, in the picked-up image, which have been inferred.