Endoscope Image Similarity Filtering for Complete Inspection-Site Recording
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing endoscope systems struggle to comprehensively record images of predetermined inspection sites during primary examinations, risking oversight of lesions in secondary examinations.
Innovation Solution
An image processing apparatus that determines similarity of endoscope images to predetermined inspection sites and suitable observation modes, automatically recording a predetermined number of images in a recording unit when conditions are met, using deep learning and condition determination.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If automatic recording is performed without determining image similarity to predetermined inspection sites, then recording speed is improved, but image omission at inspection sites occurs
Solution Approach 1:
The patent replaces manual visual inspection and selective recording with an automated image processing system that uses deep learning models to automatically determine image similarity to predetermined inspection sites and select images for recording. This substitution of manual mechanical operations with automated intelligent processing resolves the contradiction by enabling both high-speed automatic recording and reliable image capture completeness through AI-based decision making.
2Quantity of substance
If all endoscope images are recorded without filtering by observation mode, then quantity of recorded images is improved, but storage capacity requirements increase
Solution Approach 1:
The patent extracts and filters only the essential images that meet predetermined conditions (similarity to inspection sites and suitable observation modes) from the complete set of endoscope images. By taking out only the necessary images for recording rather than storing all images, the system maintains adequate quantity of useful images while significantly reducing storage capacity requirements.
Solution Approach 2:
The patent changes the selection parameters for image recording from recording all images to recording only images that satisfy specific conditions (image similarity threshold and observation mode suitability). This parameter change in the recording criteria enables the system to reduce the quantity of stored images while maintaining the essential information needed for medical diagnosis.
3Measurement precision
If manual selection of images for recording is performed, then recording accuracy is improved, but operation time increases
Solution Approach 1:
The patent implements self-service by enabling the image processing system to automatically evaluate and select images for recording without requiring manual intervention. The deep learning models independently perform image similarity assessment and observation mode determination, making the system self-sufficient in the image selection process. This eliminates the time-consuming manual selection while maintaining high recording accuracy through sophisticated automated algorithms.
Data Source
AI summary
An image processing apparatus includes a processor. The processor receives endoscope images to be determined, in each of which an image of an inside of a body of a subject is picked up, determines whether or not an endoscope image for comparison concerning a predetermined inspection site, and the endoscope images to be determined are same or similar, determines whether or not an observation mode of each of the endoscope images to be determined is a condition suitable for automatic recording, and automatically records a predetermined number of the endoscope images to be determined in a recording unit, when the endoscope images to be determined and the endoscope image for comparison are determined to be the same or similar and when the observation mode is the condition suitable for automatic recording.