Endoscope Processor Learning Models for System Abnormality Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing endoscope systems face difficulties in accurately identifying system abnormalities that affect image quality, making it challenging to correctly specify the cause of such issues.
Innovation Solution
A processor for an endoscope that utilizes a controller to acquire endoscopic images, calculate parameters, and employ learning models to discriminate subject parts and determine system information differences, enabling the detection and correction of system abnormalities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If image processing techniques are used to improve detection accuracy, then detection accuracy is improved, but the ability to correctly specify the cause of system abnormalities deteriorates
Solution Approach 1:
The patent segments the analysis into two independent parts: (1) image processing for detection accuracy, and (2) system abnormality diagnosis through separate parameter analysis. By separating these functions, the system can maintain high detection accuracy while independently improving its ability to diagnose system abnormalities through dedicated parameter monitoring of imaging elements, light sources, and optical systems.
Solution Approach 2:
The patent introduces system information as an intermediary between the image processing system and the diagnosis system. This intermediary contains metadata about the imaging elements, light sources, and optical systems that mediates between the raw image data and the final diagnosis, enabling both accurate detection and correct identification of abnormal causes.
2Difficulty of detecting and measuring
If learning models are used to discriminate subject parts and determine system information, then the ability to identify system abnormalities is improved, but device complexity increases
Solution Approach 1:
The patent applies preliminary action by pre-training learning models with comprehensive system information before actual diagnosis is needed. The models are trained on historical data containing system parameters, imaging element characteristics, and abnormality patterns, so that during actual operation they can quickly and accurately identify abnormalities without requiring complex real-time analysis, thus reducing operational complexity.
Solution Approach 2:
The patent uses parameter changes by transforming complex visual data into simplified system parameters that the learning models can process. Instead of analyzing raw images for abnormality detection, the system extracts key parameters (focus status, illumination intensity, color temperature) and feeds these to the learning models, significantly reducing computational complexity while maintaining diagnostic accuracy.
Data Source
AI summary
A processor for an endoscope according to an aspect is characterized by including: a controller executing program code to perform: acquiring, by the controller, an endoscopic image captured using first system information; calculating, by the controller, parameter on the basis of the endoscopic image acquired by the controller; discriminating a part of a subject using a first learning model that outputs a discrimination result of discriminating the part of the subject in a case in which the calculated parameter is input; outputting second system information using a second learning model that outputs the second system information in a case in which the parameter and the discriminated part of the subject are input; and determining, by the controller, a difference between the second system information output by the second learning model and the first system information.