Endoscope Processor Lesion Detection via Multi-Path Image Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing method for generating a learning model using endoscope images has insufficient ability to detect lesions, as identified in Patent Literature 1.
Innovation Solution
An endoscope processor is designed with an image acquisition unit, first and second image processing units, and an output unit that utilize a learning model to generate and output disease status by processing captured images, including generating multiple processed images through methods like Grad-CAM for enhanced lesion detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single processed image is used for lesion detection, then the device complexity is low, but the lesion detection ability is insufficient
Solution Approach 1:
The image processing is segmented into multiple independent processing units (first image processing unit and second image processing unit), each generating different types of processed images from the captured image. This segmentation allows the system to analyze different features of the captured image through different processing methods, thereby improving lesion detection ability while managing system complexity through modular design
Solution Approach 2:
Multiple processed images generated by different image processing units are merged and input together into a single learning model. The learning model integrates information from all processed images to make the final lesion detection decision, combining the strengths of different processing approaches to achieve superior detection performance
2Measurement precision
If multiple processed images are generated and input to the learning model, then the lesion detection accuracy is improved, but the processing time and computational load increase
Solution Approach 1:
Multiple image processing units generate different processed images in parallel as preliminary processing steps before the final lesion detection. By performing these processing operations simultaneously rather than sequentially, the system prepares multiple views of the data in advance, enabling the learning model to make more accurate decisions without proportionally increasing total processing time
Data Source
AI summary
An endoscope processor or the like is provided which has a high ability to detect lesions. The endoscope processor includes an image acquisition unit that acquires a captured image taken by an endoscope, a first image processing unit that generates a first processed image based on the captured image acquired by the image acquisition unit, a second image processing unit that generates a second processed image based on the captured image, and an output unit that outputs an acquired disease status using a learning model, which outputs a disease status, when the first processed image generated by the first image processing unit and the second processed image generated by the second image processing unit are input.