AI Endoscopic Size Evaluation for Objective Polyp Measurement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing endoscopic detection methods rely on visual estimation of polyp size, which is subjective and prone to disputes, lacking objective accuracy for medical diagnosis.
Innovation Solution
An auxiliary evaluation system utilizing AI-powered segmentation, depth prediction, and size prediction models to analyze and display the size of abnormal features in real-time endoscopic images, providing objective size data for accurate diagnosis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If visual estimation method is used to measure polyp size, then the measurement process is simple and quick, but the measurement precision is low and subjective
Solution Approach 1:
The patent introduces an auxiliary evaluation system as an intermediary between the endoscope and the doctor. This system includes a computing device with segmentation model, depth prediction model, and size prediction model that processes real-time images to calculate objective polyp size measurements, eliminating the need for direct visual estimation while maintaining workflow integration
Solution Approach 2:
The patent replaces the mechanical/visual estimation method with an AI-based computational system. The segmentation model identifies polyp boundaries, the depth prediction model estimates distance from the endoscope lens, and the size prediction model calculates actual dimensions, substituting human visual judgment with automated image processing and calculation algorithms
2Measurement precision
If AI-based size measurement system is implemented, then the measurement precision is improved, but the device complexity increases
Solution Approach 1:
The auxiliary evaluation system is designed to be universally applicable to different endoscopic detection instruments. The computing device receives real-time images from the detection instrument and performs multiple functions including segmentation, depth prediction, and size calculation through integrated AI models, making the complex system versatile across different medical devices
Solution Approach 2:
The system performs automated processing where the segmentation model automatically identifies abnormal features, the depth prediction model autonomously estimates distances, and the size prediction model independently calculates dimensions. The display device automatically presents results to the doctor, reducing manual intervention despite the underlying system complexity
3Reliability
If visual estimation is used for polyp size assessment, then the operation process is simple, but the reliability of diagnosis is reduced due to subjectivity
Solution Approach 1:
The system provides real-time feedback to the doctor by displaying calculated polyp sizes on the screen during the examination. The auxiliary evaluation system continuously processes real-time images and feeds back objective measurement data, allowing the doctor to make informed decisions immediately without waiting for post-processing analysis
Solution Approach 2:
The system performs preliminary size assessment and measurement calculations automatically during the endoscopic examination. By pre-calculating polyp dimensions and presenting them in real-time, the system prepares objective data before the doctor makes the final diagnosis or treatment decision, improving both reliability and workflow efficiency
Data Source
AI summary
An auxiliary evaluation system and method are provided here. The auxiliary evaluation system is electrically connected to a detection instrument. The detection instrument examines a target and generates a real-time image. The auxiliary evaluation system includes a computing device and a display device. The computing device is signally connected to the detection instrument, and includes a segmentation model, a depth prediction model, and a size prediction model. The computing device receives the real-time image, to mark a selection box for an abnormal feature on the real-time image. The segmentation model generates a bounding box and position information corresponding to the abnormal feature through calculation according to the real-time image and the selection box. The depth prediction model estimates a depth from the abnormal feature. The display device is electrically connected to the computing device to display the real-time image, the bounding box, and the size.