CNN Lumen Direction Estimation for Endoscope Operation Resumption
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing endoscope technologies face challenges in efficiently identifying the lumen direction within captured images, especially when the lumen is not present or is difficult to detect, leading to increased time and complexity in resuming normal operation.
Innovation Solution
An image processing device utilizing a Convolutional Neural Network (CNN) with a trained model that associates training images with category information including lumen direction, allowing the device to estimate the lumen direction in captured images even when the lumen is not directly visible.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If coordinate transformation is performed to obtain lumen direction from past frames, then lumen direction information can be obtained when lumen is not visible in current frame, but the process increases time complexity and reduces processing speed
Solution Approach 1:
The system performs preliminary detection of lumen direction information in advance by analyzing past frames and storing this information for future use. When the lumen becomes invisible in the current frame, the pre-detected direction information from past frames is immediately utilized, avoiding the need for time-consuming coordinate transformation and corresponding point detection at that moment.
Solution Approach 2:
The system creates a copy of lumen direction information from past frames and uses this copied information when the lumen is not visible in the current frame. Instead of performing complex coordinate transformation calculations, the system directly utilizes the copied directional data, significantly reducing processing time while maintaining reliability.
2Reliability
If coordinate transformation and corresponding point detection are performed, then lumen direction can be estimated when lumen is missing, but the device complexity and computational load increase
Solution Approach 1:
The system extracts only the essential lumen direction information from past frames and stores it for future reference. Instead of performing complete coordinate transformation and corresponding point detection when needed, the system has already extracted and stored the critical directional data in advance, simplifying the processing required at the moment of need.
Solution Approach 2:
The system performs the complex extraction and storage of lumen direction information in advance during periods when the lumen is visible. This preliminary action prepares the data structure and reduces the computational burden to minimal operations when the lumen becomes invisible, thereby reducing overall device complexity.
3Measurement precision
If chronological image analysis is performed to identify lumen direction, then accurate lumen positioning is achieved, but the processing time and operational efficiency decrease
Solution Approach 1:
The system performs preliminary analysis of chronological images to identify and store lumen direction information in advance. When the lumen becomes invisible, the pre-identified positional accuracy data is immediately retrieved and used, maintaining high measurement precision while avoiding time-consuming re-analysis and improving operation resumption efficiency.
Data Source
AI summary
An image processing device includes: one or more processors comprising hardware, wherein the one or more processors are configured to: input a captured image captured by an endoscope in a body cavity of a subject to a Convolutional Neural Network (CNN) using a trained model, the trained model having training data in which each of training images is associated with category information including a lumen direction of a region outside the each of training images in which the lumen is present; estimate, based on the CNN, category information including a lumen direction of a region outside the captured image in which the lumen is likely to be present; and output the category information estimated.