Endoscopic Image Processing with Adaptive Threshold Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing endoscopic examination systems face challenges in accurately detecting the size of regions of interest, such as lesions, which can be larger or smaller than ideal due to variations in appearance and detection conditions.
Innovation Solution
An image processing device that includes an image acquisition unit, a proportion acquisition unit to determine the region of interest, and a detection unit to generate a detection result based on the proportion, using machine learning models to adjust the threshold value for accurate detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If a fixed threshold value is used for detecting the part of interest, then the detection process is simple, but the detection accuracy deteriorates when the appearance of the part of interest varies
Solution Approach 1:
The patent applies dynamics by making the threshold value adjustable and adaptive rather than fixed. The system dynamically determines the threshold based on the proportion of the part of interest in the image, allowing the detection parameters to change according to the specific characteristics of each image, thereby resolving the contradiction between simple detection process and accurate detection results.
Solution Approach 2:
The patent changes the threshold parameter from a fixed value to a variable that is determined based on the proportion of the part of interest. By adjusting the threshold parameter according to the detected proportion, the system maintains high detection accuracy across different appearances while keeping the overall process manageable through automated parameter adjustment.
2Reliability
If the detection threshold is lowered to detect smaller lesions, then detection sensitivity improves, but false detection rate increases
Solution Approach 1:
The patent implements feedback by using the detected proportion of the part of interest to adjust the threshold value. The system first performs initial detection, calculates the proportion, and then uses this proportion information to refine the threshold for final detection. This feedback loop allows the system to maintain high sensitivity while reducing false detections by adapting the threshold to the actual content of the image.
Solution Approach 2:
The patent applies preliminary action by first detecting the proportion of the part of interest before performing the final detection with adjusted threshold. This preliminary step provides essential information about the image characteristics, enabling the system to set an appropriate threshold that balances sensitivity and false detection rate before the actual detection occurs.
3Measurement precision
If the proportion calculation is performed on the entire image, then comprehensive analysis is achieved, but processing time increases
Solution Approach 1:
The patent applies segmentation by dividing the image processing into distinct stages: initial detection, proportion calculation, and final detection with adjusted threshold. By segmenting the processing steps and performing proportion calculation based on initial detection results rather than analyzing the entire image from scratch, the system achieves comprehensive analysis while reducing overall processing time.
Data Source
AI summary
The image processing device 1X includes an image acquisition means 30X, a proportion acquisition means 31X, and a detection means 33X. The image acquisition means 30X is configured to acquire an endoscopic image obtained by photographing an examination target by a photographing unit provided in an endoscope. The proportion acquisition means 31X is configured to acquire a proportion of an interest part region in the endoscopic image where a part of interest in the examination target exists. The detection means 33X is configured to generate, based on the proportion, a detection result relating to the part of interest in the examination target in the endoscopic image. It can be used to support examiner's decision making and the like.