Endoscopic Image Processing for Malignancy Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image processing systems for endoscopic images struggle to accurately classify regions of interest within medical images, particularly in distinguishing between lesions with varying degrees of malignancy and those where malignancy judgment is difficult, leading to suboptimal diagnostic support information.
Innovation Solution
An image processing apparatus that calculates first and second feature values for pixels and regions, classifies these regions into predefined attributes, and adjusts attribute values based on surrounding regions' rates to provide enhanced diagnostic support information, incorporating a correction mechanism for regions with uncertain malignancy indicators.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If simple classification based on feature values is used, then processing speed is improved, but classification accuracy deteriorates
Solution Approach 1:
The image processing is divided into multiple stages: first feature value calculation for all pixels, then region division based on these features, followed by second feature value calculation for each region, and finally classification. This segmentation allows efficient processing at each stage while achieving accurate final classification through the multi-stage approach.
Solution Approach 2:
Different feature values are calculated for different regions based on their characteristics. The system calculates first feature values for all pixels, then identifies regions with different properties, and calculates second feature values specific to each region type. This local quality approach ensures that each region is classified using features appropriate to its characteristics, improving overall accuracy.
2Measurement precision
If detailed region analysis is performed, then diagnostic accuracy is improved, but processing time increases
Solution Approach 1:
The processing is segmented into efficient stages: quick first feature value calculation for all pixels, followed by region division, then focused second feature value calculation only for regions that need detailed analysis. This segmentation reduces processing time while maintaining diagnostic accuracy for regions requiring detailed examination.
Solution Approach 2:
The system performs detailed analysis (second feature value calculation) only for specific regions that require it, rather than analyzing all regions equally. This partial action approach focuses computational resources on areas needing detailed diagnostic analysis, reducing overall processing time while maintaining accuracy where needed.
3Reliability
If correction processing is applied to regions with uncertain attributes, then classification reliability is improved, but processing complexity increases
Solution Approach 1:
The system performs preliminary classification using second feature values to identify regions with uncertain attributes before applying correction processing. This preliminary action allows the correction to be targeted only at specific regions needing it, improving reliability without unnecessarily complicating the overall processing system.
Solution Approach 2:
The correction processing acts as an intermediary step between initial classification and final diagnostic output. It specifically addresses regions with uncertain attributes by applying correction based on surrounding regions, serving as a mediator that improves reliability without requiring complete system redesign.
Data Source
AI summary
An image processing apparatus is provided with: a first feature value calculating section calculating a first feature value for each of pixels constituting an image obtained by picking up an image of a subject; a region dividing section dividing the image into multiple regions on the basis of the first feature values; a second feature value calculating section calculating a second feature value for each of the divided regions; a classification section performing classification with regard to which of multiple kinds of attributes each region of the multiple regions has, on the basis of the second feature value; a judgment section judging whether a region having a predetermined attribute exists or not; and a diagnostic support information calculating section correcting an attribute value of the region having the predetermined attribute to calculate diagnostic support information for supporting a diagnosis.