Image Processing Device for Stable Lesion Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing techniques struggle to accurately detect abnormal areas in images, particularly in intraluminal images, due to limitations in calculating feature data and approximating distribution profiles, leading to instability in lesion detection across varying sizes and types.
Innovation Solution
An image processing device and method that calculates feature data for each pixel, approximates the distribution profile using geometric graphics like straight lines or curve lines in a feature space, and detects abnormal areas based on intra-feature-space distances and threshold processing, enabling stable detection of lesions regardless of size.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional image processing techniques are used to detect abnormal areas, then the detection process is simple, but the detection accuracy and stability are insufficient
Solution Approach 1:
The patent segments the image processing task into multiple distinct steps: feature data calculation for each pixel, distribution profile generation in feature space, approximation curve fitting, and abnormal area detection based on distance metrics. This segmentation allows each step to be optimized independently, improving overall detection accuracy while managing complexity through modular processing.
Solution Approach 2:
The patent transforms the image data from spatial domain to feature space by calculating feature data for each pixel and generating distribution profiles. This dimensional transformation enables the use of geometric approximation and distance-based detection methods that are more effective for identifying abnormal areas, thereby improving detection precision.
2Measurement precision
If feature data calculation and distribution profile approximation are performed for each pixel, then detection accuracy improves, but processing time increases
Solution Approach 1:
The patent performs preliminary calculations by pre-computing feature data for each pixel and generating distribution profiles before the actual detection process. The approximation curves are fitted in advance, allowing the final abnormal area detection to use pre-prepared reference data, thereby reducing real-time processing requirements while maintaining high accuracy.
3Reliability
If simple threshold processing is used for abnormal area detection, then processing speed is fast, but detection stability varies with image characteristics
Solution Approach 1:
The patent changes the detection parameter from simple pixel intensity thresholds to distance metrics in feature space. By calculating the distance between each pixel's feature data and the approximation curve representing normal tissue, the method adapts to varying image characteristics while maintaining consistent detection stability. This parameter transformation enables reliable detection across different image types without requiring manual threshold adjustment.
Data Source
AI summary
An image processing device includes: a feature data calculator configured to calculate the feature data of each pixel in an image; an approximate shape calculator configured to calculate an approximate shape approximating a profile of a distribution area in which the feature data is distributed in a feature space having the feature data as an element; and an abnormal area detector configured to detect an abnormal area in the image based on the approximate shape and the profile of the distribution area.