Capsule Endoscope Lesion Detection via Groove Shadow Elimination
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image processing systems for capsule endoscopes face challenges in accurately detecting lesion regions within in-vivo images due to variations in pixel values and the presence of grooves between organ walls, which can lead to false positives and negatives.
Innovation Solution
An image processing apparatus and method that includes a suspected-lesion-region extracting unit, a groove determining unit, and a lesion-region extracting unit, which analyze pixel value variations and shape features to differentiate between actual lesion regions and groove shadows, using techniques such as pixel-value-variation calculations and shape feature data analysis to refine the identification of lesion regions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If simple color data comparison is used for lesion detection, then processing speed is maintained, but measurement precision deteriorates due to false positives from groove shadows
Solution Approach 1:
The image processing is divided into multiple stages: initial lesion candidate detection using color data, followed by groove shadow detection using shape features, and final classification. This segmentation allows each stage to focus on specific characteristics, improving overall accuracy without requiring all complex processing to run simultaneously.
Solution Approach 2:
Shape feature data serves as an intermediary indicator to resolve the contradiction between color-based lesion detection and groove shadow identification. The shape features act as a mediating criterion that helps distinguish true lesions from groove shadows when color data alone is insufficient, enabling more accurate classification without directly complicating the primary detection mechanism.
2Measurement precision
If multiple processing criteria are applied to improve detection accuracy, then measurement precision improves, but processing time increases
Solution Approach 1:
The system performs preliminary detection using simple color data comparison first to identify potential lesion regions. Only regions that meet certain criteria proceed to more complex shape feature analysis. This preliminary action filters out obvious non-lesion areas early, reducing the number of images requiring full multi-criteria processing and thus minimizing overall processing time.
Solution Approach 2:
The system applies full multi-criteria processing (both color and shape analysis) only to regions where it is most needed - namely, areas that are ambiguous based on color data alone. For clear-cut cases, simpler processing suffices. This partial application of excessive action ensures high accuracy for difficult cases without unnecessarily processing all images with the most complex algorithm.
3Reliability
If color data comparison is used for lesion detection, then processing is simple, but reliability deteriorates due to variations in illumination and image quality
Solution Approach 1:
The system transitions from relying solely on color data parameters to incorporating shape feature parameters as well. By changing the detected parameters from just color-based to include both color and geometric properties, the system becomes more robust against illumination variations that affect color but not shape characteristics, thereby improving reliability.
Solution Approach 2:
The detection approach combines multiple types of data (color information and shape features) into a composite analysis framework. This composite method leverages the strengths of both data types - color for initial screening and shape for verification - creating a more reliable detection system that compensates for the weaknesses of individual parameters under varying illumination conditions.
Data Source
AI summary
An image processing apparatus includes a suspected-lesion-region extracting unit that extracts a suspected lesion region from an in-vivo image that is obtained by taking an image of inside of body; a groove determining unit that determines whether the suspected lesion region is a region corresponding to a shadow of a groove that is formed between in-vivo organ walls; and a lesion-region extracting unit that extracts a lesion region using the suspected lesion region and a result of determination by the groove determining unit.