Line Enhancing Filter for Medical Image Fissure Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image processing methods fail to accurately detect thin, faint, and non-continuous line structures in medical images, such as lobar fissures in CT scans, due to noise, anatomical anomalies, and the need for large ground truth annotations.
Innovation Solution
A line enhancing filter that calculates correlation values between a line model and image data, testing multiple hypotheses for line orientations, which is robust against noise and effective for detecting thin line structures, even when they are faint or non-continuous, using a rectangular or cuboidal line model and iterative refinement.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If Hessian matrix eigenvector analysis is used to detect bright plate-like fissures, then detection capability is improved, but the filter gives low responses for fissure voxels due to noise and thin structure challenges
Solution Approach 1:
The patent changes the detection parameters from Hessian matrix eigenvector analysis to a line structure filter that tests multiple hypotheses for line orientations. This parameter change allows the filter to adapt to different line orientations and maintain reliable responses even in noisy conditions and for thin fissure structures.
Solution Approach 2:
The patent introduces a dynamic filtering approach where the filter adapts its response based on local image characteristics. The filter dynamically adjusts its behavior by testing multiple orientation hypotheses and selecting the best match, making it reliable for detecting thin, faint lines with varying orientations.
2Measurement precision
If supervised learning filter is used for fissure detection, then detection accuracy is improved, but large set of ground truth annotations is required
Solution Approach 1:
The patent implements a self-service approach where the filter does not require external ground truth annotations for training. Instead, it uses the image data itself and tests multiple orientation hypotheses to automatically determine the best fit, making the system independent of manual annotation processes.
Solution Approach 2:
The patent extracts the essential characteristics of line structures directly from the image data without requiring separate training datasets. By taking out only the necessary image features and testing multiple orientation hypotheses, the system achieves accurate detection without the complexity of supervised learning pipelines.
3Productivity
If standard line detection algorithms are used, then processing speed is maintained, but they fail to detect non-continuous, curved, or faint line structures
Solution Approach 1:
The patent introduces a dynamic filtering approach where the filter adapts its response based on local image characteristics. The filter dynamically adjusts its behavior by testing multiple orientation hypotheses and selecting the best match, making it reliable for detecting thin, faint lines with varying orientations.
Solution Approach 2:
The patent extends the detection approach by adding the orientation dimension. Instead of checking for lines in a single fixed orientation, the filter tests multiple orientation hypotheses, effectively adding an angular dimension to the detection process. This allows detection of lines with various orientations while maintaining processing efficiency.
Data Source
Figure 1
Figure 2A~3
Figure 4A~5D
AI summary
The present invention relates to an image processing device for detecting line structures in an image data set. The device comprises a model definition unit (12) for defining a line model of a line structure to be detected, said line model comprising a number of voxels, a calculation unit (14) for calculating, per voxel of interest of said image data set, several correlation values of a correlation between said line model and an image area around said voxel of interest, said image area comprising a corresponding number of voxels as said line model, wherein for each of a number of different relative orientations of said line model with respect to said image area a respective correlation value is calculated, and a determining unit (16) for determining, per voxel of interest, the maximum correlation value from said calculated correlation values and the corresponding optimal orientation at which said maximum correlation value is obtained.