Tomographic 3D Display Filtering Reduces Blooming Effect
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for filtering tomographic 3D displays suffer from high computation complexity and the 'blooming effect,' which leads to inaccurate vessel size measurements and poor image clarity, especially in areas with calcifications.
Innovation Solution
A method that uses a combination of a single 2D filter applied uniformly across the image area, along with two different linear filters aligned with the extremes of local variances, and weighted mixing of original and filtered voxels to accelerate the filtering process while maintaining image clarity and reducing noise.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If linear low-pass filtering is used to reduce noise, then noise is effectively reduced, but the clarity of data material and quality of display of small structures deteriorates
Solution Approach 1:
The patent applies different filtering strengths to different regions of the image based on local variance calculations. Areas with high variance (likely containing small structures or edges) receive less filtering, while areas with low variance (homogeneous regions) receive stronger filtering. This resolves the contradiction by making the filtering adaptive to local image characteristics rather than applying uniform low-pass filtering throughout.
2Reliability
If two-dimensional or three-dimensional iterative filtering is used with edge information, then noise is reduced while preserving edges, but the delay time increases to minutes per axial slice
Solution Approach 1:
The patent performs only a limited number of iterative filtering steps (specifically one iteration) rather than continuing until convergence. This partial action provides sufficient noise reduction and edge preservation for clinical purposes while reducing processing time from minutes to seconds, making the method suitable for clinical use.
Solution Approach 2:
The patent changes the filtering parameters dynamically based on local variance calculations. By adjusting the filtering strength according to local image characteristics (high variance vs. low variance regions), the method achieves effective noise reduction with edge preservation without requiring excessive iterative steps, thus reducing processing time.
3Reliability
If non-linear iterative filtering based on central limit-value record is used, then Gaussian filter characteristic is achieved, but the method is rejected by radiologists as it does not correspond to normal image impression
Solution Approach 1:
The patent creates different filtering behaviors for different regions: homogeneous areas receive stronger smoothing while areas with structures or edges receive lighter filtering. This local adaptation preserves the natural appearance of anatomical structures while still providing noise reduction, making the image acceptable to radiologists.
Solution Approach 2:
The filtering method transitions from static uniform filtering to dynamic adaptive filtering where the filtering strength changes based on local variance. This dynamic adjustment allows the filter to preserve important diagnostic features while reducing noise, maintaining natural image appearance.
4Reliability
If non-defined two-dimensional filters are calculated explicitly for each voxel, then filtering is performed, but the computation complexity becomes unreasonably high
Solution Approach 1:
The patent segments the filtering process into two stages: first calculating local variance for each voxel (which is computationally simple), then using these variance values to determine filtering strength. This segmentation avoids the need to calculate complex non-defined filters for each voxel while still achieving adaptive filtering performance.
Solution Approach 2:
The patent changes from using complex non-defined filters to using simple linear filters with dynamically adjusted parameters (filtering strength based on local variance). This parameter-based approach maintains filtering effectiveness while dramatically reducing computation complexity.
5Reliability
If filtering is applied to reduce noise, then noise is suppressed, but the blooming effect causes plaques with high CT value to appear larger and vessel diameters to be measured incorrectly
Solution Approach 1:
The patent applies different filtering strengths to different regions based on local variance. Areas with high variance (such as plaque boundaries and vessel walls) receive lighter filtering that preserves sharp transitions and accurate boundaries, while homogeneous areas receive stronger filtering for noise reduction. This local adaptation prevents the blooming effect from causing measurement errors in vessel diameters.
Data Source
AI summary
A filter method for tomographic 3D displays is disclosed, in which a volume model is used for display, which reproduces the volume of the examination object in the form of a large number of three-dimensional image voxels, and the image value of each voxel reproduces one object-specific characteristic of the examination object in this volume. According to the method, the original image voxels are processed using a 2D filter which is the same over the entire image area, and two different linear filters with selected directions which are obtained from the extremes of the previously calculated variances thus resulting in three data records with differently filtered image voxels, and in which, furthermore, the original image voxels and the filtered image voxels are mixed using local weights to form a result image. In addition, original image data can be processed using a steepening linear filter with a filter direction in the direction of the maximum local variance, resulting in a data record which is mixed into the final image with locally different weighting.