Adaptively Weighted Anisotropic Diffusion for Medical Image Noise Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing medical imaging techniques, such as x-ray CT, face challenges in preserving clinically significant living body tissue structures during denoising, as low-pass filtering methods often remove high-frequency components and smooth edges, reducing image resolution and detectability of small structures.
Innovation Solution
The implementation of an adaptively weighted anisotropic diffusion (AWAD) method, which uses a diffusion coefficient dependent on image data to reduce noise while preserving edges and details of both large and small structures, by incorporating a properly weighted source term and generating adaptive weights based on edge maps to enhance image quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If low-pass filtering techniques are used to reduce noise, then noise is effectively reduced, but clinically significant living body tissue structures are lost due to smoothing edges and removing high-frequency components
Solution Approach 1:
The patent applies different filtering strengths to different regions of the image based on local characteristics. The filter adapts its behavior locally: stronger filtering in homogeneous regions and weaker filtering near edges and structures of interest, thereby reducing noise while preserving clinically significant tissue structures
Solution Approach 2:
The filtering technique transitions from static low-pass filtering to dynamic adaptive filtering. The filter parameters are adjusted dynamically based on local image statistics and edge detection, allowing the filtering strength to vary across different regions and iterations, thus preserving structures while reducing noise
2Manufacturing precision
If anisotropic diffusion filter is used to preserve living body tissue structures, then edges are preserved, but computationally complex implementation is required
Solution Approach 1:
The patent applies anisotropic diffusion iteratively with controlled parameters, performing partial diffusion actions in each iteration rather than one complex operation. This breaks down the computationally intensive task into manageable steps that can be optimized and parallelized
Solution Approach 2:
The patent dynamically adjusts diffusion parameters (such as diffusion coefficient and iteration count) based on local image characteristics. By changing parameters adaptively rather than using fixed values, the computational complexity is managed while maintaining edge preservation effectiveness
3Manufacturing precision
If prior art anisotropic smoothing techniques are used to preserve small living body tissue structures, then small structures are improved, but areas for no filtering are determined which may exclude potentially relevant structures
Solution Approach 1:
The patent incorporates feedback mechanisms where the filtering process continuously monitors image features and adjusts its behavior accordingly. Edge detection and structure analysis results feed back into the diffusion parameter selection, ensuring that potentially relevant structures are not excluded by premature filtering decisions
Solution Approach 2:
The patent performs preliminary edge detection and structure identification before applying the full diffusion process. This preliminary action identifies regions that should be protected from filtering, ensuring that small and potentially relevant tissue structures are preserved before the main denoising operation begins
Data Source
AI summary
In general, according to one embodiment, in an medical imaging apparatus including an anisotropic diffusion unit 131 and a weight generation unit 132. The anisotropic diffusion unit 131 executes anisotropic diffusion for medical image data constituted by a plurality of signal values. The weight generation unit 132 generates a plurality of weights respectively corresponding to the plurality of signal values based upon the medical image data. The anisotropic diffusion unit 131 executes the anisotropic diffusion for the medical image data by using the weights.