Phase-Sensitive Structural Similarity for MR Image Denoising
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image processing methods for magnetic resonance (MR) images discard phase information, leading to suboptimal noise distribution and biased parameter estimation, as they rely solely on magnitude images, which lose phase data and introduce noise-related biases.
Innovation Solution
The use of a phase-sensitive structural similarity index measure (PS-SSIM) that considers both the magnitude and phase information of complex MR images, allowing for the training of neural networks to reconstruct and enhance MR images by incorporating phase-sensitive comparisons.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If phase information is discarded and only magnitude images are used, then image processing becomes simpler and more compatible with conventional methods, but noise distribution becomes suboptimal and parameter estimation becomes biased
Solution Approach 1:
The patent changes the parameter space from real-valued magnitude images to complex-valued images preserving phase information. By transforming the image representation from magnitude-only to magnitude-phase preserving format, the system enables accurate parameter estimation while maintaining compatibility with neural network processing frameworks through complex number operations.
2Device complexity
If phase information is discarded, then computational complexity is reduced, but image quality deteriorates due to noise distribution changes and loss of structural information
Solution Approach 1:
The patent replaces conventional real-valued image processing mechanisms with complex-valued processing. By substituting magnitude-only operations with complex number operations that preserve phase, the system achieves superior noise distribution characteristics and image quality while maintaining computational efficiency through optimized complex arithmetic.
3Adaptability or versatility
If conventional magnitude-only processing is used, then compatibility with existing imaging modalities is maintained, but structural similarity and phase information are lost
Solution Approach 1:
The patent creates a composite image representation that combines magnitude and phase information into a single complex-valued image structure. This composite format maintains compatibility with existing MRI processing pipelines while preserving the phase information that would otherwise be lost, enabling both structural and phase-based analysis.
Data Source
AI summary
A computer-implemented method of processing complex magnetic resonance (MR) images is provided. The method includes receiving a pair of corrupted complex data and pristine complex images. The method also includes training a neural network model using the pair by inputting the corrupted complex data to the neural network model, setting the pristine complex images as target outputs, and processing the corrupted complex data using the neural network model to derive output complex images of the corrupted complex data. Training a neural network model also includes comparing the output complex images with the target outputs by computing a phase-sensitive structural similarity index measure (PS-SSIM) between each of the output complex images and its corresponding target complex image, wherein the PS-SSIM is real-valued and varies with phases of the output complex image and phases of the target complex image, and adjusting the neural network model based on the comparison.