Tissue Electrical Property Estimation Using MR Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional MRI-based methods for estimating tissue electrical properties, such as conductivity and permittivity, face challenges due to noise in RF magnetic field data, leading to poor resolution and inaccurate results, as they rely on approximating the phase of the transmit RF magnetic field and discarding non-physical values.
Innovation Solution
A method using least squared error estimation to determine electrical properties by segmenting MR images into sub-regions of constant properties, generating complex values from transmit and receive RF magnetic field data, and applying these estimates to calculate permittivity and conductivity with improved noise immunity and resolution.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional Laplacian operation is used to calculate electrical properties from B1+ data, then the measurement can be performed, but noise in the data leads to poor results and low resolution
Solution Approach 1:
The image is divided into multiple sub-regions, and the Laplacian operation is performed separately in each sub-region rather than globally. This segmentation allows the calculation to be more robust to local noise variations while maintaining measurement precision. Each sub-region's electrical properties are determined independently, reducing the impact of noise on overall results.
Solution Approach 2:
Multiple B1+ data measurements are combined through averaging or other statistical methods before performing the Laplacian operation. This merging of multiple measurements reduces the impact of random noise, leading to more reliable electrical properties estimates while maintaining measurement accuracy.
2Reliability
If skip factors are used to increase SNR by considering data points far apart, then noise immunity improves, but image resolution is reduced
Solution Approach 1:
The calculation domain is segmented into multiple sub-regions, allowing the use of smaller skip factors within each sub-region. This maintains local resolution while the segmentation itself provides noise immunity through distributed sampling. The fine-grained segmentation enables high-resolution results without requiring large skip factors that would reduce resolution.
Solution Approach 2:
The problem is shifted from a single global calculation to multiple local calculations across the spatial dimension. By performing Laplacian operations in multiple sub-regions rather than one global operation, the method achieves both noise immunity (through multiple samples) and high resolution (through fine-grained spatial processing).
3Stability of the object's composition
If non-physical values from noise are discarded and replaced with average values, then measurement stability improves, but resolution and accuracy are reduced
Solution Approach 1:
By segmenting the image into sub-regions and performing calculations locally, the method maintains stability through consistent local processing while preserving resolution. Each sub-region's electrical properties are calculated independently, avoiding the need to discard or replace values with averages, thus maintaining both stability and precision.
Solution Approach 2:
The approach changes from global parameter estimation to local parameter estimation. By calculating electrical properties in each sub-region separately, the method maintains measurement stability through local consistency while preserving spatial resolution and accuracy, avoiding the need to replace values with global averages.
4Reliability
If smoothing of B1+ data is applied to remove noise, then measurement reliability improves, but image resolution is degraded
Solution Approach 1:
The method segments the data processing into local sub-region operations rather than applying global smoothing. This allows noise reduction through local statistical processing while maintaining the sharp boundaries and fine details that would be blurred by global smoothing operations, thus preserving resolution while improving reliability.
Solution Approach 2:
The noise is extracted and removed through statistical processing of multiple measurements before the Laplacian operation, rather than applying smoothing filters to the final image. This extraction of noise from the raw data maintains measurement reliability while preserving the resolution of the resulting electrical properties map.
Data Source
AI summary
Exemplary embodiments of the present disclosure are directed to estimating an electrical property of tissue using MR images. Complex values having real components and imaginary components are generated and are associated with pixels in one or more MR images that corresponding to a region of tissue for which the electrical property is constant. An estimated value of the electrical property for the region of tissue is determined based on a least squared error estimation applied to the complex values.