DKI Parameter Estimation Using Separated Least Square Fitting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In diffusion kurtosis imaging (DKI) analysis, existing methods face challenges in stabilizing and speeding up non-linear least square fitting for diffusion-related parameter estimation, often resulting in blurring of images due to the application of smoothing filters and repeated calculations with constrained processing.
Innovation Solution
Separate the least square fitting from constraint processing, correcting only pixel values that do not meet the constraint condition, and re-estimating diffusion-related parameters using the corrected values to generate parameter images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a smoothing filter is applied to stabilize the non-linear least square fitting calculation, then the calculation stability is improved, but the diffusion weighted image undergoes blurring which reduces image quality
Solution Approach 1:
The patent segments the image processing into two distinct stages: first applying a smoothing filter to stabilize the non-linear least square fitting calculation, then applying a deconvolution filter to the resulting parameter image to restore image quality. This segmentation allows each processing stage to optimize for its specific purpose without compromising the other.
Solution Approach 2:
The patent introduces an intermediary parameter image (such as ADC or kurtosis coefficient map) that serves as a bridge between the diffusion weighted images and the final output. The smoothing filter is applied during parameter estimation, and then a deconvolution filter is applied to the parameter image to compensate for blurring, effectively using the parameter image as an intermediary that can be processed separately from the original diffusion weighted images.
2Reliability
If constrained non-linear least square fitting is used to stabilize the calculation, then the calculation stability is improved, but the processing time is extended due to repeated calculations
Solution Approach 1:
The patent applies a smoothing filter as a preliminary action before performing the non-linear least square fitting calculation. This preprocessing step stabilizes the input data, allowing the subsequent constrained fitting process to converge faster and with fewer iterations, thereby reducing the overall processing time while maintaining calculation stability.
Solution Approach 2:
The patent employs an iterative feedback mechanism where the constrained non-linear least square fitting is performed, the results are evaluated against constraint conditions, and if constraints are not satisfied, the calculation is repeated with adjusted parameters. This feedback loop ensures calculation stability while minimizing unnecessary repeated calculations by efficiently converging to a valid solution.
3Reliability
If the non-linear least square fitting is performed with constraint conditions, then the calculation stability is improved, but the number of repeated calculations increases which reduces processing speed
Solution Approach 1:
The patent applies partial constraint conditions selectively rather than enforcing all constraints uniformly across all pixels. By identifying which constraint conditions are most critical for calculation stability and applying those preferentially, the method achieves sufficient calculation stability with fewer repeated calculations, thereby improving processing speed.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enables the generation of high-quality images at high speed in DKI analysis by minimizing blurring and reducing the number of repetitive calculations, thus enhancing image quality and processing speed.
Implementation Method 1
An MRI apparatus is a medical image acquisition system that utilizes mainly nuclear magnetic resonance phenomenon of protons
Implementation Method 2
the MPG pulse induces reduction of signal intensity due to phase dispersion on a nuclear spin that moves randomly
Implementation Method 3
Since the nuclear spin that diffuses in the MPG pulse application direction causes the reduction of signal intensity due to the phase dispersion
Data Source
AI summary
The present invention provides a technique for obtaining a high-quality image at high speed in DKI analysis. In the DKI analysis, upon estimating a parameter relating to diffusion in an application direction of an MPG pulse, a least square fitting is separated from a constraint processing, and only a value of the pixel that does not meet the constraint condition in the least square fitting is targeted for the correction. Then, with regard to this pixel, a diffusion-related parameter is re-estimated using the pixel value after the correction, and a parameter image is generated by using the diffusion-related parameter thus obtained.