Ultrasound Motion Estimation Using Iterative Phase-Shift Refinement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current tissue motion estimation techniques in ultrasound imaging, such as block matching and phase-shift estimation, face challenges including high sampling frequency requirements, susceptibility to false minimums and peaks, memory bottlenecks, and increased variance in motion estimates, which affect the accuracy and reliability of elastography imaging.
Innovation Solution
An iterative motion processor that utilizes phase-shift estimation with cross-correlation and envelope compression to refine tissue motion estimates, reducing noise artifacts and improving precision, while also employing phase unwrapping and mean frequency estimation to enhance the accuracy of motion maps and reduce zebra artifacts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If block matching or phase-shift estimation is used for tissue motion estimation, then motion can be estimated, but high sampling frequency (eight times carrier frequency) is required which is four times the Nyquist requirement
Solution Approach 1:
The patent changes the parameter of sampling frequency by using a reduced sampling rate that meets or exceeds the Nyquist rate instead of the previously required eight-times carrier frequency. This parameter change maintains motion estimation capability while reducing the sampling burden by a factor of four.
2Ease of operation
If sum of absolute differences or sum of squared differences algorithms are used for block matching, then motion estimation can be performed, but the techniques are susceptible to identifying false minimums
Solution Approach 1:
The patent applies feedback through an iterative refinement process where initial motion estimates are obtained using simple block matching, then these estimates are used to guide subsequent phase-shift estimation iterations. Each iteration refines the motion estimate by eliminating false minimums through phase information, with the process continuing until convergence to the true minimum is achieved.
3Measurement precision
If maximum cross-correlation approach is used for block matching, then motion estimation can be performed, but the approach is susceptible to identifying false peaks especially when performed on RF data
Solution Approach 1:
The patent introduces phase information as an intermediary that mediates between the cross-correlation peak identification and the final motion estimate. By using phase-shift estimation on RF data, the method identifies the true correlation peak through phase continuity constraints, eliminating false peaks that plague traditional maximum cross-correlation approaches.
4Measurement precision
If polynomial approximation is fitted to maximum cross-correlation or sum of absolute differences functions for sub-sample precision, then sub-sample precision can be achieved, but large amounts of data must be transferred which requires increased RAM and memory access becomes the bottleneck
Solution Approach 1:
The patent extracts only the essential phase information from the RF data at each iteration, rather than processing and storing large amounts of raw data. By working with phase angles and incremental motion estimates, the method achieves sub-sample precision while minimizing memory requirements and eliminating the memory access bottleneck associated with polynomial approximation methods.
5Measurement precision
If phase shift estimation is used with narrow-band signal assumption, then displacement can be estimated from phase of complex correlation function, but the method is susceptible to aliasing and requires robust phase unwrapping algorithm
Solution Approach 1:
The patent applies preliminary action by using block matching to obtain coarse motion estimates before performing phase-shift estimation. These preliminary estimates serve as initial conditions that constrain the phase unwrapping process, allowing the algorithm to correctly interpret phase jumps and eliminate aliasing artifacts without requiring complex post-processing.
6Reliability
If iterative phase-shift estimation with cross-correlation is used to refine motion estimates, then noise artifacts are reduced and precision is improved, but the processing complexity increases
Solution Approach 1:
The patent applies dynamics by making the processing complexity adaptive rather than static. The iterative phase-shift estimation process dynamically adjusts the amount of processing required based on the convergence behavior of the motion estimates. The algorithm continues iterating only as long as meaningful refinement is occurring, automatically reducing complexity when convergence is achieved while maintaining high reliability.
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
The solution achieves smooth motion maps with reduced noise and increased contrast, enabling high-precision tissue motion estimation with reduced data sets, improving the reliability of elastography imaging by addressing the limitations of existing methods.
Implementation Method 1
The fundament for phase-shift displacement estimation is the assumption that the measured signals are narrow-band and can be described as x(t) = a(t) exp(-jωt)... The time shift can be found from the phase of the complex correlation function R12(0): φm = ∠(R12(0)) = -jωtm
Implementation Method 2
The lag-zero cross correlation between x1(t) and x2(t) can be estimated as: R12(0) = ∫ a(t) exp(-jωt) · a(tm - t) exp(jωt - jωtm) dt
Implementation Method 3
An envelope compressor compresses the envelope, while preserving the phase, producing xic(m, n) and xjc(m, n)
Implementation Method 4
The phase is unwrapped using a standard or other unwrapping procedure... The phase unwrapper receives, as an input, an input matrix Φi and outputs a matrix Φo
Implementation Method 5
A beamformer 112 processes the received echoes, e.g., by applying time delays and weights to the echoes and summing the resulting echoes, producing an RF signal
Data Source
Figure 1~2
Figure 3~4
Figure 5
AI summary
A motion processor (118) includes a motion estimator (306) that iteratively estimates a motion between a pair of consecutive frames of pre-processed echoes, wherein the motion estimator (306) generates the estimated motion based on at least one iteration. A method includes iteratively estimating tissue motion. between a pair of consecutive frames of pre-processed echoes over at least one iteration.