An adaptive displacement estimation algorithm for improved reconstruction of thermal strain

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Ultrasound Elastography

Background:

• Thermal strain imaging (TSI) differentiates arterial tissues but requires precise displacement estimation.
• Existing methods like Loupas' estimator and normalized cross-correlation (NXcorr) have limitations (phase-wrapping, low SNR performance).
• Accurate detection of small lipid pools in vivo is crucial for atherosclerotic plaque characterization.

Purpose of the Study:

• To develop and evaluate an adaptive displacement estimation algorithm for enhanced TSI.
• To combine the strengths of Loupas' estimator and NXcorr for improved accuracy and robustness.
• To assess the algorithm's performance in computer simulations and ex vivo human tissue.

Main Methods:

• Developed an adaptive algorithm integrating Loupas' estimator and NXcorr based on displacement magnitude and SNR.
• Validated the algorithm using 1-D simulations to compare bias and variance against individual estimators.
• Evaluated performance in computer simulations of TSI and an ex vivo human atherosclerotic artery sample.

Main Results:

• The adaptive estimator demonstrated reduced bias compared to Loupas' estimator and NXcorr in simulations.
• Strain reconstruction using adaptive estimates showed significant improvements: 43.7-350% increase in strain SNR.
• Spatial accuracy improved by 1.2-23.0% (P < 0.001), with ex vivo results comparable to simulations.

Conclusions:

• The novel adaptive displacement estimation algorithm enhances TSI accuracy and robustness.
• This combined approach overcomes limitations of individual estimators for better lipid pool detection.
• Improved strain reconstruction holds significant potential for in vivo atherosclerotic artery assessment.