Comparison of one-dimensional and two-dimensional least-squares strain estimators for phased array displacement data

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Ultrasound Elastography

Background:

• Ultrasound elastography quantifies tissue stiffness.
• Accurate strain estimation is critical for diagnostic imaging.
• One-dimensional (1D) and two-dimensional (2D) least-squares strain estimators (LSQSE) are common methods.

Purpose of the Study:

• To compare the performance of 1D and 2D LSQSE.
• To evaluate the impact of kernel size on LSQSE performance.
• To assess strain estimation using simulated ultrasound data.

Main Methods:

• Simulated raw frequency data from a hard lesion/soft tissue model.
• Performance evaluation using root-mean-squared error (RMSE), elastographic signal-to-noise ratio (SNRe), and contrast-to-noise ratio (CNRe).
• Analysis of 1D and 2D LSQSE with varying kernel sizes.

Main Results:

• 2D LSQSE demonstrated superior performance over 1D LSQSE for phased array data, particularly at larger insonification angles.
• 2D LSQSE allowed processing of unfiltered displacement data, improving lateral/horizontal strain components.
• SNRe and CNRe analysis indicated enhanced precision with minimal contrast loss using 2D LSQSE.
• Optimal 2D kernel size is region-dependent and should be limited to preserve resolution.

Conclusions:

• 2D LSQSE offers significant advantages in accuracy and precision for ultrasound elastography.
• Kernel size selection is a critical parameter for optimizing 2D LSQSE performance and maintaining image resolution.
• These findings support the use of 2D LSQSE for improved tissue stiffness characterization.