Noise analysis and improvement of displacement vector estimation from angular displacements

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Ultrasound Elastography

Background:

• Elastography techniques typically measure strain along the insonification direction.
• Recent methods use angular displacement estimates from multiple insonifications for improved strain imaging.
• Existing angular strain estimation assumes independent and identically distributed noise, limiting accuracy.

Purpose of the Study:

• To develop a modified least-squares approach for elastography that does not assume noise distribution.
• To incorporate displacement noise artifacts into strain estimation using a cross-correlation matrix.
• To improve the accuracy of axial and lateral strain tensor estimation in ultrasound elastography.

Main Methods:

• A modified least-squares approach was developed to estimate strain tensors.
• The method incorporates displacement noise artifacts using a cross-correlation matrix.
• Two techniques for estimating noise artifacts from displacement images were employed.

Main Results:

• The modified approach demonstrated improved strain tensor estimation performance.
• Improvements were validated using finite-element analysis simulations and experimental phantom data.
• Quantified improvements were shown in elastographic signal-to-noise and contrast-to-noise ratios.

Conclusions:

• The proposed method enhances strain tensor estimation accuracy in elastography.
• This technique effectively addresses limitations of previous methods by not assuming noise distribution.
• The findings suggest a significant advancement for quantitative elasticity imaging.