Investigation of parametric spectral estimation techniques for elasticity imaging

Area of Science:

• Biomedical Engineering
• Signal Processing
• Medical Imaging

Background:

• Tissue strain estimation is crucial for diagnosing various medical conditions.
• Parametric spectral estimators offer potential advantages over traditional methods.
• Evaluating autoregressive (AR) and autoregressive moving average (ARMA) models is necessary.

Purpose of the Study:

• To evaluate AR and ARMA parametric spectral estimators for tissue strain estimation.
• To compare their performance against nonparametric discrete Fourier transform (DFT) and coherent elastography.
• To identify optimal model orders using a modified strain filter approach.

Main Methods:

• 1-D simulations and in vitro phantom experiments were conducted.
• Parametric spectral estimators (AR, ARMA) and DFT were implemented.
• A modified strain filter approach determined model orders, analyzing trade-offs.
• Elastographic signal-to-noise ratio (SNR(e)) and dynamic range were key performance metrics.

Main Results:

• The Yule-Walker AR spectral strain estimator demonstrated superior performance among parametric methods.
• No parametric method outperformed the DFT-based spectral strain estimator.
• Both spectral estimation techniques achieved higher SNR(e) and dynamic range than conventional elastography.
• The area under the strain filter served as a performance measure across a 0.1-50% strain range.

Conclusions:

• AR and ARMA parametric spectral estimators are viable for tissue strain estimation.
• The Yule-Walker AR estimator is a strong candidate within parametric approaches.
• Spectral estimation techniques offer significant improvements in SNR(e) and dynamic range for elastography.