A comparison of direct and iterative finite element inversion techniques in dynamic elastography.

Area of Science:

• Biomedical Engineering
• Medical Imaging
• Computational Mechanics

Background:

• Tissue elasticity imaging (elastography) requires solving an inverse problem to determine elasticity distribution from displacement measurements.
• The finite element method (FEM) is a prevalent technique for inverse problems in dynamic elastography, with both direct and iterative schemes available.

Purpose of the Study:

• To investigate and compare the performance of direct and iterative FEM schemes for solving the inverse problem in dynamic elastography.
• To evaluate the impact of data resolution and excitation frequency on elasticity estimation accuracy.
• To introduce and assess a novel iterative FEM requiring lower mesh density and compare two direct FEM derivative calculation methods (exact vs. least squares fit).

Main Methods:

• Comparative analysis of direct and iterative FEM schemes under varying excitation frequencies.
• Simulation-based study on spatial resolution of different FEM methods.
• Validation of computational findings using a phantom experiment with controlled noise levels (20 dB).

Main Results:

• The direct FEM with least squares fit demonstrates superior robustness to noise compared to iterative methods, achieving approximately half the RMS error in noisy homogenous regions.
• A critical ratio between voxel size and wavelength (0.1-0.2 for 20 dB noise) is identified for reliable elasticity estimation with the direct least squares method.
• Iterative FEM methods exhibit higher computational costs and dependency on initial guesses, while direct methods offer a more noise-resilient alternative.

Conclusions:

• The direct FEM with least squares fit is recommended for linear elasticity cases due to its noise robustness, lower computational demand, and independence from initial guesses.
• Optimizing data resolution and excitation frequency is crucial for accurate tissue elasticity estimation using FEM.
• The identified voxel size to wavelength ratio provides a guideline for reliable elastography results.