SNR Weighting for Shear Wave Speed Reconstruction in Tomoelastography

Area of Science:

• Medical Imaging
• Biophysics
• Signal Processing

Background:

• Quantitative MRI elastography (MRE) combines data from multiple directions and frequencies to generate wave speed maps.
• Weighting schemes are essential for combining these reconstructions effectively.
• Existing amplitude-weighting methods can introduce bias, especially in low signal-to-noise ratio (SNR) scenarios.

Purpose of the Study:

• To propose and validate a novel SNR-maximizing weighting scheme for MRE wave speed map reconstruction.
• To analytically derive and experimentally verify the SNR of reconstructed wave speed maps.
• To assess the impact of the proposed weighting on accuracy, spatial resolution, and SNR performance.

Main Methods:

• Developed an SNR-maximizing weighting scheme using the squares of estimated SNR values.
• Formulated the SNR of reconstructed wave speed maps using an analytical approach, assuming high SNR and Gaussian noise.
• Verified analytical SNR derivations with Monte Carlo simulations (MCSs).
• Evaluated the weighting scheme on simulated phantoms and experimental 3D human abdomen MRE data.

Main Results:

• The SNR-weighted inversion improved wave speed map SNR by a factor of two compared to amplitude-weighted reconstruction.
• The proposed method introduced no bias in low SNR conditions, unlike amplitude weighting which caused ~10% bias.
• Accuracy and spatial resolution remained unaffected by the SNR weighting scheme.

Conclusions:

• The proposed SNR-maximizing weighting scheme significantly enhances MRE wave speed map quality.
• This method offers a robust improvement in SNR performance without compromising image accuracy or resolution.
• The SNR weighting is particularly beneficial in low SNR MRE data, reducing reconstruction bias.