Dictionary Representations for Electrode Displacement Elastography.

Area of Science:

• Medical imaging
• Biomedical engineering
• Signal processing

Background:

• Ultrasound electrode displacement elastography (EDE) shows promise for monitoring microwave ablation in patients.
• Decorrelation noise artifacts commonly degrade displacement estimates in EDE, hindering accurate monitoring.
• Effective denoising is crucial for reliable displacement estimation in EDE.

Purpose of the Study:

• To propose and evaluate a global dictionary learning approach for denoising displacement estimates in EDE.
• To improve the signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) of displacement maps.
• To enhance the delineation of ablated regions using denoised EDE data.

Main Methods:

• Developed a global dictionary learning algorithm using adaptively learned dictionaries from EDE phantom displacement maps.
• The algorithm sparsely represents low-noise displacement patches and averages remaining patches for denoising.
• Compared denoised displacements against initial, median-filtered, and spline-smoothed displacements.

Main Results:

• Dictionary-represented displacements showed significantly higher SNR, CNR, and improved contrast compared to other methods.
• Quantitative improvements included 1.77 dB higher SNR, 4.56 dB higher CNR, and 4.35 dB improved contrast.
• The method successfully reduced noise while preserving important edge information and improving phantom inclusion delineation.
• Validation using in vivo patient data confirmed the efficacy of the dictionary learning approach.

Conclusions:

• Global dictionary learning offers an effective strategy for denoising displacement estimates in EDE.
• This approach significantly enhances image quality and ablated region delineation with minimal degradation.
• Dictionary learning represents a promising new application for improving EDE-based monitoring of minimally invasive procedures.