[Paper Review] Morphology-, Noise-, and Resolution-Robust Ultrasound Elasticity Imaging with Fourier Neural Operators

Heekyu Kim, Hugon LEe+2 more|arXiv (Cornell University)|Jan 21, 2026

Ultrasound Imaging and Elastography0 citations

TL;DR

This paper investigates applying Fourier Neural Operators (FNO) to ultrasound elasticity imaging, evaluating robustness to morphology, noise, and resolution, and showing FNO outperforms baselines in simulations.

ABSTRACT

Ultrasound-based elasticity imaging is a non-invasive technique for estimating tissue stiffness fields from displacement fields obtained by comparing ultrasound signals before and after compression. While recent deep learning approaches have enabled faster and more accurate elasticity estimation compared to traditional methods, several challenges remain for clinical translation. In this study, we employ finite element simulations of free-hand palpation to investigate the applicability of the Fourier neural operator (FNO). Four practical scenarios were investigated: (1) prediction across diverse lesion morphologies, (2) generalization to cases with lesion counts differing from those in the training data, (3) robustness to noise in measured displacement fields, and (4) resilience to variations in ultrasound device resolution. Across these tasks, FNO consistently outperformed baseline models such as U-Net and DeepONet in predictive accuracy and generalization, while maintaining robustness under noise and resolution changes. Validated through in silico simulations, these findings demonstrate the potential of FNO as a framework that could facilitate translation of elasticity imaging toward clinical practice.

Motivation & Objective

• Investigate applicability of Fourier Neural Operators for ultrasound elasticity imaging.
• Assess generalization across lesion morphologies and lesion counts.
• Evaluate robustness to noisy displacement measurements.
• Assess resilience to ultrasound device resolution changes.

Proposed method

• Use finite element simulations of free-hand palpation to generate data.
• Apply Fourier Neural Operators to map displacement fields to elasticity fields.
• Compare FNO performance against U-Net and DeepONet across scenarios.
• Test generalization to varying lesion morphologies and counts.
• Assess robustness to measurement noise and resolution variations.

Experimental results

Research questions

• RQ1 Can FNO accurately estimate tissue elasticity from displacement fields across diverse lesion morphologies?
• RQ2 Does FNO generalize when lesion counts differ from training data?
• RQ3 Is FNO robust to noise in measured displacement fields?
• RQ4 Is FNO robust to variations in ultrasound device resolution?

Key findings

• FNO consistently outperforms baseline models (U-Net and DeepONet) in predictive accuracy and generalization across tasks.
• FNO maintains robustness under noise in displacement measurements.
• FNO demonstrates resilience to variations in ultrasound device resolution.
• In silico simulations validate the potential of FNO for clinical translation of elasticity imaging.