Deep learning-based passive elastography using U-Net architecture

Maud Legrand

doi:10.1063/5.0302399

Deep learning-based passive elastography using U-Net architecture

2026-02-21

0

PRE

AI

Emmanuel Martins Seromenho

DOI：10.1063/5.0302399

Original

Original request for help

Save

Abstract

En 中文

Since the early days of medical practice, assessing tissue stiffness has been a key component in evaluating tissue health. To estimate this parameter quantitatively and non-invasively, a variety of elastography techniques have been developed. Among them, methods based on the estimation of local shear wave speed have yielded highly promising results in ultrasound, MRI, and optical imaging modalities. In this paper, we introduce a proof-of-concept study that combines a deep learning approach with noise correlation elastography to estimate mechanical properties from a diffuse shear wave field. While the focus of this work is limited to stiffness estimation, the proposed framework can be extended to other mechanical parameters, such as anisotropy or viscoelasticity.

Keywords:

NOISE-CORRELATION

TOMOGRAPHY

TISSUES

Deep learning-based passive elastography using U-Net architecture

Abstract

Journal

Organization