SOURCE-LINKED INTELLIGENCE

Streamlining Clinical Applicability of Ultrasound Localization Microscopy via Advanced Reconstruction Models

CORDIS · observation · Publication date unknown

crobubbles (MBs) accurately, stretching the capabilities of current clinical ultrasound systems. ULMARM aims to solve these challenges by developing advanced reconstruction techniques and integrating deep learning (DL) models. The goal is to relax frame rate demands while maintaining high-quality super-resolved imaging and accurate motion tracking, making ULM more feasible for clinical applications. In ULMARM, I propose recasting ULM as a mathematical inverse problem to achieve super-resolution in both space and time jointly. By treating the relationship between MB movement and ultrasound frames as a generative forward model, I will infer the most probable MB trajectories from observations using a Bayesian posterior maximization approach. To tackle inherent challenges, I will utilize advanced deep generative models and strong data-driven priors to streamline the inference process. The ULMARM project is organized into six work packages (WPs). WP1 focuses on new sparse coding methods to relax frame rate while preserving image quality. WP2 develops mathematical models for direct reconst

Read original source ↗ Open in workspace

recordType

award

status

SIGNED

region

EU

value

217076.16

unit

EUR