[Paper Review] 4D-UNet improves clutter rejection in human transcranial contrast enhanced ultrasound

Tristan Beruard, Armand Delbos+3 more|arXiv (Cornell University)|Feb 20, 2026

Ultrasound and Hyperthermia Applications0 citations

TL;DR

This work introduces a 4D U-Net for clutter rejection in transcranial 3D CEUS, leveraging spatial and temporal information to enhance microbubble detection in adult human data.

ABSTRACT

Transcranial ultrasound imaging is limited by high skull absorption, limiting vascular imaging to only the largest vessels. Traditional clutter filters struggle with low signal-to-noise ratio (SNR) ultrasound datasets, where blood and tissue signals cannot be easily separated, even when the echogenicity of the blood is improved with contrast agents. Here, we present a novel 4D U-Net approach for clutter filtering in transcranial 3D Contrast Enhanced Ultrasound (CEUS) exploiting spatial and temporal information via a 4D-UNet implementation to enhance microbubble detection in transcranial data acquired in human adults. Our results show that the 4D-UNet improves temporal clutter filters. By integrating deep learning into CEUS, this study advances neurovascular imaging, offering improved clutter rejection and visualization. The findings underscore the potential of AI-driven approaches to enhance ultrasound-based medical imaging, paving the way for more accurate diagnostics and broader clinical applications.

Motivation & Objective

• Address clutter in transcranial CEUS with low SNR datasets.
• Develop a 4D U-Net architecture that utilizes spatial and temporal information to improve microbubble detection.
• Validate the approach on transcranial CEUS data from adult humans and assess improvements in clutter filtering.

Proposed method

• Design a 4D-UNet architecture that extends 3D CEUS processing to include temporal information.
• Leverage spatial and temporal features across 3D CEUS frames to suppress clutter.
• Train the network to discriminate microbubble signals from tissue clutter within transcranial data.
• Integrate deep learning into CEUS processing to enhance visualization of vascular signals.

Experimental results

Research questions

• RQ1 Can a 4D-UNet improve clutter rejection in human transcranial CEUS compared to traditional methods?
• RQ2 Does incorporating temporal information enhance microbubble visualization under low SNR conditions?
• RQ3 How does the 4D-UNet affect the quality of CEUS visualization and potential diagnostic utility in neurovascular imaging?

Key findings

• The 4D-UNet improves temporal clutter filters in transcranial CEUS.
• The approach enhances microbubble detection in adult human transcranial data.
• Integration of deep learning into CEUS shows potential for improved clutter rejection and visualization.
• The study supports AI-driven improvements in neurovascular imaging and clinical diagnostics.