English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Robotic ultrasound (US) systems have shown great potential to make USexaminations easier and more accurate. Recently, various machine learningtechniques have been proposed to realize automatic US image interpretation forrobotic US acquisition tasks. However, obtaining large amounts of real USimaging data for training is usually expensive or even unfeasible in someclinical applications. An alternative is to build a simulator to generatesynthetic US data for training, but the differences between simulated and realUS images may result in poor model performance. This work presents a Sim2Realframework to efficiently learn robotic US image analysis tasks based only onsimulated data for real-world deployment. A style transfer module is proposedbased on unsupervised contrastive learning and used as a preprocessing step toconvert the real US images into the simulation style. Thereafter, atask-relevant model is designed to combine CNNs with vision transformers togenerate the task-dependent prediction with improved generalization ability. Wedemonstrate the effectiveness of our method in an image regression task topredict the probe position based on US images in robotic transesophagealechocardiography (TEE). Our results show that using only simulated US data anda small amount of unlabelled real data for training, our method can achievecomparable performance to semi-supervised and fully supervised learningmethods. Moreover, the effectiveness of our previously proposed CT-based USimage simulation method is also indirectly confirmed.

Style Transfer Enabled Sim2Real Framework for Efficient Learning of Robotic Ultrasound Image Analysis Using Simulated Data