DopUS-Net: Quality-Aware Robotic Ultrasound Imaging Based on Doppler Signal

Medical ultrasound (US) is widely used to evaluate and stage vascular diseases, in particular for the preliminary screening program, due advantage of being radiation-free. However, automatic segmentation small tubular structures (e.g., ulnar artery) from cross-sectional US images still challenging. To address this challenge, paper proposes DopUS-Net a vessel re-identification module that leverage Doppler effect enhance final result. Firstly, combines with B-mode increase accuracy robustness blood vessels. It incorporates two encoders exploit maximum potential signal recurrent neural network modules preserve sequential information. Input first encoder two-channel duplex image representing combination grey-scale ensure anatomical spatial correctness. The second operates on pure provide region proposal. Secondly, benefiting signal, work introduces an online artery qualitatively real-time results automatically optimize probe pose enhanced images. This quality-aware enables closed-loop control robotic further improve confidence segmentation. experimental demonstrate proposed approach process can significantly (Dice score: $0.54$ notation="LaTeX">$0.86$ ; intersection over union: notation="LaTeX">$0.47$ notation="LaTeX">$0.78$ ). Note Practitioners —The important diagnosis disease, e.g., peripheral arterial clinical practices, nevertheless it not similar significance state-of-the-art examination systems yet. explores various effectively extract vessels by incorporating into process. DopUS structure extracting differentiated information different inputs fusing latent feature representations bottleneck layer also inspire other tasks like multi-senor fusion. In addition, developed Doppler-based tracker assess quality real-time. assessment subsequently screening. Preliminary physical experiments suggest system autonomous results. future, could be support diagnosis. We believe development large-scale programs. will only benefit limb arteries but structures, carotid or aorta.