In fetal ultrasound screening, Doppler images on the umbilical artery (UA) are important for monitoring blood supply through the umbilical cord. However, to capture UA Doppler images, a number of steps need to be done correctly: placing the gate at a proper location in the ultrasound image to obtain blood flow waveforms, and judging the Doppler waveform quality. Both of these rely on the operator's experience. The shortage of experienced sonographers thus creates a demand for machine assistance. We propose an automatic system to fill this gap. Using a modified Faster R-CNN we obtain an algorithm that suggests Doppler flow gate locations. We subsequently assess the Doppler waveform quality. We validate the proposed system on 657 scans from a national ultrasound screening database. The experimental results demonstrate that our system is useful in guiding operators for UA Doppler image capture and quality assessment.

本研究提出了一种自动系统，使用改进的Faster R-CNN算法协助超声医生对胎儿超声筛查中脐动脉峡部多普勒图像的获取和质量评估。对国家超声筛查数据库中657个样本进行实验验证，结果表明该系统的使用可以有效指导超声医生，提高脐动脉多普勒图像的质量。

脐动脉多普勒成像自动引导和质量评估系统