Cookies on this website

We use cookies to ensure that we give you the best experience on our website. If you click 'Accept all cookies' we'll assume that you are happy to receive all cookies and you won't see this message again. If you click 'Reject all non-essential cookies' only necessary cookies providing core functionality such as security, network management, and accessibility will be enabled. Click 'Find out more' for information on how to change your cookie settings.

OBJECTIVE: Despite decades of obstetric scanning, the study of sonographer workflow remains largely unexplored. In the second trimester for example, sonographers use scan guidelines to guide their acquisition of standard planes and structures; however, the scan acquisition order is not prescribed. Using deep learning-based video analysis, the aim of this study was to develop a deeper understanding of the clinical workflow sonographers undertake during second-trimester anomaly scans. METHODS: We prospectively collected full-length video recordings of routine second-trimester anomaly scans. Important scan events in the videos were identified by automatically detecting image freeze and image/clip save. The video immediately preceding the important events was extracted and labelled as one of 11 commonly acquired anatomical structures. We developed and used automatic labelling of the large number of scan events using a purposely trained and tested deep-learning annotation model. Thus, anomaly scans were partitioned as a sequence of anatomical planes or fetal structures obtained over time. RESULTS: A total of 496 anomaly scans performed by 14 sonographers were available for analysis. The UK guidelines specify that an image or video clip of five different anatomical regions must be stored, and these were detected in the majority of scans: head and brain 97.2%, coronal face view 86.1%, abdomen 93.1%, spine 95%, and femur 92.3%. Analyzing the clinical workflow, we observed that sonographers begin their scan by capturing the head/brain, spine, or thorax-heart in 24.4%, 23.2%, or 22.8% of the scans, respectively. The most identified structure transitions were placenta-amniotic fluid to maternal anatomy, head-brain to coronal face (nose/lips), abdomen to thorax-heart, and 3D/4D face to sagittal face in 44.5%, 42.7%, 26.1%, and 23.7% of the scans, respectively. Three or more consecutive structure sequences were uncommon, occurring in up to 13% of the scans. No consistent sequence of an entire anomaly scan was identified among the captured scans. CONCLUSIONS: We present a novel analysis of the anomaly scan acquisition process based on deep learning-based analysis of ultrasound video. We note a wide variation in the number and sequence of structures obtained during routine second-trimester anomaly scans. Overall, each anomaly scan was found to be unique in its scanning sequence, suggesting that sonographers take advantage of the fetal position and acquire the standard planes according to their visibility rather than following a strict acquisition order. This article is protected by copyright. All rights reserved.

Original publication




Journal article


Ultrasound Obstet Gynecol

Publication Date



anatomy, artificial intelligence, automation, big data, clinical workflow, computer vision, data science, deep learning, image analysis, machine learning, neural network, obstetrics, pregnancy, screening, sonography, ultrasound