Toward point-of-care ultrasound estimation of fetal gestational age from the trans-cerebellar diameter using CNN-based ultrasound image analysis

Abstract

Obstetric ultrasound is a fundamental ingredient of modern prenatal care with many applications including accurate dating of a pregnancy, identifying pregnancy-related complications, and diagnosis of fetal abnormalities. However, despite its many benefits, two factors currently prevent wide-scale uptake of this technology for point-of-care clinical decision-making in low-and middle-income country (LMIC) settings. First, there is a steep learning curve for scan proficiency, and second, there has been a lack of easy-to-use, affordable, and portable ultrasound devices. We introduce a framework toward addressing these barriers, enabled by recent advances in machine learning applied to medical imaging. The framework is designed to be realizable as a point-of-care ultrasound (POCUS) solution with an affordable wireless ultrasound probe, a smartphone or tablet, and automated machine-learning-based image processing. Specifically, we propose a machine-learning-based algorithm pipeline designed to automatically estimate the gestational age of a fetus from a short fetal ultrasound scan. We present proof-of-concept evaluation of accuracy of the key image analysis algorithms for automatic head transcerebellar plane detection, automatic transcerebellar diameter measurement, and estimation of gestational age on conventional ultrasound data simulating the POCUS task and discuss next steps toward translation via a first application on clinical ultrasound video from a low-cost ultrasound probe.

Artificial Intelligence in Pregnancy: A Scoping Review

“…• Prediction of gestational age [47] (model), [168] (framework). • Pregnancy detection [169] (ML model).…”

The PRECISE (PREgnancy Care Integrating translational Science, Everywhere) Network’s first protocol: deep phenotyping in three sub-Saharan African countries

“…Whenever possible, at the booking visit women will be approached for participation. Consenting women will have their pregnancies dated using, in order of accuracy, ultrasound dating (recognised to be very uncommon in these settings), certain last menstrual period, estimated last menstrual period and symphysis-fundal height (supplemented, for study purposes, by the ultrasound-based transcerebellar diameter (TraCer) device that determines gestational age using the highly-conserved transcerebellar diameter [41]); the TraCer dating will be used as the gold standard for the academic enterprise. As the TraCer device does not yet have a CE mark it is not appropriate that these data be used to direct clinical care; however, if during the scan there are concerns about the fetus or pregnancy, the women will be referred down the standard clinical care pathway for further investigation.…”

Cross-Device Cross-Anatomy Adaptation Network for Ultrasound Video Analysis

“…To address this challenge, recent academic research on solutions for LMIC setting has proposed a three-component approach: i) the adoption of inexpensive and portable US equipment, ii) designing simple-to-use US scanning protocols, e.g. the obstetric sweep protocol (OSP) [1] and iii) innovating intelligent image analysis algorithms [6,10,9]. Arguably, the third innovation plays a bridging role in this approach, enabling the other two cost-effective components of the solution.…”

“…Even experienced sonographers can struggle to interpret and analyze data obtained using simple protocols on inexpensive machines, which are often equipped with older generation transducers and processing units. Furthermore this data degrades modern machine learning models compared to those developed with well-curated datasets [5,4,9]. Refining these existing models trained with site-specific data is an option, but requires additional data to be acquired and manually annotated, which may present substantial logistic challenges in expertise and cost.…”

“…1b, hereafter referred to as the single-sweep dataset. Classifying video frames in such single-sweep data into multiple anatomical classes is useful for assisting a range of clinical applications, including anomaly detection, gestational age estimation and pregnancy risk assessment [9].…”