Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2024 Jun;165(3):1013-1021.
doi: 10.1002/ijgo.15321. Epub 2024 Jan 8.

Deep learning to estimate gestational age from fly-to cineloop videos: A novel approach to ultrasound quality control

Ambika V Viswanathan  1 , Teeranan Pokaprakarn  1   2 , Margaret P Kasaro  1   3 , Hina R Shah  4 , Juan C Prieto  4 , Chiraz Benabdelkader  1 , Yuri V Sebastião  1 , Ntazana Sindano  3 , Elizabeth Stringer  1   3 , Jeffrey S A Stringer  1   3
Affiliations

Deep learning to estimate gestational age from fly-to cineloop videos: A novel approach to ultrasound quality control

Ambika V Viswanathan et al. Int J Gynaecol Obstet. 2024 Jun.

Abstract

Objective: Low-cost devices have made obstetric sonography possible in settings where it was previously unfeasible, but ensuring quality and consistency at scale remains a challenge. In the present study, we sought to create a tool to reduce substandard fetal biometry measurement while minimizing care disruption.

Methods: We developed a deep learning artificial intelligence (AI) model to estimate gestational age (GA) in the second and third trimester from fly-to cineloops-brief videos acquired during routine ultrasound biometry-and evaluated its performance in comparison to expert sonographer measurement. We then introduced random error into fetal biometry measurements and analyzed the ability of the AI model to flag grossly inaccurate measurements such as those that might be obtained by a novice.

Results: The mean absolute error (MAE) of our model (±standard error) was 3.87 ± 0.07 days, compared to 4.80 ± 0.10 days for expert biometry (difference -0.92 days; 95% CI: -1.10 to -0.76). Based on simulated novice biometry with average absolute error of 7.5%, our model reliably detected cases where novice biometry differed from expert biometry by 10 days or more, with an area under the receiver operating characteristics curve of 0.93 (95% CI: 0.92, 0.95), sensitivity of 81.0% (95% CI: 77.9, 83.8), and specificity of 89.9% (95% CI: 88.1, 91.5). These results held across a range of sensitivity analyses, including where the model was provided suboptimal truncated fly-to cineloops.

Conclusions: Our AI model estimated GA more accurately than expert biometry. Because fly-to cineloop videos can be obtained without any change to sonographer workflow, the model represents a no-cost guardrail that could be incorporated into both low-cost and commercial ultrasound devices to prevent reporting of most gross GA estimation errors.

Keywords: artificial intelligence; biometry; deep learning; gestational age; quality control; ultrasound.

PubMed Disclaimer

Conflict of interest statement

Conflict of interest

The authors have no conflict of interest.

Figures

Figure 1:
Figure 1:. Components of a fly-to cineloop video
a The last frame of a fly-to cineloop is what the sonographer freezes and measures. b The frames immediately preceding the last frame are often very similar to the last frame. c The earlier frames in a fly-to cineloop may contain other structures adjacent to the structure of interest or they may contain no usable information
Figure 2:
Figure 2:. Study flow chart
After applying participant and visit-level exclusions, we created a training set to develop and tune the deep learning model and a test set to assess its performance. To be eligible for inclusion in a test set a participant must have been dated by a prior scan or in vitro fertilization (IVF). Additionally, a scan must include full biometry (head circumference, biparietal diameter, abdominal circumference, and femur length) as well as a full set of fly-to scans (head defined as either head circumference or biparietal diameter fly-to, abdominal circumference fly-to, and femur fly-to). TCD fly-to scans were not included in the test set. The test set was selected at random from among all eligible participants. Train and tuning sets were created from all remaining participants.
Figure 3.
Figure 3.. Sources of novice sonographer biometry error
Panel A shows optimal and suboptimal image capture. Panel B shows optimal image for measuring head circumference with correct caliper placement. Biometry errors can result from either incorrect caliper placement (Panel C) or from capturing an imperfect plane that distorts the structure of interest (Panel D).
Figure 4:
Figure 4:. Error of deep learning model, expert biometry, and simulated novice biometry in the test set
Dashed horizontal lines in each panel denote established error bounds of ultrasound biometry as defined by the American College of Obstetricians and Gynecologists.[11] Novice biometry simulated by introducing 7.5% average uniform random error into expert measurements (see Methods).
Figure 5:
Figure 5:
Receiver operating characteristics of deep learning model to identify 10-day discrepancy between novice and expert gestational age estimates
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

References

    1. WHO recommendations on antenatal care for a positive pregnancy experience. https://www.who.int/publications/i/item/9789241549912. Accessed 2023 Mar 30. - PubMed
    1. Whitworth M, Bricker L, Mullan C. Ultrasound for fetal assessment in early pregnancy. Cochrane Database Syst Rev. 2015. - PMC - PubMed
    1. Becker DM, Tafoya CA, Becker SL, Kruger GH, Tafoya MJ, Becker TK. The use of portable ultrasound devices in low- and middle-income countries: a systematic review of the literature. Trop Med Int Health. 2016;21(3):294–311. - PubMed
    1. Lamprecht H, Lemke G, van Hoving D, Kruger T, Wallis L. Poor return on investment: investigating the barriers that cause low credentialing yields in a resource-limited clinical ultrasound training programme. Int J Emerg Med. 2018;11(1). - PMC - PubMed
    1. Ginsburg AS, Liddy Z, Khazaneh PT, May S, Pervaiz F. A survey of barriers and facilitators to ultrasound use in low- and middle-income countries. Scientific Reports. 2023;13(1):1–11. - PMC - PubMed