Abdominal Circumference Operator-agnostic UltraSound measurement Banner
Join

Clinical and societal problem

Diagnosing fetal growth restriction is challenging in low-resource settings

Fetal growth restriction (FGR), affecting up to 10% of pregnancies, is a critical factor contributing to perinatal morbidity and mortality (1-3). Strongly linked to stillbirths, FGR can also lead to preterm labor, posing risks to the mother (4,5). This condition often results from an impediment to the fetus' genetic growth potential due to various maternal, fetal, and placental factors (6). Measurements of the fetal abdominal circumference (AC) as seen on prenatal ultrasound are a key aspect of monitoring fetal growth. When smaller than expected, these measurements can be indicative of FGR, a condition linked to approximately 60% of fetal deaths (4). FGR diagnosis relies on repeated measurements of either the fetal abdominal circumference (AC), the expected fetal weight, or both. These measurements must be taken at least twice, with a minimum interval of two weeks between them for a reliable diagnosis (7). Additionally, an AC measurement that falls below the third percentile is, by itself, sufficient to diagnose FGR (7-9). However, the routine practice of biometric obstetric ultrasounds, crucial for AC measurements, is limited in low-resource settings due to the high cost of sonography equipment and the scarcity of trained sonographers. 

AI-powered low-cost prenatal ultrasound for novice operators

The use of low-cost ultrasound devices and standardized blind-sweep protocols has been proposed for novice operators to acquire obstetric data in low-resource settings (10-12). Blind-sweep acquisition protocols are characterized by operators performing scans without viewing the ultrasound images. These protocols yield sequences of 2D ultrasound frames that are captured as the ultrasound probe follows specific trajectories across the gravid abdomen. Unlike traditional clinical sonography, where experienced sonographers search for the standard plane to conduct biometry measurements, blind-sweep data poses a distinct set of challenges. The quality of the image data is limited and may not contain the precise standard planes conventionally used for measurements (13).

Figure 1: The obstetric sweep protocol (DeStigter et al, 2011). An example standardized blind-sweep acquisition protocol for prenatal ultrasound.

Addressing these limitations, a growing body of literature focuses on the use of artificial intelligence (AI) to automate prenatal assessment tasks on free-hand ultrasound sequences acquired following standardized protocols, bypassing the need for expert sonographic interpretation. Such tasks include fetal biometry measurements (13,14), gestational age estimation (13,15-17) and pregnancy risk detection (14,15,18-22). These AI solutions have the potential to be embedded into mobile devices, offering a complete, offline, low-cost, and portable solution suitable for resource-limited settings, as demonstrated in (15,21). 

The ACOUSLIC-AI challenge

The ACOUSLIC-AI (Abdominal Circumference Operator-agnostic UltraSound measurement) challenge is a classification and segmentation MICCAI2024 challenge. It is the first challenge to propose the use of blind-sweep data for fetal biometry tasks. The goal is to develop and benchmark AI models for the automated measurement of fetal abdominal circumference on this specific data type, with the aim to broaden the accessibility of prenatal care in areas with limited resources.  Participants in this challenge will develop AI models to estimate AC in blind sweep 2D prenatal abdominal ultrasound sequences, acquired by novice operators in five African peripheral healthcare units and one European hospital. The models must identify the optimal frame for measurement and accurately segment the fetal abdomen within that frame. They must provide the identified frame and the corresponding segmentation mask, which will be used to precisely measure the fetal abdominal circumference. The models will be evaluated against expert estimates derived from blind-sweep data. This challenge represents a first step into FGR detection in low-resource settings. Its main aim is to accurately estimate AC from blind sweep data acquired by novice operators. These estimates could eventually be used to detect FGR, though FGR detection is beyond the scope of the challenge itself. Our end goal is to create effective AI applications for ultrasound imaging that will help improve the care provided to pregnant women and neonates in these regions.

Publication

The results of the ACOUSLIC-AI challenge will be published in a journal article. Teams whose algorithms rank in the top three of the Final Leaderboard (Final Test Phase) will be invited to co-author the challenge paper, with a maximum of three members per team.

Contact information

Please feel free to post any questions about the challenge in our discussion forum. This platform is available for you to interact with fellow participants and to ask questions to the organizers.

Partners and organizers

This challenge is organized by Radboud University Medical Center and sponsored by Delft Imaging Systems.

References

1.            Gardosi J, Chang A, Kalyan B, Sahota D, Symonds EM. Customised antenatal growth charts. The Lancet. 1992 Feb 1;339(8788):283-7.

2.            Bernstein IM, Horbar JD, Badger GJ, Ohlsson A, Golan A. Morbidity and mortality among very-low-birth-weight neonates with intrauterine growth restriction. Am J Obstet Gynecol. 2000 Jan 1;182(1):198-206.

3.            Unterscheider J, O'Donoghue K, Daly S, Geary MP, Kennelly MM, McAuliffe FM, et al. Fetal growth restriction and the risk of perinatal mortality-case studies from the multicentre PORTO study. BMC Pregnancy Childbirth. 2014 Feb 11;14(1):63.

4.            Lawn JE, Blencowe H, Pattinson R, Cousens S, Kumar R, Ibiebele I, et al. Stillbirths: Where? When? Why? How to make the data count? The Lancet. 2011 Apr 23;377(9775):1448-63.

5.            Lawn JE, Ohuma EO, Bradley E, Idueta LS, Hazel E, Okwaraji YB, et al. Small babies, big risks: global estimates of prevalence and mortality for vulnerable newborns to accelerate change and improve counting. The Lancet. 2023 May 20;401(10389):1707-19.

6.            Society for Maternal-Fetal Medicine (SMFM). Electronic address: pubs@smfm.org, Martins JG, Biggio JR, Abuhamad A. Society for Maternal-Fetal Medicine Consult Series 52: Diagnosis and management of fetal growth restriction: (Replaces Clinical Guideline Number 3, April 2012). Am J Obstet Gynecol. 2020 Oct;223(4):B2-17.

7.            van Scheltinga JAT, Scherjon SA, van Dillen J. Nederlandse Vereniging voor Obstetrie en Gynaecologie (NVOG). 2017 [cited 2024 Jan 17]. NVOG-Richtlijn Foetale Groeirestrictie (FGR). Available from: https://www.nvog.nl/wp-content/uploads/2017/12/Foetate-groeirestricie-FGR-15-09-2017.pdf

8.            Gordijn SJ, Beune IM, Thilaganathan B, Papageorghiou A, Baschat AA, Baker PN, et al. Consensus definition of fetal growth restriction: a Delphi procedure. Ultrasound Obstet Gynecol. 2016;48(3):333-9.

9.            Lees CC, Stampalija T, Baschat AA, Da Silva Costa F, Ferrazzi E, Figueras F, et al. ISUOG Practice Guidelines: diagnosis and management of small-for-gestational-age fetus and fetal growth restriction. Ultrasound Obstet Gynecol. 2020 Aug;56(2):298-312.

10.         Abuhamad A, Zhao Y, Abuhamad S, Sinkovskaya E, Rao R, Kanaan C, et al. Standardized Six-Step Approach to the Performance of the Focused Basic Obstetric Ultrasound Examination. Am J Perinatol. 2016 Mar;02(1):90-8.

11.         DeStigter KK, Morey GE, Garra BS, Rielly MR, Anderson ME, Kawooya MG, et al. Low-Cost Teleradiology for Rural Ultrasound. In: 2011 IEEE Global Humanitarian Technology Conference. 2011. p. 290-5.

12.         Self A, Chen Q, Desiraju BK, Dhariwal S, Gleed A, Mishra D, et al. Developing Clinical Artificial Intelligence for Obstetric Ultrasound to Improve Access in Underserved Regions: Protocol for a Computer-Assisted Low-Cost Point-of-Care UltraSound (CALOPUS) Study. JMIR Res Protoc. 2022 Sep;11(9):e37374.

13.         van den Heuvel T, Petros H, Santini S, de Korte C, van Ginneken B. AUTOMATED FETAL HEAD DETECTION AND CIRCUMFERENCE ESTIMATION FROM FREE-HAND ULTRASOUND SWEEPS USING DEEP LEARNING IN RESOURCE- LIMITED COUNTRIES. ULTRASOUND Med Biol. 2019 Mar;45(3):773-85.

14.         Arroyo J, Marini TJ, Saavedra AC, Toscano M, Baran TM, Drennan K, et al. No sonographer, no radiologist: New system for automatic prenatal detection of fetal biometry, fetal presentation, and placental location. PLOS ONE. 2022 Feb 9;17(2):e0262107.

15.         Gomes RG, Vwalika B, Lee C, Willis A, Sieniek M, Price JT, et al. A mobile-optimized artificial intelligence system for gestational age and fetal malpresentation assessment. Commun Med. 2022 Oct 11;2(1):1-9.

16.         Lee C, Willis A, Chen C, Sieniek M, Watters A, Stetson B, et al. Development of a Machine Learning Model for Sonographic Assessment of Gestational Age. JAMA Netw OPEN. 2023 Jan 4;6(1):e2248685.

17.         Pokaprakarn T, Prieto JC, Price JT, Kasaro MP, Sindano N, Shah HR, et al. AI Estimation of Gestational Age from Blind Ultrasound Sweeps in Low-Resource Settings. NEJM Evid. 2022 Apr 26;1(5):EVIDoa2100058.

18.         Gleed AD, Chen Q, Jackman J, Mishra D, Chandramohan V, Self A, et al. Automatic Image Guidance for Assessment of Placenta Location in Ultrasound Video Sweeps. Ultrasound Med Biol. 2023 Jan 1;49(1):106-21.

19.         Gleed AD, Mishra D, Chandramohan V, Fu Z, Self A, Bhatnagar S, et al. Towards Multi-Sweep Ultrasound Video Understanding: Application in Detection of Breech Position Using Statistical Priors. In: 2023 IEEE 20th International Symposium on Biomedical Imaging (ISBI) [Internet]. Cartagena, Colombia: IEEE; 2023 [cited 2023 Dec 29]. p. 1-5. Available from: https://ieeexplore.ieee.org/document/10230662/

20.         Maraci MA, Bridge CP, Napolitano R, Papageorghiou A, Noble JA. A framework for analysis of linear ultrasound videos to detect fetal presentation and heartbeat. Med Image Anal. 2017 Apr;37:22-36.

21.         Schilpzand M, Neff C, van Dillen J, van Ginneken B, Heskes T, de Korte C, et al. AUTOMATIC PLACENTA LOCALIZATION FROM ULTRASOUND IMAGING IN A RESOURCE-LIMITED SETTING USING A PREDEFINED ULTRASOUND ACQUISITION PROTOCOL AND DEEP LEARNING. ULTRASOUND Med Biol. 2022 Apr;48(4):663-74.

22.         Self A, Chen Q, Noble J a., Papageorghiou A t. OC10.03: Computer-assisted low-cost point of care ultrasound: an intelligent image analysis algorithm for diagnosis of malpresentation. Ultrasound Obstet Gynecol. 2020;56(S1):28-28.

23.         Plotka S, Klasa A, Lisowska A, Seliga-Siwecka J, Lipa M, Trzcinski T, et al. Deep learning fetal ultrasound video model match human observers in biometric measurements. Phys Med Biol. 2022 Feb 21;67(4).