More than 800 women globally die each day of preventable pregnancy- and childbirth-related complications, and 20 times this number have peripartum injuries, infections, and disabilities.1,2 Improving access to obstetric ultrasonography is one proposed means to reduce maternal morbidity and mortality by revealing high-risk conditions, and the World Health Organization recommends that obstetric ultrasonography be incorporated into all prenatal care.3 Limited obstetric ultrasonography allows for the detection of fetal cardiac activity, number of gestations, amniotic fluid volume, fetal presentation, and placental position with respect to the internal cervical os, as well as estimation of fetal size.4,5 Detection of pregnancy complications with ultrasonography can allow modification of the care plan, including referral to higher-resource centers; however, this valuable diagnostic imaging is often unavailable in rural areas and resource-poor countries.6–11

Challenges to widespread deployment of ultrasonography, particularly to global rural settings, include lack of trained ultrasonographers, lack of equipment, and lack of infrastructure.7,10,12 To eliminate these barriers, one proposed solution is volume sweep imaging with a portable, low-cost ultrasound system combined with telemedicine for remote interpretation by an offsite trained specialist or artificial intelligence. Volume sweep imaging uses blind sweeps of the ultrasound probe guided by external anatomic landmarks to direct transducer movements and does not require the ultrasound operator to have any knowledge of internal anatomy or pathology. It can therefore be performed by operators without prior medical knowledge who undergo a brief training program, with competency expected within hours (as opposed to months or years for traditional training). Volume sweep imaging has been shown to adequately demonstrate obstetric, lung, thyroid, breast, and right upper quadrant pathologies with excellent agreement with formal diagnostic ultrasonography and potential for artificial intelligence for automated diagnosis.13–25

This type of protocolized blind sweep obstetric ultrasound examination has been tested previously in individuals with normal, uncomplicated pregnancies.4,13,14,21,23 Here, we build on these studies by using this technology to identify pregnancies with abnormalities. We hypothesized that the blinded ultrasound sweeps using a portable ultrasound probe and performed by individuals without prior formal ultrasound training will allow high diagnostic accuracy relative to a standard diagnostic ultrasonogram for the detection of complications of pregnancy such as multiple gestations, placenta previa, abnormal amniotic fluid volume, and noncephalic fetal presentation.

METHODS

This is a single-center, prospective cohort study of people with second- and third-trimester pregnancies (gestational ages greater than 14 0/7 weeks) recruited from a single university tertiary care institution in New York. The setting was a regional referral center for high-risk pregnancy conditions and was selected because of an enriched population available for recruitment. Adult patients were eligible if undergoing diagnostic obstetric ultrasonography (the clinical reference standard) ordered and scheduled by their primary obstetric care practitioner at one of three university-affiliated clinics or in the inpatient obstetric hospital unit. Investigators screened patients for eligibility from the daily clinical schedule on the basis of the stated indication for ultrasound examination. Patients were eligible for inclusion if the clinical reference standard ultrasonogram confirmed that at least one of four prespecified pregnancy complications was present (study group) or absent (control group). Exclusion criteria included fetal congenital anomaly(ies) or fetal death, English not the primary language, positive for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection and in quarantine, medical emergency identified on reference standard ultrasonogram, or unable to be approached because of ultrasound unit clinical volumes or patient time constraints. Eligible participants were enrolled sequentially in a convenience sample on a volunteer basis. Enrollment continued until the intended sample size was reached for each of the four target conditions and for the control group and occurred from October 2020 to January 2022. Participants who underwent multiple diagnostic ultrasonograms during the same pregnancy during this time period were eligible to be included multiple times in data collection. Demographic characteristics of included patients were abstracted from the medical record.

This study was approved by the University of Rochester Medical Center IRB, and informed consent was obtained from all participants before enrollment. The STARD (Standards for Reporting Diagnostic Accuracy) guidelines were used in manuscript preparation.26

All participants underwent obstetric volume sweep imaging ultrasound examination with a handheld Butterfly iQ ultrasound system as the index test. Volume sweep imaging allows the acquisition of ultrasound imaging of a target area based on a series of probe sweeps and arcs with external anatomic landmarks used as reference points.27 The obstetric volume sweep imaging protocol index test used in this study involves sequential performance of eight standardized blind sweeps of the ultrasound probe over the maternal abdomen guided solely by external body landmarks and is intended to record short cine clips of the gravid uterus and fetus (Fig. 1). Details of this protocol have been previously published.13,14 All imaging is performed transabdominally, and an examination can be completed in less than 10 minutes. The intended use of the obstetric volume sweep imaging protocol with the Butterfly iQ ultrasound system is for an individual without formal ultrasound training to record standard imaging sweeps of the gravid uterus, which may then be transmitted remotely over low internet bandwidths (roughly equivalent to dial-up internet) with reasonable transmission times for asynchronous remote interpretation through a previously described telediagnostic system.14

Fig. 1.:

Illustration of the obstetrics volume sweep imaging (VSI) protocol. The Obstetrics protocol involves eight sweeps each beginning and ending with an arc (fan) of the probe that has not been illustrated for simplicity. Arcing of the probe allows maximal visualization. Steps 1–3 involve transverse sweeps of the probe from the pelvis to the upper abdomen. Steps 4 and 5 are sagittal sweeps at the base of the pelvis. Steps 6–8 are sagittal sweeps on the upper gravid abdomen. Reprinted from Toscano M, Marini TJ, Drennan K, Baran TM, Kan J, Garra B, et al. Testing telediagnostic obstetric ultrasound in Peru: a new horizon in expanding access to prenatal ultrasound. BMC Pregnancy Childbirth 2021;21:328. doi: 10.1186/s12884-021-03720-w. This article was published under a CC BY license, which allows copying and redistributing the material in any medium or format and permits remixing, transforming, and building on the material for any purpose, even commercially.

Prespecified pregnancy complications included abnormal placentation, multiple gestations, abnormal amniotic fluid volume, or noncephalic presentation. A positive index test consisted of the presence of one or more of these conditions, and a negative index test consisted of the absence of all of these conditions. Abnormal placentation was defined as complete placenta previa (placenta covering internal cervical os) or low-lying placenta (placental margin within 2 cm of the internal cervical os). Multiple gestation was defined as any pregnancy with more than one viable fetus. Abnormal amniotic fluid volume was defined as either oligohydramnios (amniotic fluid index 5 cm or less or single deepest vertical pocket of fluid less than 2 cm) or polyhydramnios (amniotic fluid index 24 cm or greater or deepest vertical pocket 8 cm or greater).5,28Noncephalic fetal presentation was defined as any fetal presentation other than vertex and was assigned only in pregnancies at 28 weeks of gestation or later.

In this study, because of the coronavirus disease 2019 (COVID-19) pandemic and logistical access, ultrasound operators who performed the volume sweep imaging scans were medical students without prior formal ultrasound training. Medical students were chosen as surrogates for the intended training program participant, a community member or health worker in a low-resource, global, or rural health setting without prior training, because of availability and the potential for rigorous assessment of the protocol before implementation in the field. Previous studies have shown that rural global health workers without a medical background are able to learn similar volume sweep imaging scanning protocols in brief time periods without difficulty.15,29

Medical student ultrasonographers included three fourth-year students, two third-year students, and one first-year student. Students were surveyed regarding prior ultrasonography experience. About half reported prior exposure to observing an ultrasound scan (of any organ system) in a clinical setting, and the remaining reported personally performing point-of-care ultrasound scanning (of any organ system) in fewer than 10 patients in their lifetime. Students were instructed not to look at ultrasound images on the screen as they performed the examination or target any internal anatomic structures but instead focus on moving the ultrasound probe across the maternal abdomen according to the specific external anatomic landmarks of the protocol.

Ultrasound operators received training on the volume sweep imaging blind sweep protocol as described previously.30 The training curriculum was delivered to each person individually as a one-time, stand-alone session performed over 3 hours and included a didactic portion (about 60 minutes) using a visual aid (Fig. 1) and a video aid (https://cdn-links.lww.com/permalink/ruq/a/ruq_00_00_2022_09_15_toscano_usq-d-22-00088_sdc2.mov) and supervised, instructional clinical scanning portion (about 120 minutes) using volunteer pregnant people. In this study, no ongoing quality assessments or training refreshers were performed after trainees completed the training program, but all didactic materials were easily accessible to them for individual review at any time.

,,]}

All ultrasound scans were performed with the Butterfly iQ ultrasound probe with an Apple iPad Air with the Butterfly iQ application downloaded to the device. Scans were performed with the single, multifrequency 1- to 9-mHz probe on obstetric preset using B-mode scanning only.16 Operators were explicitly instructed not to adjust any of the preset settings. Sample sweeps using the ultrasound system are shown in Video 1.

The reference standard in this study was a diagnostic obstetric transabdominal ultrasound examination performed by registered diagnostic medical ultrasonographers with accreditation in obstetric ultrasonography from American Institute of Ultrasound in Medicine. Reference ultrasonograms were performed in accordance with the American Institute of Ultrasound in Medicine Practice Parameter for the Performance of Standard Diagnostic Obstetric Ultrasound Examinations with a Voluson E10 machine.5 This reference standard was chosen because it is the current standard of care for establishing the presence or absence of the pregnancy complications selected for this study. All volume sweep imaging blind sweep ultrasound examinations were performed immediately after the reference standard diagnostic obstetric ultrasound examination.

All volume sweep imaging studies were interpreted by five maternal–fetal medicine physicians with specialized training in interpretation of obstetric ultrasonograms (range 1–14 years of experience) who were blinded to both images and results of the diagnostic ultrasonogram reference standard. After completion of all data collection, studies were randomized before interpretation with a random number generator to ensure blinding and to decrease any possible bias arising from sequential enrollment.

Volume sweep imaging index tests were evaluated for the ability to detect a live fetus (according to cardiac activity), fetal number, fetal presentation, placental location, and amniotic fluid volume. Fetal biometry was measured as an exploratory outcome. Image quality was assigned to each complete study with the use of a three-point Likert scale (1=excellent, 2=acceptable, 3=poor). “Excellent” was designated when there was sufficient image contrast resolution, image depth, and gain to define fetal structures, fluid, and placenta from one another. “Acceptable” was designated when there was some noise, decreased contrast resolution, decreased penetration, or other aspect of the image that degraded image quality and made interpretation less optimal but still diagnostic. “Poor” was designated when there was significant noise, decreased contrast resolution, decreased penetration, or other aspect of the image that made image interpretation challenging or impossible. Examination quality was also assigned to each scan with a three-point Likert scale (1=diagnostic, 2=limited, 3=nondiagnostic). When all five key categorical variables of a limited obstetric ultrasound scan (fetal heart motion, fetal presentation, placental location relative to internal cervical os, fetal number, and amniotic fluid volume) could be seen on the scan, it was designated as “diagnostic.” When some but not all of these five key categorical variables were seen on the scan, it was designated as “limited.” When none of the categorical variables could be seen, it was designated as “nondiagnostic.”

All reference standard diagnostic ultrasound examinations were interpreted by separate board-certified maternal–fetal medicine physicians who were blinded to the images and results from the volume sweep imaging study. Reference standard diagnostic ultrasonography reports written by these individuals were abstracted for pregnancy complications and fetal biometry measurements. No missing data or indeterminate results were present among reference standard ultrasound examinations.

The primary objective of this study was to examine the sensitivity, specificity, positive predictive value, and negative predictive value of volume sweep imaging ultrasound examination with the Butterfly iQ ultrasound system relative to the reference standard for identifying prespecified pregnancy complications. Sensitivity for multiple concurrent pathologies was also investigated as a subanalysis.

Secondary objectives included assessment of overall agreement and interreader agreement of volume sweep imaging relative to the reference standard for pregnancy complications. Prespecified subanalyses were planned for participants in different body mass index (BMI, calculated as weight in kilograms divided by height in meters squared) categories (lower than 30, lower than 35, and lower than 40) and for maternal–fetal medicine physician experience level (less than 5 years and more than 5 years). Additional secondary outcomes included comparison of volume sweep imaging fetal biometry to reference standard fetal biometry measurements (also subanalyzed by BMI lower than 30, lower than 35, and lower than 40), as well as determination of overall image quality and examination quality. Estimated fetal weight was calculated from Shepard and Hadlock equations, and estimated gestational age was calculated with the Hadlock equation.31–36

A priori sample size calculations were performed by defining an anticipated proportion of pregnancy complications and a one-sided lower-limit 95% CI of this proportion for each prespecified complication. The goal of the study was to determine detection proportions of specific pregnancy complications compared with a reference standard rather than explicitly demonstrating noninferiority to this reference standard. Therefore, sample sizes were calculated from a desired CI width rather than using typical methods for noninferiority. For a desired detection proportion of 90% and a lower-limit CI of 80%, a target sample size of 50 for each group was needed. The Clopper–Pearson method was used to calculate CIs. Sample sizes were calculated with PASS 11.0.2.

Reported results are summarized as proportions with 95% CI for categorical variables and mean±SD for continuous variables. Agreement between volume sweep imaging and diagnostic ultrasound reference test was determined with the Cohen κ for categorical variables and the intraclass correlation coefficient and Bland–Altman bias for continuous biometry values. Intraclass correlation was calculated with a two-way random model for absolute agreement. Bland–Altman bias was calculated as index test minus reference test. In both cases, reported P values are for a one-sample t test comparing with a theoretical value of 0. Interreader agreement was calculated with the Fleiss multirater κ. All statistical analyses were performed with MATLAB R2019b.

RESULTS

Trainees performed volume sweep imaging ultrasound examinations on 168 unique pregnant people (248 fetuses) who underwent 194 separate ultrasound examinations with eight sweeps per examination for a total of 1,552 blinded sweep cine clips (Fig. 2). There were no adverse events. Included participants were on average 31.7 (SD 5.42) years old, were parous, and had obesity (mean BMI 33.3±7.95). The majority (56.7%) underwent ultrasound examination in the third trimester. Three quarters of included participants (74.7%) had a medical condition complicating their pregnancy, and one quarter (25.3%) had a pregnancy-related complication. There were 49 ultrasonograms with normal results (control group) and 145 ultrasonograms with abnormal results with pregnancy complications, including multiple gestations (26.8%, n=52), placenta previa (15.4%, n=30), polyhydramnios (10.8%, n=21), oligohydramnios (7.7%, n=15), and noncephalic presentation (41.7%, n=81) (Table 1). Image quality was rated as excellent or acceptable in 88.2% of scans. Only 7 scans (3.6%) were rated as nondiagnostic (Table 2).

Fig. 2.:

Flow diagram of participant selection. Index test is the blind sweep by an individual without training in volume sweep imaging. Reference standard is a standard-of-care ultrasonogram performed by an ultrasonographer in the clinical setting. Target condition is a prespecified group of pregnancy complications. COVID-19, coronavirus disease 2019.

Table 1.:

Characteristics of Included Participants

Table 2.:

Image and Examination Quality of Blind Ultrasound Sweeps

Overall, the volume sweep imaging index test demonstrated a sensitivity of 91.7% (95% CI 87.2–96.2%) for detection of one of the prespecified pregnancy complications. Complications with the highest detection rates were multiple gestations (sensitivity 100%, specificity 95.1%) and noncephalic presentation (sensitivity 91.8%, specificity 83.9%). Detection of low-lying placenta or placenta previa or abnormal fluid volume was lower (sensitivity 75.8% and 56.9%, respectively), but the test had a high negative predictive value for these conditions (96.1% and 89.5%, respectively) (Table 3).

Table 3.:

Detection Rates of Pregnancy Complications Identified by Blind Ultrasound Sweeps

High mean agreement between interpretation of obstetric volume sweep imaging index test by five maternal–fetal medicine subspecialists compared with formal diagnostic ultrasound reference test was demonstrated (Table 4). Interreader agreement was substantial for noncephalic fetal presentation; moderate for an overall abnormal examination results, multiple gestations, and abnormal amniotic fluid volume; and fair for low placental implantation (Table 5).

Table 4.:

Agreement Between Blind Ultrasound Volume Sweep Imaging and Reference Diagnostic Ultrasonogram

Table 5.:

Interreader Agreement*

In a planned subanalysis of obstetric volume sweep imaging interpretation performed by more experienced (more than 5 years of experience) compared with less experienced (less than 5 years of experience) maternal–fetal medicine subspecialists, agreement with the diagnostic ultrasound reference test and interreader agreement was not significantly different according to experience level (Appendix 1, available online at https://links.lww.com/AOG/D84). Similarly, in a planned subanalysis comparing obstetric volume sweep imaging interpretation in participants with varying levels of obesity (BMI lower than 30, lower than 35, or lower than 40), agreement with the diagnostic ultrasound reference test was not significantly different according to BMI (Appendix 2, available online at https://links.lww.com/AOG/D84).

The intraclass correlation between volume sweep imaging and diagnostic ultrasound measurements of fetal size was found to be excellent. Estimated fetal weight was within a mean error of 116±62 g (intraclass correlation 0.93, 95% CI 0.892–0.953, P<.0001), and estimated gestational age was within a mean error of 11±11 days (intraclass correlation 0.93, 95% CI 0.895–0.947, P<.0001). Bland–Altman analysis demonstrated similar results, with 95% limits of agreement between measurements identified as −5.03 to 3.48 weeks (P<.001) for estimated gestational age. A clinically irrelevant but statistically significant bias of −11.1 g was identified in Bland–Altman analysis of estimated fetal weight (P=.61) with an agreement range of −401 to 379 g. There was also excellent correlation between each individual biometric measurement (femur length, biparietal diameter, head circumference, and abdominal circumference), with the intraclass correlation ranging from 0.9 to 0.95. The mean difference in measurements was lowest for biparietal diameter (2.89±3.91 mm), followed by femur length (4.74±3.95 mm), head circumference (12.2±14.2 mm), and abdominal circumference (17.1±14.5 mm) (Table 6). Fetal measurement correlation was stronger for examinations performed in the second trimester than for those performed in the third trimester (Appendix 3, available online at https://links.lww.com/AOG/D84) but was not significantly affected by the inclusion of participants of increasing classes of obesity (Appendix 4, available online at https://links.lww.com/AOG/D84).

Table 6.:

Fetal Measurements Obtained by Blind Ultrasound Sweeps Compared With Reference Diagnostic Ultrasonogram

DISCUSSION

Blinded ultrasound sweeps performed by operators without prior formal ultrasound training after a brief training program demonstrated substantial diagnostic accuracy for the identification of pregnancy complications. Furthermore, these examinations were performed with an inexpensive handheld, battery-powered device. This technology has the potential to increase the availability of ultrasonography in resource-limited settings. Although this approach previously has been demonstrated to be feasible in people with normal, low-risk pregnancies, the contribution of this study is the demonstration of effective performance in abnormal pregnancy states. Potentially serious causes of maternal and fetal morbidity and mortality, such as placenta previa, noncephalic fetal position, abnormal amniotic fluid volume, and multiple gestations, were detected by the blinded ultrasound sweeps in this study. Furthermore, demonstration of accuracy of fetal size measurements suggests that this protocol could be potentially useful in establishing gestational age or identifying fetal growth disorders.

The primary intended use of this approach would be to screen for pregnancy complications to identify patients who would benefit from being directed to appropriate care settings for further assessment and delivery. Diagnostic ultrasound examinations performed by an expert ultrasonographer would still be required to follow-up cases of positive screening tests. Telemedicine systems incorporating volume sweep imaging or blinded ultrasound sweeps have already demonstrated efficacy in normal pregnancies.13–23 The advantage of these systems is that they are asynchronous, requiring neither high-speed internet nor the presence of an onsite specialist. Blind ultrasound sweeps can be performed offline, and the acquisitions can be stored locally until internet becomes available. In contrast, synchronous teleultrasound systems require high internet speeds and the concomitant presence of the specialist for direct video chat and direction of the ultrasound examination.12

Obtaining the enriched cohort of pregnancies with abnormalities to answer the primary question of this study necessitated its performance at a high-resource hospital in the United States to overcome logistical constraints related to the COVID-19 pandemic. This setting allowed rigorous testing of the technology while overcoming barriers to study such as adequate reference standards, low volumes of patients with pregnancies with abnormalities presenting to rural health centers, and the substantial expenses required to conduct high-quality studies in a global setting. Replicating the results of the current study in a larger trial conducted in a rural or low-resource setting using local community workers as ultrasound operators is a crucial next step.

Similarly, ultrasound operators in the current study were medical students (without prior formal ultrasound training) as opposed to members of rural communities who would perform this approach if put in practice. The use of medical students was necessary for logistical reasons and for regulatory purposes with the IRB. Students were blinded to and explicitly instructed not to look at the ultrasound screen during scanning. Although it is a theoretical possibility that their medical knowledge could have produced bias leading to improved examination quality, those precautions and the standardized nature of the examination make it less likely their medical experience significantly affected the quality of the examinations. Furthermore, we have previously trained ultrasound operators in rural Peru who have decreased familiarity with clinical medicine and ultrasound technology without issues with diagnostic examination quality or accuracy.15,29

Results demonstrate that this current version of the scanning protocol has diagnostic accuracy comparable with the current standard of care for detection of multiple gestation and fetal cardiac activity. The sensitivity for noncephalic presentation (in general) was somewhat decreased overall when a transverse- or breech-presenting second twin was included, but among third-trimester singletons alone (as would be most clinically useful), the sensitivity was also equivalent to that of standard of care. Transvaginal imaging is the optimal approach to visualize the internal cervical os to detect low placental implantation. However, this study used a pragmatic and real-world approach to screening for this with transabdominal imaging. The sensitivity for low-lying placenta by transabdominal imaging in our study was higher than previously reported with the use of diagnostic ultrasonography and a trained ultrasonographer (reported sensitivity 20%), and the sensitivity for placenta previa is comparable (reported sensitivity 81–94%).37–40 Despite this, the transabdominal approach to identifying low placental implantation used in this study would have missed eight cases. Future iterations will modify training and the blind sweep protocol to more effectively visualize the lower uterine segment and cervical os to improve sensitivity. Similarly, diagnostic accuracy for the identification of abnormal amniotic fluid volume was limited with this protocol. This is likely attributable to composite grouping of participants with oligohydramnios and polyhydramnios. There was an over-representation of patients with mild polyhydramnios in our study cohort, which is difficult to discern from normal fluid volume without formal caliper measurement of amniotic fluid pockets. The sensitivity for oligohydramnios and anhydramnios was much higher than that for polyhydramnios in the current cohort.

These findings need to be replicated in global low-resource and rural areas with trainees without a clinical background to determine the preferred implementation strategies to enhance sustainability. In addition, larger-scale clinical trials to investigate the diagnostic accuracy for these abnormal findings in an unselected population are necessary. The public health and health services components that will be needed to effectively disseminate and implement this technology on a larger scale are also a major consideration. Expanding to image interpretation by general obstetrics and gynecology physicians or potentially advanced practice health care professionals, as opposed to maternal–fetal medicine subspecialists alone, could increase the pool of available interpreters and the success of such a program.

The majority of people in the world lack access to obstetric ultrasonography. The ability to scale up and deploy this technology could offer an effective means to screen for pregnancy complications and to decrease associated maternal and neonatal morbidity and mortality. Integration of artificial intelligence with this approach could potentially allow rapid and automatic diagnosis of pregnancy complications without a radiologist or an ultrasonographer. Closing the gap of the significant health care disparities in obstetrics will require new and innovative strategies. Obstetric volume sweep imaging shows significant promise as a tool to diagnose pregnancy complications and merits further investigation and development.

REFERENCES 1. World Health Organization. Trends in maternal mortality 2000 to 2017: estimates by WHO, UNICEF, UNFPA, World Bank Group and the United Nations Population Division. Accessed October 31, 2022. https://apps.who.int/iris/bitstream/handle/10665/327596/WHO-RHR-19.23-eng.pdf?sequence=13&isAllowed=y 2. Geller SE, Koch AR, Garland CE, MacDonald EJ, Storey F, Lawton B. A global view of severe maternal morbidity: moving beyond maternal mortality. Reprod Health 2018;15:98. doi: 10.1186/s12978-018-0527-2 3. World Health Organization. WHO antenatal care recommendations for a positive pregnancy experience: maternal and fetal assessment update: imaging ultrasound before 24 weeks of pregnancy. 2022. Accessed October 31, 2022. https://apps.who.int/iris/handle/10665/352620 4. Sinkovskaya E, Zhao Y, Abuhamad S, Rao R, Kanaan C, Platt L. Standardized six-step approach to the performance of the focused basic obstetric ultrasound examination. Am J Perinatol 2015;33:90–8. doi: 10.1055/s-0035-1558828 5. AIUM–ACR–ACOG–SMFM–SRU practice parameter for the performance of standard diagnostic obstetric ultrasound examinations. J Ultrasound Med 2018;37:E13–24. doi: 10.1002/jum.14831 6. WHO recommendations on antenatal care for a positive pregnancy experience. World Health Organization; 2016. 7. Harris RD, Marks WM. Compact ultrasound for improving maternal and perinatal care in low-resource settings: review of the potential benefits, implementation challenges, and public health issues. J Ultrasound Med 2009;28:1067–76. doi. 10.7863/jum.2009.28.8.1067 8. Bauserman M, Nathan R, Lokangaka A, McClure EM, Moore J, Ishoso D, Polyhydramnios among women in a cluster-randomized trial of ultrasound during prenatal care within five low and low-middle income countries: a secondary analysis of the first look study. BMC Pregnancy Childbirth 2019;19:258. doi: 10.1186/s12884-019-2412-6 9. Sippel S, Muruganandan K, Levine A, Shah S. Review article: use of ultrasound in the developing world. Int J Emerg Med 2011;4:72. doi: 10.1186/1865-1380-4-72 10. Maru DSR, Schwarz R, Andrews J, Basu S, Sharma A, Moore C. Turning a blind eye: the mobilization of radiology services in resource-poor regions. Global Health 2010;6:18. doi: 10.1186/1744-8603-6-18 11. Yadav H, Shah D, Sayed S, Horton S, Schroeder LF. Availability of essential diagnostics in ten low-income and middle-income countries: results from national health facility surveys. Lancet Glob Health 2021;9:e1553–60. doi: 10.1016/S2214-109X(21)00442-3 12. Britton N, Miller MA, Safadi S, Siegel A, Levine AR, McCurdy MT. Tele-ultrasound in resource-limited settings: a systematic review. Front Public Health 2019;7:244. doi: 10.3389/fpubh.2019.00244 13. Marini TJ, Oppenheimer DC, Baran TM, Rubens DJ, Toscano M, Drennan K. New ultrasound telediagnostic system for low-resource areas: pilot results from Peru. J Ultrasound Med 2021;40:583–95. doi: 10.1002/jum.15420 14. Toscano M, Marini TJ, Drennan K, Baran TM, Kan J, Garra B. Testing telediagnostic obstetric ultrasound in Peru: a new horizon in expanding access to prenatal ultrasound. BMC Pregnancy Childbirth 2021;21:328. doi: 10.1186/s12884-021-03720-w 15. Marini TJ, Castaneda B, Baran T, O’Connor TP, Garra B, Tamayo L, et al. Lung ultrasound volume sweep imaging for pneumonia detection in rural areas: piloting training in rural Peru. J Clin Imaging Sci 2019;9:35. doi: 10.25259/jcis_29_2019 16. Åhman A, Edvardsson K, Lesio Kidanto H, Ngarina M, Small R, Mogren I. “Without ultrasound you can’t reach the best decision”: midwives’ experiences and views of the role of ultrasound in maternity care in Dar Es Salaam, Tanzania. Sex Reprod Healthc 2018;15:28–34. doi: 10.1016/j.srhc.2017.11.007 17. DeStigter KK, Morey GE, Garra BS, Rielly MR, Anderson ME, Kawooya MG. Low-cost teleradiology for rural ultrasound. IEEE Global Humanitarian Technology Conference 2011:290–5. doi: 10.1109/GHTC.2011.39 18. Arroyo J, Marini TJ, Saavedra AC, Toscano M, Baran TM, Drennan K. No sonographer, no radiologist: new system for automatic prenatal detection of fetal biometry, fetal presentation, and placental location. PLoS One 2022;17:e0262107. doi: 10.1371/journal.pone.0262107 19. Marini TJ, Weiss SL, Gupta A, Zhao YT, Baran TM, Garra B. Testing telediagnostic thyroid ultrasound in Peru: a new horizon in expanding access to imaging in rural and underserved areas. J Endocrinol Invest 2021;44:2699–708. doi: 10.1007/s40618-021-01584-7 20. Marini TJ, Weis JM, Baran TM, Kan J, Meng S, Yeo A, et al. Lung ultrasound volume sweep imaging for respiratory illness: a new horizon in expanding imaging access. BMJ Open Respir Res 2021;8:e000919. doi: 10.1136/bmjresp-2021-000919 21. Self A, Chen Q, Desiraju BK, Dhariwal S, Gleed AD, 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;11:e37374. doi: 10.2196/37374 22. Marini TJ, Castaneda B, Iyer R, Baran TM, Nemer O, Dozier AM, et al. Breast ultrasound volume sweep imaging: a new horizon in expanding imaging access for breast cancer detection. J Ultrasound Med 2022 Jul 8 [epub ahead of print]. doi: 10.1002/jum.16047 23. 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;1:EVIDoa2100058. doi: 10.1056/EVIDoa2100058 24. Marini TJ, Castaneda B, Parker K, Baran TM, Romero S, Iyer R, et al. No sonographer, no radiologist: assessing accuracy of artificial intelligence on breast ultrasound volume sweep imaging scans. PLoS Digital Health 2022;1:e0000148.doi: 10.1371/journal.pdig.0000148 25. Marini TJ, Oppenheimer DC, Baran TM, Rubens DJ, Dozier A, Garra B, et al. Testing telediagnostic right upper quadrant abdominal ultrasound in Peru: a new horizon in expanding access to imaging in rural and underserved areas. PLoS One 2021;16:e0255919. doi: 10.1371/journal.pone.0255919 26. Bossuyt PM, Reitsma JB, Bruns DE, Gatsonis CA, Glasziou PP, Irwig LM, et al. Towards complete and accurate reporting of studies of diagnostic accuracy: the STARD initiative. BMJ 2003;326:41–4. doi: 10.1136/bmj.326.7379.41 27. Mollura DJ, Azene EM, Starikovsky A, Thelwell A, Iosifescu S, Kimble C, et al. White Paper Report of the RAD-AID Conference on International Radiology for Developing Countries: identifying challenges, opportunities, and strategies for imaging services in the developing world. J Am Coll Radiol 2010;7:495–500. doi: 10.1016/j.jacr.2010.01.018 28. Dashe JS, Pressman EK, Hibbard JU. SMFM consult series #46: evaluation and management of polyhydramnios. Am J Obstet Gynecol 2018;219:B2–8. doi: 10.1016/j.ajog.2018.07.016 29. Marini TJ, Kaproth-Joslin K, Ambrosini R, Baran TM, Dozier AM, Zhao YT, et al. Volume sweep imaging lung teleultrasound for detection of COVID-19 in Peru: a multicentre pilot study. BMJ Open 2022;12:e061332. doi: 10.1136/bmjopen-2022-061332 30. Erlick M, Marini TJ, Drennan K, Castaneda B, Baran T, Toscano M, et al. Assessment of a brief standardized obstetric ultrasound training program for individuals without prior ultrasound experience. Ultrasound Q 2022 Oct 8 [epub ahead of print]. doi: 10.1097/RUQ.0000000000000626 31. Hadlock FP, Shah YP, Kanon DJ, Lindsey JV. Fetal crown-rump length: reevaluation of relation to menstrual age (5-18 weeks) with high-resolution real-time US. Radiology 1992;182:501–5. doi: 10.1148/radiology.182.2.1732970 32. Hadlock FP, Deter RL, Harrist RB, Park SK. Estimating fetal age: computer-assisted analysis of multiple fetal growth parameters. Radiology 1984;152:497–501. doi: 10.1148/radiology.152.2.6739822 33. Skupski DW, Owen J, Kim S, Fuchs KM, Albert PS, Grantz KL. Estimating gestational age from ultrasound fetal biometrics. Obstet Gynecol 2017;130:433–41. doi: 10.1097/aog.0000000000002137 34. Shepard MJ, Richards VA, Berkowitz RL, Warsof SL, Hobbins JC. An evaluation of two equations for predicting fetal weight by ultrasound. Am J Obstet Gynecol 1982;142:47–54. doi: 10.1016/s0002-9378(16)32283-9 35. Hadlock FP. Sonographic estimation of fetal age and weight. Radiol Clin North Am 1990;28:39–50. doi: 10.1016/s0033-8389(22)01218-0 36. Hadlock FP, Harrist RB, Sharman RS, Deter RL, Park SK. Estimation of fetal weight with the use of head, body, and femur measurements: a prospective study. Am J Obstet Gynecol 1985;151:333–7. doi: 10.1016/0002-9378(85)90298-4 37. Westerway SC, Hyett J, Henning Pedersen L. Measuring leading placental edge to internal cervical os: transabdominal versus transvaginal approach. Australas J Ultrasound Med 2017;20:163–7. doi: 10.1002/ajum.12065 38. Sunna E, Ziadeh S. Transvaginal and transabdominal ultrasound for the diagnosis of placenta praevia. J Obstet Gynaecol 1999;19:152–4. doi: 10.1080/01443619965471 39. Lahoria K, Malhotra S, Bagga R. Transabdominal and transvaginal ultrasonography of placenta previa. Int J Gynecol Obstet 2007;98:258–9. doi: 10.1016/j.ijgo.2007.04.024 40. Petpichetchian C, Pranpanus S, Suntharasaj T, Kor-Anantakul O, Hanprasertpong T. Comparison of transabdominal and transvaginal sonography in the diagnosis of placenta previa. J Clin Ultrasound 2018;46:386–90. doi: 10.1002/jcu.22600.