Ultrasound has become an integral part of medical practice, providing timely diagnostic information. Unlike imaging using ionizing radiation such as X-ray or computer tomography (CT) scans, and at a fraction of the cost of Magnetic Resonance Imaging (MRI), ultrasound is used widely due to its cost effectiveness, portability and real-time nature. Many current clinical uses include identifying free fluid in the abdomen or pelvis (a sign of internal haemorrhage) in trauma patients; diagnosing ectopic pregnancies and ovarian cysts in gynaecology; imaging the liver, gallbladder, pancreas and appendix in general surgery; assessing of the adult heart in cardiology; and many others. However, it is in modern obstetric practice that ultrasound is the real cornerstone of care. For this reason, in this desk review we focus on obstetric ultrasound. However, we recognize that the ultrasound devices discussed offer great benefits across the medical field.

The World Health Organization (WHO)’s Recommendations on Antenatal Care for a Positive Pregnancy Experience underscores the need for ultrasound assessment during pregnancy ¹ . One of the most important benefits of ultrasound is the ability to estimate gestational age which is a critical piece of information for maternity care providers ² . The gold standard for dating a pregnancy is to measure the crown-rump length (CRL) between 11–14 weeks. For women presenting later in pregnancy a combination of head circumference (HC) and femur length (FL) measurements can be used ³ . Measurements of the HC, biparietal diameter (BPD), abdominal circumference (AC) and FL also provide useful information on the growth and wellbeing of the fetus. Importantly ultrasound also allows the provider to check for factors that could indicate that the pregnancy is at high risk, and benefit the mother and unborn child from increased monitoring; this includes identifying multiple pregnancies, problems with the placenta (such as placenta praevia), non-cephalic fetal presentation, increased and decreased amniotic fluid levels as well as congenital anomalies ⁴ .

Given that maternal and perinatal mortality remain unacceptably high, particularly in low- and middle-income countries (LMICs), the significance of the WHO guidelines cannot be understated ⁵ . However, in resource-constrained environments, there is a critical lack of ultrasound equipment and skilled operators ⁶ . The cost of devices has decreased in recent years, but without trained health personnel who are able to use the ultrasound device proficiently and interpret findings accurately, quality implementation of antenatal ultrasound services in global settings will continue to be limited ⁷ .

In response to limitations related to user-dependence associated with ultrasound scanning, a research area that has seen considerable growth is artificial intelligence (AI) applied to guidance and image analysis ⁸ . This has particular relevance for point-of-care ultrasound since it has the potential to guide a non-expert user to collect high quality data ⁹ and automatically interpret images to assist with clinical decision making in areas such as obstetrics ¹⁰ . Specifically, object detection capabilities of AI are especially useful in this case, since this allows for tracking objects in real time to allow for automated image or object classification ¹¹ . To this end, automatic measurements of fetal crown-rump length (CRL) ¹² , biparietal diameter (BPD) and head circumference (HC) ¹³ , femur length (FL) ¹⁴ , transcerebellar diameter (TCD) ¹⁵ , abdominal circumference (AC) ¹⁶ , and fetal weight ¹⁷ are all under active investigation. In addition, advances have been made in automated gestational age estimation independent of fetal measurement ¹⁸ , and algorithms to automatically determine fetal heart rate, placental position, fetal presentation, multiple gestation and other crucial obstetric findings are in development ^{19, 20} . Given the rapid development of such AI methods to address challenges of user-dependence in global settings, utilizing ultrasound devices with the capacity to integrate AI algorithms is an important consideration for investment and future deployment.

In recent years, portable ultrasound devices have become increasingly popular in LMICs due to their cost-effectiveness, and usability in resource-constrained environments ²¹ . Many of these new models are hand-held, mobile-based systems that are under $10,000, which is a substantially lower cost than standard full-sized clinical systems ²² . The user-interface of such devices has been simplified and streamlined so that healthcare workers at varying level of experience may be able to use them ²³ . As these devices are becoming more available, funding agencies and local healthcare staff will need to evaluate the strengths and weaknesses to determine which device is optimal for their intended setting.

Application Program Interface (API) and software development kit (SDK) capabilities are another important consideration when it comes to selecting a device. An API is the intermediary software that transfers information between two or more applications to communicate with each other. Modifying an ultrasound system’s API enables the extraction of specific data from that system to a mobile application. Unlike an API, which is a set of programmable libraries that come with the ultrasound system and its corresponding mobile application, an SDK is a set of tools (some of which may include APIs) that allow technical developers to create their own applications. While an API allows users to enhance the built-in application that comes with the ultrasound system, the inclusion of an SDK allows developers to build completely new applications. An SDK often comes with one or more APIs. All in all, SDKs allow for further flexibility and the ability to develop fully functioning applications from scratch using the data provided by the ultrasound system, while APIs have a more limited but still effective capability in customizing an application. Mobile based devices should have an integrable REST (representational state transfer) API capable of accessing an ultrasound system’s SDK.

Ultrasound systems may rely heavily on AI models to track/detect inputs from the system’s sensor. The SDK/API associated with that particular ultrasound system will include options to modify that AI model to, for example, use object detection or perform additional computations to optimize system performance. In this desk review, we present the capabilities and costs of currently available portable ultrasound devices. This is meant to serve as a resource for clinical AI researchers and AI imaging technologists who are weighing the benefits and disadvantages of various ultrasound models for antenatal care in resource-constrained environments.

A list of ideal technical features for portable ultrasound devices, with a focus on antenatal care in an LMICs setting, was developed. The list of features was established with input from subject matter experts (SMEs) including clinicians and academic researchers who actively practice or have research programs in LMICs. All features were defined a priori, and are outlined in Table 1. Each feature is divided into two categories: (1) ‘minimum viable product’ which represent baseline required capabilities, and (2) ‘target state’ which identify ideal attributes. The identified features consist of image acquisition modes (including extra modes for obstetrics), cost, portability, power, compatibility, connectivity, data storage and security, and certification status.

Candidate devices for assessment were identified through internet searches, company websites, discussions with SMEs and industry partners in the field. Only low-cost (less than $10,000) portable, mobile- or tablet-based devices were considered for this desk review. Mid- and full-size ultrasound systems were excluded given cost constraints.

Information regarding the ideal technical features of each of the devices was collected from publicly available information, input from SMEs and/or discussions with sales representatives from the respective companies. The review was conducted and verified in March 2022.

The search identified 14 portable mobile-based ultrasound systems. Ten of these employ convex probes, which are most commonly used for obstetrics due to the width of the field of view allowing improved imaging in later pregnancy; three are linear probes, which may also be useful for non-obstetric applications; while in one case a dual probe (convex and linear) has been engineered in the same device. Ten devices offered color Doppler in addition to B-mode imaging. Calculation of gestational age from fetal biometry and estimation of fetal weight were not available on all devices.

Costs ranged from $700 to $10,000 and three were available on a subscription model. Importantly, from the perspective of building new applications (including AI enabled software), no SDK or API was available for eight of the devices at the time the search was performed; while access to SDK or API was available in limited fashion or by agreement for four. In two devices, this was available by agreement with purchase.

Detailed Information for each feature and for the identified products is summarized in Table 2. The output of the desk review is meant to provide objective information on available ultrasound systems for researchers and clinicians working in obstetric ultrasound in LMICs. The authors’ intention is not to make endorsements for any particular technology, and recommend that stakeholders utilize this information to make the most appropriate decisions for their given context.

This desk review provides a review of the landscape of low-cost portable ultrasound systems for use in obstetrics in LMIC settings. Ideal technical features ( Table 1) were identified a priori to guide the review of each technology. 14 low-cost portable ultrasound systems were identified, and a review of the technical features for each are presented in Table 2. The authors acknowledge that new devices are being commercialized all the time and that any review of rapidly evolving technology is of temporally finite benefit. Nevertheless, we believe that the framework we have developed can be useful for researchers and clinicians in the field to evaluate future options. Additionally, we acknowledge the limitation that information on the devices was based on that available by manufacturers. We considered that field testing was prohibitive given the number of devices available and the aforementioned rapidly developing field, and instead opted to use this review to shortlist devices for future field testing. Finally, we did not evaluate specific imaging characteristics as these can be subjective and difficult to describe in a reproducible manner.

In addition, methods for mapping the landscape, as well as the clearly defined criteria that resulted from SME discussions, may be modified for other clinical applications. For example, a comparable set of ideal technical features could be adapted to create a landscape of portable ultrasound technology for other clinical applications such as cardiac, lung, or musculoskeletal assessment. Depending on the needs of the research community, this review may also serve as a framework for mapping other types of medical technologies for applications to global healthcare.

Numerous areas for future work are highlighted because of this desk review. In particular, the authors suggest convening a focus group of clinicians, academic clinical and AI imaging researchers, and technologists to establish trials for a selection of probes (e.g., multiple probes side-by-side at research sites of varying resources) and provide targeted feedback. As with this study, key domains for assessment must be determined in advance. It will also be essential to test probes with both male and female healthcare workers (as differences in hand sizes contribute to ergonomic performance), and with clinical healthcare workers of diverse skill levels (e.g., doctors, midwives, nurses, sonographers).

In summary, this desk review provides objective information on current available ultrasound systems. It was designed to be a resource for researchers and clinicians working in obstetric ultrasound in LMICs.