CoChat
Sign Up
Cited by 251Artificial Intelligence in Healthcare and EducationAI in cancer detectionMedical Image Segmentation Techniques

Introduction to artificial intelligence in ultrasound imaging in obstetrics and gynecology

Open access
Drukker, L., Noble, J. A., Papageorghiou, A. T.·2020·Ultrasound in Obstetrics & Gynecology
DOI 10.1002/uog.22122PMC PMC7702141
Open this paper in CoChat — free accountOpen PDF

Abstract

Artificial intelligence (AI) uses data and algorithms to aim to draw conclusions that are as good as, or even better than, those drawn by humans. AI is already part of our daily life; it is behind face recognition technology, speech recognition in virtual assistants (such as Amazon Alexa, Apple's Siri, Google Assistant and Microsoft Cortana) and self‐driving cars. AI software has been able to beat world champions in chess, Go and recently even Poker. Relevant to our community, it is a prominent source of innovation in healthcare, already helping to develop new drugs, support clinical decisions and provide quality assurance in radiology. The list of medical image‐analysis AI applications with USA Food and Drug Administration or European Union (soon to fall under European Union Medical Device Regulation) approval is growing rapidly and covers diverse clinical needs, such as detection of arrhythmia using a smartwatch or automatic triage of critical imaging studies to the top of the radiologist's worklist. Deep learning, a leading tool of AI, performs particularly well in image pattern recognition and, therefore, can be of great benefit to doctors who rely heavily on images, such as sonologists, radiographers and pathologists. Although obstetric and gynecological ultrasound are two of the most commonly performed imaging studies, AI has had little impact on this field so far. Nevertheless, there is huge potential for AI to assist in repetitive ultrasound tasks, such as automatically identifying good‐quality acquisitions and providing instant quality assurance. For this potential to thrive, interdisciplinary communication between AI developers and ultrasound professionals is necessary. In this article, we explore the fundamentals of medical imaging AI, from theory to applicability, and introduce some key terms to medical professionals in the field of ultrasound. We believe that wider knowledge of AI will help accelerate its integration into healthcare. © 2020 The Authors. Ultrasound in Obstetrics & Gynecology published by John Wiley & Sons Ltd on behalf of the International Society of Ultrasound in Obstetrics and Gynecology.

AI snapshot

Written from the paper’s full text.

TL;DR

This narrative review introduces artificial intelligence (AI) concepts and terminology to obstetric and gynecologic ultrasound professionals, emphasizing deep learning’s strength in image pattern recognition. It argues that AI has had little impact on obstetric/gynecologic ultrasound so far, but has potential for repetitive tasks, quality assurance, and workflow support if interdisciplinary communication, validation, and ethical/regulatory issues are addressed.

Key findings

  • As of May 2020, more than 50 deep learning-based imaging applications had been approved by the USA FDA or European Union, spanning modalities including ultrasound; the first ultrasound AI application approved by the FDA guides users during adult echocardiography image acquisition.
  • In obstetric and gynecologic ultrasound, AI impact remains limited: at the 29th ISUOG World Congress in 2019 there were 14 abstracts specifically mentioning AI, compared with a total of 13 AI-related abstracts across the preceding six ISUOG World Congresses (2013–2018).
  • Potential Ob/Gyn ultrasound applications listed include probe guidance, fetal biometric plane finding, anomaly scan completeness, anomaly highlighting, ovarian cyst classification according to IOTA criteria, and lung scans for Ob/Gyn in COVID-19 patients.
  • Deep learning performs especially well in image pattern recognition; the ImageNet annual competition reached an error rate below 5% in 2015, described as the average human error rate, and the medical imaging AI market was forecasted to top $2 billion by 2023.

Methods

  • Narrative review/introduction, not a systematic review or primary study; it explains AI-related concepts and terms to clinicians in ultrasound in obstetrics and gynecology.
  • No sample, dataset, or formal quantitative analysis is reported; the paper summarizes selected literature and examples, including a table of reported and expected future AI applications in obstetric and gynecologic ultrasound.
  • Examples are drawn from radiology as the closest sister field and from cited obstetric/gynecologic ultrasound studies, including adnexal masses, endometrial cancer lymph node metastases, pelvic organ function, breast lesions, aneuploidy risk, fetal lung maturity, perinatal outcome, shoulder dystocia, brain damage, gestational age estimation, and fetal brain image classification.

Limitations

  • Stated: a systematic review would be premature at this stage; most published peer-reviewed scientific articles appear in the engineering literature, usually focus on AI methodology, and few studies have assessed clinical applicability.
  • Stated: algorithms and results of approved AI applications are often not published in scientific journals due to commercial sensitivities; most reported AI applications use data from a single site and focus on algorithm performance rather than clinical utility or health economics, and human expert accuracy for the same task is often difficult to determine or unknown.
  • Evident: as a narrative overview, the paper has no systematic search strategy, inclusion criteria, or quantitative synthesis; its examples are illustrative rather than exhaustive, and many implementation examples come from radiology rather than obstetric/gynecologic ultrasound.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1002/uog.22122

Cites

63 of 63 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Claims the paper makes

0 of 26 cited sources checked · 0 supported · 0 flagged

  • Findingreasoning

    Few studies have assessed clinical applicability.

    “In addition, performing a clinical systematic review in this area is challenging because most of the published peer‐reviewed scientific articles appear in the engineering literature which usually focuses on the AI methodology and few studies have assessed clinical applicability.”
    Quote in the paperSubstantial claim
  • Findingnone given

    At the 29th ISUOG World Congress in 2019, there were 14 abstracts specifically mentioning AI, compared with a total of 13 abstracts in the preceding six ISUOG World Congresses (2013-2018).

    “at the 29th World Congress of the International Society of Ultrasound in Obstetrics and Gynecology (ISUOG) in 2019, there were 14 abstracts specifically mentioning AI, in comparison to a total of 13 abstracts in the preceding six ISUOG World Congresses (2013–2018).”
    Quote in the paperNot a substantial claim
  • Interpretationreasoning

    Since human involvement can potentially introduce bias to the learning process, the resulting model could also be biased.

    “Since human involvement can potentially introduce bias to the learning process, the resulting model could also be biased.”
    Quote in the paperSubstantial claim
  • Premisereasoning

    Imaging features alone are often not sufficient to determine diagnosis and management.

    “Imaging features alone are often not sufficient to determine diagnosis and management.”
    Quote in the paperSubstantial claim
  • Interpretationcitation

    As operators seem to be at risk of expected-value bias when acquiring fetal biometry measurements, an algorithm training to measure standard biometric planes by supervised learning might end up having a built-in bias when automatically calculating fetal biometry.

    “For example, as operators seem to be at risk of expected‐value bias when acquiring fetal biometry measurements, an algorithm training to measure standard biometric planes by supervised learning might end up having a built‐in bias when automatically calculating fetal biometry 59 .”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [59] Drukker L, Droste R, Chatelain P, Noble JA, Papageorghiou AT. Expected‐value bias in routine third‐trimester growth scans. Ultrasound Obstet Gynecol 2020; 55: 375– 382. 31763735 PMC7079033 source

      Not checked in the free run.

  • Classified as probably not a substantial claim. 30
  • May combine two claims. 18

  • Probably not a substantial claim: may not pass S2 (checkable), S4 (based).

    Findingnone given

    As interdisciplinary research related to AI in clinical medicine increases, readers and reviewers with a clinical background have difficulties critically assessing the methodology of scientific AI papers because the field is largely unfamiliar to many medical professionals.

    “As we begin to see more interdisciplinary research related to AI in clinical medicine, difficulties arise when readers and reviewers with a clinical background attempt to critically assess the methodology of scientific AI papers in a field that is, for now, largely unfamiliar to many medical professionals.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: Ultrasound professionals may struggle with technical descriptions of AI methods.

  • Probably not a substantial claim: may not pass S4 (based).

    Findingnone given

    The engineering literature usually focuses on AI methodology.

    “In addition, performing a clinical systematic review in this area is challenging because most of the published peer‐reviewed scientific articles appear in the engineering literature which usually focuses on the AI methodology and few studies have assessed clinical applicability.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S4 (based).

    Findingnone given

    At this stage, AI applications are in the early stages of deployment.

    “At this stage, AI applications are in the early stages of deployment and a systematic review would be premature.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S4 (based).

    Findingnone given

    Most published peer-reviewed scientific articles on AI in this area appear in the engineering literature.

    “In addition, performing a clinical systematic review in this area is challenging because most of the published peer‐reviewed scientific articles appear in the engineering literature which usually focuses on the AI methodology and few studies have assessed clinical applicability.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Interpretationcitation

    The drop in ImageNet error rate below 5% was largely due to advances in deep learning.

    “This was largely due to advances in deep learning, the branch of AI that learns from large amounts of data.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [17] Langlotz CP, Allen B, Erickson BJ, Kalpathy‐Cramer J, Bigelow K, Cook TS, Flanders AE, Lungren MP, Mendelson DS, Rudie JDJR. A roadmap for foundational research on artificial intelligence in medical imaging: From the 2018 NIH/RSNA/ACR/The Academy Workshop. Radiology 2019; 291: 781– 791. 30990384 PMC6542624 source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Premisenone given

    Supervised AI applications require a large volume of labeled training data to achieve good performance.

    “In order to achieve good performance, supervised AI applications require a large volume of labeled training data (usually images) from which to learn.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Premisenone given

    Most methods of deep learning are supervised (training data are manually annotated).

    “Deep learning excels in pattern recognition, but it is important to recognize that most methods are supervised (training data are manually annotated).”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Premisecitation

    An AI application has higher capacity, theoretically being able to read thousands of scans, while a radiographer reads 50-100 scans per day.

    “Additionally, an AI application has higher capacity, theoretically being able to read thousands of scans, while a radiographer reads 50–100 scans per day 49 .”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [49] Topol EJ. Chapter six: Doctors and Patterns In: Deep medicine: how artificial intelligence can make healthcare human again (first edn). Basic Books: New York, 2019; 111– 135.

      Not checked in the free run.

  • May combine two claims.

    Interpretationcitation

    AI can enhance both the value and the professional satisfaction of sonographers and maternal-fetal medicine experts by reducing the time needed for routine tasks and allowing more time to perform tasks that add value and influence patient care.

    “AI can enhance both the value and the professional satisfaction of sonographers and maternal–fetal medicine experts by reducing the time needed for routine tasks and allowing more time to perform tasks that add value and influence patient care 49 , 50 .”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites [49] Topol EJ. Chapter six: Doctors and Patterns In: Deep medicine: how artificial intelligence can make healthcare human again (first edn). Basic Books: New York, 2019; 111– 135.

      Not checked in the free run.

    • cites [50] Recht M, Bryan RN. Artificial Intelligence: Threat or Boon to Radiologists? J Am Coll Radiol 2017; 14: 1476– 1480. 28826960 source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S4 (based). May combine two claims.

    Interpretationreasoning

    Deep learning performs particularly well in image pattern recognition and can be of great benefit to doctors who rely heavily on images, such as sonologists, radiographers and pathologists.

    “Deep learning, a leading tool of AI, performs particularly well in image pattern recognition and, therefore, can be of great benefit to doctors who rely heavily on images, such as sonologists, radiographers and pathologists.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

    Premisenone given

    Ultrasound AI software needs to fit into the workflow differently from CT; in ultrasound, real-time analysis at the point of acquisition is ideally needed, while in CT, automated reading is only needed at the end of the examination.

    “Ultrasound AI software needs to fit into the workflow differently from, for example, in the analysis of a CT scan; in ultrasound, real‐time analysis at the point of acquisition is ideally needed, while in CT, automated reading is only needed at the end of the examination.”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim.

    Premisenone given

    Most AI models are derived by using supervised learning, meaning that the model learns from data annotated by humans.

    “Most AI models are derived by using ‘supervised learning’, meaning that the model learns from data annotated by humans (Box 1).”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S2 (checkable), S4 (based).

    Interpretationnone given

    There is huge potential for AI to assist in repetitive ultrasound tasks, such as automatically identifying good-quality acquisitions and providing instant quality assurance.

    “Nevertheless, there is huge potential for AI to assist in repetitive ultrasound tasks, such as automatically identifying good‐quality acquisitions and providing instant quality assurance.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: AI has had little impact on obstetric and gynecological ultrasound so far.

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Interpretationreasoning

    Medical professions which rely on imaging will be the first to see the benefits of deep learning.

    “Deep learning excels in pattern recognition and we believe that medical professions which rely on imaging will be the first to see the benefits of this tool (Appendix S1).”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationreasoning

    Most academic publications use data annotated a single time by one or more human annotators, which means that the derived model will be biased by the human annotator's preferred method of annotation.

    “Most academic publications use data annotated a single time by one or more human annotators, which means that the derived model will be biased by (or skewed towards) the human annotator's preferred method of annotation.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Premisenone given

    Machines retain absolute consistency over time whereas the performance of a clinician varies depending on many factors, such as years of experience, fatigue or simple distractions.

    “An important advantage that machines have over humans is reproducibility: machines retain absolute consistency over time whereas the performance of a clinician varies depending on many factors, such as years of experience, fatigue or simple distractions, such as a late‐running clinic or a ringing phone.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

    Interpretationnone given

    No known current AI method is generic enough to be applied on a wide range of tasks; for example, an AI application designed for the second trimester is unlikely to be applicable to the first-trimester scan.

    “Compared to the image acquisition and analysis abilities of a sonologist, no known current AI method is generic enough to be applied on a wide range of tasks (e.g. an AI application designed for the second trimester is unlikely to be applicable to the first‐trimester scan).”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

    Premisenone given

    Establishing a clinically relevant, well-curated dataset that can be used to train an algorithm can be very time-intensive, and the accuracy of such curation determines the quality of the derived model.

    “Establishing a clinically relevant, well‐curated dataset that can be used to train an algorithm can be a very time‐intensive process, and the accuracy of such curation determines the quality of the derived model.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S4 (based). May combine two claims.

    Premisenone given

    Manual annotation is resource-intensive and is often subjective.

    “Manual annotation is resource‐intensive and is often subjective.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    In 2015, the ImageNet annual competition reached a milestone when the error rate of automatic classification of images dropped below 5%, which is the average human error rate.

    “In 2015, the ImageNet annual competition reached a milestone when the error rate of automatic classification of images dropped below 5%, which is the average human error rate (Figure S1) 17 .”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [17] Langlotz CP, Allen B, Erickson BJ, Kalpathy‐Cramer J, Bigelow K, Cook TS, Flanders AE, Lungren MP, Mendelson DS, Rudie JDJR. A roadmap for foundational research on artificial intelligence in medical imaging: From the 2018 NIH/RSNA/ACR/The Academy Workshop. Radiology 2019; 291: 781– 791. 30990384 PMC6542624 source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    The first ultrasound AI application that guides the user received FDA approval; the software uses AI to help the user capture images of acceptable diagnostic quality during adult echocardiography.

    “Recently, the first ultrasound AI application that guides the user received FDA approval; the software uses AI to help the user capture images of acceptable diagnostic quality during adult echocardiography 38 .”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [38] Food and Drug Administration. FDA Authorizes Marketing of First Cardiac Ultrasound Software That Uses Artificial Intelligence to Guide User https://www.fda.gov/news‐events/press‐announcements/fda‐authorizes‐marketing‐first‐cardiac‐ultrasound‐software‐uses‐artificial‐intelligence‐guide‐user. 2020.

      Not checked in the free run.

  • May combine two claims.

    Interpretationcitation

    All current AI applications are static models that can no longer adapt, and therefore the approved product does not change over time.

    “This means that all current AI applications are static models that can no longer adapt, and therefore, the approved product does not change over time.”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites [56] Chiappa V, Bogani G, Ditto A, Martinelli F, Murru G, Raspagliesi F. OP07.10: Artificial intelligence weights the importance of clinical and sonographic factors predicting nodal metastasis in endometrial cancer. Ultrasound Obstet Gynecol 2019; 54 (S1): 107– 107. source

      Not checked in the free run.

    • cites [57] Babic B, Gerke S, Evgeniou T, Cohen IG. Algorithms on regulatory lockdown in medicine. Science 2019; 366: 1202– 1204. 31806804 source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    As of May 2020, there are more than 50 deep learning-based imaging applications approved by the USA FDA or the European Union, spanning most imaging modalities including X-ray, CT, MRI, retinal OCT and ultrasound.

    “As of May 2020, there are more than 50 deep learning‐based imaging applications 37 approved by the USA Food and Drug Administration (FDA) or the European Union, spanning across most imaging modalities, including X‐ray, computerized tomography (CT), magnetic resonance imaging, retinal optical coherence tomography and ultrasound.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [37] American College of Radiology Data Science Institute. FDA Cleared AI Algorithms. https://www.acrdsi.org/DSI‐Services/FDA‐Cleared‐AI‐Algorithms [Accessed May 7th, 2020].

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

    Premisenone given

    Traditional deep-learning models can have millions of parameters and take up lots of computer memory.

    “Another issue deep‐learning scientists have to consider is deployability, as traditional deep‐learning models can have millions of parameters and take up lots of computer memory.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Premisecitation

    Most reported AI applications to date use data from a single site and focus on algorithm performance rather than looking at clinical utility or health economics.

    “Indeed, most of the reported AI applications to date use data from a single site and focus on algorithm performance rather than looking at clinical utility or health economics 60 .”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [60] Nagendran M, Chen Y, Lovejoy CA, Gordon AC, Komorowski M, Harvey H, Topol EJ, Ioannidis JPA, Collins GS, Maruthappu M. Artificial intelligence versus clinicians: systematic review of design, reporting standards, and claims of deep learning studies. BMJ 2020; 368: m689. 32213531 PMC7190037 source

      Not checked in the free run.

  • Classified as probably not a substantial claim.

    Interpretationreasoning

    Deep learning-based solutions can benefit healthcare professionals who depend heavily on information obtained from images.

    “Deep learning performs particularly well in image pattern recognition and solutions based on this approach can benefit healthcare professionals who depend heavily on information obtained from images, such as radiographers, pathologists and sonologists.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: Deep learning performs particularly well in image pattern recognition.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    The report calls for introduction of third-party auditing of AI systems, creating a system for reporting AI incidents, and encouraging researchers in academia to verify claims made by industry.

    “Among the recommendations, the report calls for introduction of third‐party auditing of AI systems, creating a system for reporting AI incidents and encouraging researchers in academia to verify claims made by industry.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [61] Brundage M, Avin S, Wang J, Belfield H, Krueger G, Hadfield G, Khlaaf H, Yang J, Toner H, Fong R, Maharaj T, Koh PW, Hooker S, Leung J, Trask A, Bluemke E, Lebensold J, O'Keefe C, Koren M, Ryffel T, Rubinovitz JB, Besiroglu T, Carugati F, Clark J, Eckersley P, de Haas S, Johnson M, Laurie B, Ingerman A, Krawczuk I, Askell A, Cammarota R, Lohn A, Krueger D, Stix C, Henderson P, Graham L, Prunkl C, Martin B, Seger E, Zilberman N, Ó hÉigeartaigh S, Kroeger F, Sastry G, Kagan R, Weller A, Tse B, Barnes E, Dafoe A, Scharre P, Herbert‐Voss A, Rasser M, Sodhani S, Flynn C, Gilbert TK, Dyer L, Khan S,

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S2 (checkable), S4 (based).

    Interpretationnone given

    Improved quality through automatic categorization or interpretation of images and ensuring images are fit for purpose can increase confidence in imaging-based diagnosis.

    “Improved quality through automatic categorization or interpretation of images and ensuring images are fit for purpose can increase confidence in imaging‐based diagnosis.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: Deep learning performs particularly well in image pattern recognition.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Ethical concerns surround the issue of privacy.

    “Ethical concerns also surround the issue of privacy 65 , 66 .”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites [65] Mittelstadt B. Principles alone cannot guarantee ethical AI. Nat Mach Intell 2019; 1: 501– 507. source

      Not checked in the free run.

    • cites [66] Geis JR, Brady AP, Wu CC, Spencer J, Ranschaert E, Jaremko JL, Langer SG, Borondy Kitts A, Birch J, Shields WF, van den Hoven van Genderen R, Kotter E, Wawira Gichoya J, Cook TS, Morgan MB, Tang A, Safdar NM, Kohli M. Ethics of Artificial Intelligence in Radiology: Summary of the Joint European and North American Multisociety Statement. Radiology 2019; 293: 436– 440. 31573399 source

      Not checked in the free run.

    • Rests on an assumption: Developing AI applications typically requires a large volume of data about patients and their diagnoses.

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

    Premisenone given

    Traditional deep-learning methods require a considerable volume of data to build accurate models, which are not always available.

    “Furthermore, traditional deep‐learning methods require a considerable volume of data to build accurate models, which are not always available.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S4 (based). May combine two claims.

    Interpretationreasoning

    In under-resourced settings, AI opens the prospect of strengthening ultrasound imaging by replicating basic obstetric ultrasound where there is none, which could allow gestational-age estimation or diagnosis of placenta previa.

    “In under‐resourced settings, it opens the prospect of strengthening ultrasound imaging by replicating basic obstetric ultrasound where there is none which could allow, for example, gestational‐age estimation or diagnosis of placenta previa.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    A recent multinational academic report provides a list of measures and mechanisms for AI developers and regulatory bodies to ensure responsible AI development.

    “Towards this goal, a recent multinational academic report addressing both medical and non‐medical AI systems, entitled ‘Toward Trustworthy AI Development: Mechanisms for Supporting Verifiable Claims’ 61 , provides a list of measures and mechanisms for AI developers and regulatory bodies to ensure responsible AI development.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [61] Brundage M, Avin S, Wang J, Belfield H, Krueger G, Hadfield G, Khlaaf H, Yang J, Toner H, Fong R, Maharaj T, Koh PW, Hooker S, Leung J, Trask A, Bluemke E, Lebensold J, O'Keefe C, Koren M, Ryffel T, Rubinovitz JB, Besiroglu T, Carugati F, Clark J, Eckersley P, de Haas S, Johnson M, Laurie B, Ingerman A, Krawczuk I, Askell A, Cammarota R, Lohn A, Krueger D, Stix C, Henderson P, Graham L, Prunkl C, Martin B, Seger E, Zilberman N, Ó hÉigeartaigh S, Kroeger F, Sastry G, Kagan R, Weller A, Tse B, Barnes E, Dafoe A, Scharre P, Herbert‐Voss A, Rasser M, Sodhani S, Flynn C, Gilbert TK, Dyer L, Khan S,

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Potential professional liability for physicians using AI is another challenge.

    “Potential professional liability for physicians using AI is another challenge 67 .”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [67] Price WN 2nd, Gerke S, Cohen IG. Potential Liability for Physicians Using Artificial Intelligence. JAMA 2019 DOI: 10.1001/jama.2019.15064. 31584609

      Not checked in the free run.

    • Rests on an assumption: Algorithms, like humans, can err.

  • Probably not a substantial claim: may not pass S2 (checkable).

    Interpretationreasoning

    In high-income settings, AI could contribute to healthcare efficiency and workflow improvements in screening.

    “In high‐income settings, this could contribute to healthcare efficiency and workflow improvements in screening.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Premisecitation

    Several original studies published in this Journal have used AI methodology to evaluate adnexal masses, risk of lymph node metastases in endometrial cancer, pelvic organ function, breast lesions, aneuploidy risk, fetal lung maturity, perinatal outcome, shoulder dystocia, brain damage, gestational age in late pregnancy, and classify standard fetal brain images as normal or abnormal.

    “Several original studies published in this Journal have used AI methodology to evaluate adnexal masses 21 , the risk of lymph node metastases in endometrial cancer 22 , pelvic organ function 23 , 24 and breast lesions 25 , 26 , 27 , assess aneuploidy risk 28 , predict fetal lung maturity 29 , perinatal outcome 30 , shoulder dystocia 31 and brain damage 32 , estimate gestational age in late pregnan”
    Quote in the paperNot a substantial claim0 of 3 sources checked
    • cites [21] Timmerman D, Verrelst H, Bourne TH, De Moor B, Collins WP, Vergote I, Vandewalle J. Artificial neural network models for the preoperative discrimination between malignant and benign adnexal masses. Ultrasound Obstet Gynecol 1999; 13: 17– 25. 10201082 source

      Not checked in the free run.

    • cites [22] Eriksson LSE, Epstein E, Testa AC, Fischerova D, Valentin L, Sladkevicius P, Franchi D, Fruhauf F, Fruscio R, Haak LA, Opolskiene G, Mascilini F, Alcazar JL, Van Holsbeke C, Chiappa V, Bourne T, Lindqvist PG, Van Calster B, Timmerman D, Verbakel JY, Van den Bosch T, Wynants L. Ultrasound‐based risk model for preoperative prediction of lymph‐node metastases in women with endometrial cancer: model‐development study. Ultrasound Obstet Gynecol 2020; 56: 443– 452. 31840873 source

      Not checked in the free run.

    • cites [23] Huang YL, Chen HY. Computer‐aided diagnosis of urodynamic stress incontinence with vector‐based perineal ultrasound using neural networks. Ultrasound Obstet Gynecol 2007; 30: 1002– 1006. 17948232 source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S2 (checkable), S4 (based).

    Interpretationnone given

    AI opens an opportunity to introduce in the patient-carer relationship a third 'participant' that is able to contribute to healthcare.

    “AI opens an opportunity to introduce in the patient–carer relationship a third ‘participant’ that is able to contribute to healthcare.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: AI uses data and algorithms to derive computational models of tasks that are often as good as (or better than) humans.
    • Rests on an assumption: Deep learning performs particularly well in image pattern recognition.
    • Rests on an assumption: Despite advances in women's ultrasound, the fundamental way in which ultrasound images are acquired and interpreted has remained relatively unchanged.

  • Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).

    Premisecitation

    The use of AI-based solutions may raise several moral questions in medicine.

    “Similarly, the use of AI‐based solutions may raise several moral questions in medicine 64 , 65 : would we trust computers to screen for disease, prioritize treatment, diagnose, treat, discharge?”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites [64] Char DS, Shah NH, Magnus D. Implementing Machine Learning in Health Care ‐ Addressing Ethical Challenges. N Engl J Med 2018; 378: 981– 983. 29539284 PMC5962261 source

      Not checked in the free run.

    • cites [65] Mittelstadt B. Principles alone cannot guarantee ethical AI. Nat Mach Intell 2019; 1: 501– 507. source

      Not checked in the free run.

37 of 37

  • The paper relies on this without support.

    AI has had little impact on obstetric and gynecological ultrasound so far.

    “Although obstetric and gynecological ultrasound are two of the most commonly performed imaging studies, AI has had little impact on this field so far.”

  • The paper relies on this without support.

    Relatively few AI-based ultrasound applications have advanced the whole way from academic concept to clinical application and commercialization.

    “It is also important to appreciate that relatively few AI‐based ultrasound applications have advanced the whole way from academic concept to clinical application and commercialization.”

  • The paper relies on this without support.

    AI uses data and algorithms to derive computational models of tasks that are often as good as (or better than) humans.

    “AI uses data and algorithms to derive computational models of tasks that are often as good as (or better than) humans.”

  • The paper relies on this without support.

    Deep learning performs particularly well in image pattern recognition.

    “Deep learning performs particularly well in image pattern recognition and solutions based on this approach can benefit healthcare professionals who depend heavily on information obtained from images, such as radiographers, pathologists and sonologists.”

  • The paper relies on this without support.

    Despite advances in women's ultrasound, the fundamental way in which ultrasound images are acquired and interpreted has remained relatively unchanged.

    “Despite all these advances, the fundamental way in which ultrasound images are acquired and interpreted has remained relatively unchanged.”

  • The paper relies on this without support.

    Algorithms and results of approved AI applications are often not published in scientific journals due to commercial sensitivities.

    “Lastly, algorithms and results of approved AI applications are often not published in scientific journals due to commercial sensitivities.”

  • The paper relies on this without support.

    Developing AI applications typically requires a large volume of data about patients and their diagnoses.

    “Developing AI applications typically requires a large volume of data about patients and their diagnoses.”

  • The paper relies on this without support.

    Algorithms, like humans, can err.

    “However, algorithms, like humans, can err.”

  • The paper relies on this without support.

    Ultrasound professionals may struggle with technical descriptions of AI methods.

    “Ultrasound professionals understand the full meaning of ‘sonographer with 10 years of experience’ or ‘images were reviewed by two specialists’, but may struggle with descriptions such as ‘A feed‐forward network of neurons consisting of a number of layers that are connected to each other was built.’ 28”
9 of 9

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • The subspecialty of obstetrics is profoundly dependent on US because of its non-invasive, cost-effective, real-time, and low-radiation characteristics for fetal scanning.

    “In particular, the subspecialty of obstetrics is profoundly dependent on the diagnostic imaging tool of US because of its non-invasive, cost-effective, real-time, and low-radiation characteristics for fetal scanning [3,4].”

    From Evolving the Era of 5D Ultrasound? A Systematic Literature Review on the Applications for Artificial Intelligence Ultrasound Imaging in Obstetrics and Gynecology (2023) · cites this paper as [3]

    The Abstract (page unavailable) describes obstetric ultrasound as one of the most commonly performed imaging studies, and the ‘Artificial intelligence and medical imaging’ section describes ultrasound analysis at the point of acquisition as real-time. These passages support its extensive obstetric use and real-time character, but our copy does not attribute that use to non-invasiveness, cost-effectiveness, or low radiation. (LLM, conf=0.85)

    Experimental — accuracy not yet measured

  • Clinical applicability of AI algorithms is still limited due to fears and concerns of clinicians regarding safety or stability of the algorithms, trustworthiness, ethical background, privacy, and professional liability.

    “the clinical applicability of AI algorithms is still limited due to fears and concerns of clinicians regarding the safety or stability of the algorithms, trustworthiness, ethical background, privacy, and professional liability [3].”

    From Evolving the Era of 5D Ultrasound? A Systematic Literature Review on the Applications for Artificial Intelligence Ultrasound Imaging in Obstetrics and Gynecology (2023) · cites this paper as [3]

    The Challenges section (page unavailable) discusses clinicians’ concerns about algorithm bias and failure, as well as safety, ethics, privacy and professional liability. The Summary says clinical deployment is at an early stage, but the paper does not establish that clinicians’ fears are the reason clinical applicability remains limited; it also points to a lack of clinical studies assessing applicability. (LLM, conf=0.83)

    Experimental — accuracy not yet measured

  • No AI method exists that can be generalized to different tasks compared with an OB/GYN specialist capable of performing US scans of different organs and the fetus in various GA periods.

    “to date, no AI method exists that can be generalized to different tasks compared with an OB/GYN specialist capable of performing US scans of different organs and the fetus in various GA periods.”

    From Evolving the Era of 5D Ultrasound? A Systematic Literature Review on the Applications for Artificial Intelligence Ultrasound Imaging in Obstetrics and Gynecology (2023) · cites this paper as [3]

  • AI is not meant to replace human work and input, but should assist them and reduce the increasing workload.

    “existing literature still emphasizes the fact that AI is not meant to replace human work and input, but should assist them and reduce the increasing workload [3,6].”

    From Evolving the Era of 5D Ultrasound? A Systematic Literature Review on the Applications for Artificial Intelligence Ultrasound Imaging in Obstetrics and Gynecology (2023) · cites this paper as [3]

  • Artificial neural networks are particularly good at image pattern recognition and can be utilized to improve the speed and accuracy of diagnosing patient conditions in medical fields that depend heavily on images and video, as in ultrasound.

    “Artificial neural networks are particularly good at image pattern recognition and can be utilized to improve the speed and accuracy of diagnosing patient conditions in medical fields that depend heavily on images and video, as in the case of ultrasound 11 , 12 .”

    From Clinical workflow of sonographers performing fetal anomaly ultrasound scans: Deep-learning-based analysis (2022) · cites this paper as [11]

Cites

Flagged first, then by citations

Research works printed but not linked: 1

  1. Gobet F., 2018, The Future of Expertise: The Need for a Multidisciplinary Approach, Journal of Expertise, 1, 107

Cited by

Sorted by citations

Related

Authors
Drukker, L., Noble, J. A., & Papageorghiou, A. T.
Affiliations
  • John Radcliffe Hospital (GB)
  • University of Oxford (GB)
Corresponding author
Papageorghiou, A. T.
ORCID
Venue
Ultrasound in Obstetrics & Gynecology
Publisher
Wiley
Published
2020-10
Volume
56
Issue
4
Pages
498-505
ISSN
  • 0960-7692 (ISSN-L)
  • 1469-0705
DOI
10.1002/uog.22122
PMC
PMC7702141
OpenAlex
W3034855613
Semantic Scholar
00fd24bbdfc975b07b0a40994df65a436f5992c2
Cited by
240
References
63
Citations by year
2026: 45 · 2025: 44 · 2024: 50 · 2023: 38 · 2022: 32 · 2021: 26 · 2020: 5
FWCI
7.43
Type
journal article
Language
en
Open access
Licence
Locations
  • Ultrasound in Obstetrics and Gynecology (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • St George's Online Research Archive (St George's University of London) (acceptedVersion)
Funders
  • Bill and Melinda Gates Foundation (49038)
  • Engineering and Physical Sciences Research Council (EP/M013774/1, EP/R013853/1)
  • H2020 European Research Council (ERC‐ADG‐2015 694581)
Topics
  • Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
  • Physical Sciences › Computer Science › Artificial Intelligence › AI in cancer detection
  • Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Medical Image Segmentation Techniques
Keywords
artificial intelligence, prenatal ultrasound, gynecologic ultrasound, deep learning, medical image analysis, clinical decision support, AI adoption in healthcare, interdisciplinary collaboration
MeSH terms
Artificial Intelligence, Female, Gynecology, Humans, Obstetrics, Pregnancy, Ultrasonography

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-10.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-10.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review