Article Text

Download PDFPDF

Abdominal ultrasound should become part of standard care for early diagnosis and management of necrotising enterocolitis: a narrative review
  1. Jacqueline van Druten1,
  2. Minesh Khashu2,3,
  3. Sherwin S Chan4,
  4. Saeed Sharif1,
  5. Hassan Abdalla1
  1. 1 School of Architecture Computing and Engineering, University of East London, London, Greater London, UK
  2. 2 Perinatal Health, Bournemouth University, Poole, Dorset, UK
  3. 3 Department of Neonatology, Poole Hospital NHS Foundation Trust, Poole, UK
  4. 4 Department of Radiology, Children’s Mercy Hospitals and Clinics, Kansas City, Kansas, USA
  1. Correspondence to Mrs Jacqueline van Druten, University of East London School of Architecture Computing and Engineering, London E16 2RD, UK; Jacqueline.vandruten{at}


Necrotising enterocolitis (NEC) is a leading cause of death and disability in preterm newborns. Early diagnosis through non-invasive investigations is a crucial strategy that can significantly improve outcomes. Hence, this review gives particular attention to the emerging role of abdominal ultrasound (AUS) in the early diagnosis of NEC, its performance against abdominal radiograph and the benefits of AUS use in daily practice. AUS has been used in the diagnosis and management of NEC for a couple of decades. However, its first-line use has been minimal, despite growing evidence demonstrating AUS can be a critical tool in the early diagnosis and management of NEC. In 2018, the NEC group of the International Neonatal Consortium recommended using AUS to detect pneumatosis and/or portal air in preterm NEC as part of the ‘Two out of three’ model. To facilitate widespread adoption, and future improvement in practice and outcomes, collaboration between neonatologists, surgeons and radiologists is needed to generate standard operating procedures and indications for use for AUS. The pace and scale of the benefit generated by use of AUS can be amplified through use of computer-aided detection and artificial intelligence.

  • necrotising enterocolitis
  • abdominal radiograph
  • abdominal ultrasound
  • differential diagnosis
  • big data
  • artificial intelligence
  • acquired neonatal intestinal disease
  • spontaneous intestinal perfusion
  • real world evidence
  • machiene learning
  • standard operating procedures
  • point of care
  • diagnostics

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

What is already known on this topic?

  • Abdominal radiographs remain the first-line modality in imaging investigations in necrotising enterocolitis (NEC).

  • Abdominal ultrasound (AUS) has been used in the diagnosis and management of NEC for a couple of decades, but not regularly or in standard fashion.

What this study adds?

  • Presents evidence of AUS as a critical tool in the early diagnosis of NEC.

  • Suggests needed procedures for the routine use of AUS and research.


Timely diagnosis and monitoring of necrotising enterocolitis (NEC) is a strategic research area aiming to improve outcomes.1 2 NEC is considered the most devastating gastrointestinal emergency in the neonatal unit. It can be unpredictable, progressing to intestinal perforation, peritonitis to shock in a rapid fashion. However, evidence suggests that earlier diagnosis and treatment can improve outcomes.3 4 Therefore, advancement of more effective point-of-care diagnostic tools is key strategic area.

NEC can present similarly to other acquired neonatal intestinal diseases (ANID). Clinical symptoms, laboratory tests and imaging help differentiate between various ANIDs.1 Without validated serum biomarkers to differentiate or monitor ANIDs, imaging is often used to confirm NEC diagnosis and initiate treatment.1 5 6 Bedside imaging has the advantages of real-time feedback and a faster turnaround time compared with serum biomarkers, at least until a point-of-care serum biomarker test is identified and adopted.7 However, imaging is challenging because imaging equipment and protocols vary and there is high variability in interpretation and reporting.8 Overcoming these challenges with standardisation of imaging protocols and interpretations would support further advancement.

The International Neonatal Consortium’s NEC subgroup was convened in 2017 to revisit the diagnostic and classification criteria for NEC.9 10 Addressing the absence of an appropriate case definition for preterm NEC, the work group proposed the ‘Two out of three’ model1 11 (see table 1). These criteria are based on clinical presentation and the visualisation of pneumatosis intestinalis (PI) or portal venous gas (PVG) by abdominal radiograph (AR) and/or abdominal ultrasound (AUS). AUS was included because various studies (including two systemic reviews and meta-analyses) demonstrated that AUS outperformed AR for diagnosis and monitoring for NEC.12–15 However, initial observations suggest significant barriers remain before AUS is routinely used to diagnose in NEC in clinical practice.12 13 16

Table 1

The VON case definition65 for NEC in comparison with the INC interim ‘Two out of three’ rule for identifying preterm NEC71

According to a recent international multispecialist survey by Ahle et al that investigated the use of various NEC ‘diagnostic’ criteria and AUS, only 75% of radiologists and 88% of clinicians would subclassify NEC into stages of Bell’s or Vermont Oxford Network (VON) criteria. Eighty-eight per cent used the VON criteria, 59% still used variants of Bell’s criteria and 8% used Gordon’s classification.3 In a follow-up multispecialist survey, Ahle et al 17 investigated the current use of imaging (AR and AUS) in NEC. There was a resounding agreement of AR as the first-line modality to visualise PI and PVG, despite clear evidence that AUS is superior. The selection of AR versus AUS was often made by the neonatologists in consultation with radiologists. In current practice, AUS was mainly requested when AR findings were inconclusive. Neonatologists reported that many performed the AUS alone, but about half of the responders reported doing the assessment with a radiology colleague.17

Ultrasound (US) has become a widely used first-line diagnostic imaging modality. Advantages over AR, CT and MRI include relatively low-cost, bedside availability and no ionising radiation. Furthermore, US is well suited for further development as a point-of-care application because it is portable, uses conventional electrical power sources and requires no shielding. These advantages are especially useful in under-resourced settings.18

During the past decade, more publications have highlighted the advantages of AUS in NEC.13–15 19–25 Recently, Cuna et al published a meta-analysis underlining the advantages of AUS over AR in NEC diagnosis.26 Similarly, a comparative trial by Chen et al showed AUS significantly outperforming AR in the prognostic prediction of NEC.12 Despite various publications demonstrating compelling evidence for the use of AUS,14 16 27 AR remains as the imaging ‘gold standard’ in NEC. This is why we ask: Should AUS become part of standard investigations for suspected NEC to support earlier NEC diagnosis and to help determine when to escalate from medical to surgical management of NEC?

Ahle et al 17 highlighted significant barriers to adopting AUS as the first-line screening modality for NEC including lack of consensus among neonatologists, radiologists and surgeons.17 Consensus is needed to indicate AUS for first-line use and provide standardised procedures in ANIDs. The survey found considerable variation in frequency of assessment, standard procedures, indications and what visualisation signs beyond PI/PVG were deemed helpful in clinical decision-making. These questions are important in defining the role of both imaging modalities in relation to other diagnostic parameters and evaluation of various imaging routines in relation to timing of surgery, complications and mortality.17

The rest of this review will largely focus on the most important sonographic findings in the course of NEC and current knowledge concerning its diagnostic, prognostic and monitoring value. It will also explore how cleaner data sets and artificial intelligence (AI) may further improve the pace and scale of benefit of using AUS.

Use of AUS in NEC for diagnosis and management


While only the ‘Two out of three’ model explicitly mentions AUS, recent surveys show that it is deployed on an ad hoc basis when findings from AR are inconclusive.17 Use is further limited due to limited availability. Based on the current evidence, AUS should be used as first-line imaging1 for NEC diagnosis and to inform decisions regarding escalation to surgical management.13 23–25


Neonatal AUS examination requires high-resolution transducers with 8–15 MHz frequency to visualise the bowel wall.20 27 The high-resolution linear array transducer is useful in assessing the bowel wall, PVG and pneumoperitoneum in smaller infants (figures 1–6).20 In larger infants, a lower frequency curved transducer can improve imaging of deeper abdominal free fluid, PVG and free abdominal air. Greyscale AUS is best for assessing the bowel wall thickness, bowel wall echotexture, bowel peristalsis and for diagnosing PI, PVG, free abdominal air and abdominal fluid collections.20 27 28 Gut perfusion can be measured with colour Doppler (CD).14 15 Spectral Doppler can be used to assess flow patterns in the superior mesenteric artery and can be useful for confirming PVG (figures 1–6).19 21 26

Figure 1

Late preterm (32–37 weeks’ gestational age) female neonate with necrotising enterocolitis (NEC) and small amount of pneumoperitoneum  on abdominal ultrasound. Greyscale ultrasound shows an echogenic interface just below the abdominal wall (arrow) with dirty posterior acoustic shadowing is from a small amount of free air in the abdomen. This air was not visible on the corresponding supine abdominal radiograph.

Figure 2

Very preterm (28–32 weeks’ gestational age) female neonate with NEC and diffuse portal venous gas. Greyscale ultrasound shows echogenic foci within the liver parenchyma that are indicative of gas bubbles within the portal veins (arrows). Note that the gas bubbles extend all the way to the periphery of the liver parenchyma. This helps differentiate portal venous gas from pneumobilia. Also note that the imaging was performed with a curved transducer allowing for complete evaluation of the liver.

Figure 3

Extremely preterm (<28 weeks’ gestational age) male neonate who was part of a set of quadruplets with NEC and thinned bowel with absent perfusion (arrows). Colour Doppler ultrasound shows thinning of the bowel wall in the epigastric region with absent blood flow with the bowel wall.

Figure 4

Very preterm (28–32 weeks’ gestational age) female neonate with NEC and diffuse pneumatosis intestinalis. Greyscale ultrasound shows two layers of echogenic foci (arrows and arrowheads) within the bowel wall indicative of gas bubbles within the bowel wall. Note that the imaging was performed with a linear transducer to get the best spatial resolution possible to see the gas within the bowel wall.

Figure 5

Extremely preterm (<28 weeks’ gestational age) female neonate with suspicion for necrotising enterocolitis . Patient was taken to surgery after these radiograph and ultrasound images were obtained and large segments of necrotic bowel were resected. This case shows (A) complex fluid collections (arrows) and (B) marked bowel wall thickening (arrows) on abdominal ultrasound. An anteroposterior abdominal radiograph from the same day (C) shows centralised loops of bowel (arrows) consistent with ascites. 

Figure 6

Term male neonate with histry of cerebral venous malformation now with acute abdominal bleeding, ultrasound obtained for concern for necrotising enterocolitis. (A) Greyscale ultrasound shows portal venous gas. Note the linear echogenic foci in the liver periphery that represent gas bubbles in the portal venous system (arrow). (B) Spectral Doppler ultrasound from the same time confirms the presence of portal venous gas with vertical lines (arrowheads) in the Doppler tracing. These lines sound like ‘pops’ if the volume is turned on while the spectral Doppler is being acquired. (C) A colour Doppler image shows bowel wall thickening (blue callipers). (D) An anteroposterior abdominal radiograph taken just 30 min before the first ultrasound images were acquired does not show portal venous gas or pneumatosis. This case demonstrates how ultrasound is more sensitive for portal venous gas since you can image over longer time periods and the contrast resolution is better so you do not need as much gas to be present to detect it.

Standard operating procedures: routines

Numerous studies13–15 19 29–34 support the use of Faingold et al 35 and Epelman et al’s27 techniques for a standardised sonographic protocol for greyscale and CD AUS. Faingold recently published a standard protocol for performing AUS for NEC,20 recommending systematically assessing each quadrant of the abdomen by scanning in transverse and sagittal planes. CD AUS at a velocity of 0.029–0.11 m/s evaluates intestinal mural blood flow with a standard protocol. Important parameters include using the lowest possible pulse repetition frequency without aliasing, a low wall filter and the highest Doppler gain settings without flash artefacts. Absent bowel wall perfusion can be assumed when no CD AUS signal is detected at the slowest possible velocity (0.029 m/s). Comparison of blood flow on CD AUS between normal and abnormal bowel loops on greyscale US imaging may also be helpful in flow assessment.20

Clinical expertise

Ahle et al 17 indicate a preference for AUS assessment as a joint multidisciplinary assessment by radiologist and neonatologists.17 36 An important limitation of AUS is that interpretation can be difficult for a novice with a steep initial learning curve. Due to limited AUS use in NEC, the required interpretation expertise can be hard to obtain in low-volume centres contributing to the high interexpert variability in interpretation. Paediatric radiologists are the specialty that has the most experience with AUS interpretation. One way to disseminate the required interpretation expertise to other interested clinicians would be to develop a robust teaching package under radiologist guidance and supervision.

Differential clinical and imaging findings

The three forms of ANID that occur most often are spontaneous intestinal perforations (SIP), classic NEC (preterm NEC) and term NEC (primarily seen in term infants).1 4 31 37 38 Initial monitoring should aim to differentiate normal feeding intolerance of prematurity (FIP) from ANIDs like NEC based on the clinical picture, radiographic findings, degree of prematurity and age at onset as illustrated in table 2 and figure 7.17 A detailed review of FIP beyond the scope of this manuscript and review by Fanaro et al.39 and other publications are recommended for further reading.39–47

Table 2

Clinical and imaging differential features for FIP39–47 and ANIDs22 31 33 38 to assist use of ‘Two out of three’ model1 11 37

Figure 7

Simplified schema of radiographic and ultrasonographic findings in necrotising enterocolitis (NEC) with correlation to the pathophysiology of NEC. The pathophysiology of NEC involves dysmature innate immune responses to intestinal microbiota or other triggers from the premature infant’s intestinal tract, leading to inflammation and injury.64 66 67 74 The onset of NEC is characterised by inflammation, increased blood flow and mucosal oedema. On CD (colour Doppler) abdominal ultrasound (AUS) this presents with a hyperaemic appearance, increased perfusion, progressive bowel wall thickening and increasing echogenicity. Bacteria penetrate the mucosal defence, and their by-products of metabolism lead to the formation of intramural gas, which can be exquisitely depicted with greyscale AUS as pneumatosis intestinalis (PI). Portal venous gas (PVG) arises from leakage of intramural gas into the mesenteric venous system, which drains into the portal vein, or even from gas-producing bacteria entering into the portal circulation. Greyscale AUS using a curved transducer allows for complete evaluation of the liver. PVG is diagnosed by seeing echogenic foci extending all the way to the periphery of the liver parenchyma. In contrast, pneumobilia is only seen centrally near the hepatic hilum. In the progression of NEC, platelet-activating factors produced by inflammatory cells and bacteria propagate the inflammatory cascade, mainly that of cytokines and complement leading to extensive transmural involvement. Eventually, there is compromise of the microvasculature such that ischaemic changes to the tissue occur. Colour Doppler AUS findings of absent transmural blood flow are indicative of this later stage. Finally, greyscale AUS showing bowel wall thinning and absent perfusion is highly suggestive of impending bowel perforation.32 33 48 73 Black framed boxes indicate abdominal radiograph (AR)/AUS features of NEC. Red framed boxes indicate pathophysiology cascade of NEC. Grey eye indicates visualisation on AUS. Note that circulation and peristalsis studies are done with colour Doppler. Black eye indicates visualisation on AR.

NEC: the visualisation of pathophysiology

Figures 1–6 show a simple schematic illustration and detailed description of the pathophysiological changes as visualised by AUS and AR.12 14 15 17 27 29 34 48–52 It is important to draw parallels between timing of physiological progression and the accompanying findings on imaging to diagnose NEC early and monitor for the need for surgery (table 3 and figures 1–6).34 48 49 51 PI and PVG remain the hallmarks of NEC on ARs.7 17 19 27 30 Bowel wall thickness, bowel wall texture, bowel necrosis53 and bowel peristalsis are all much more difficult to evaluate by AR compared with AUS.20 32 35 54 Changes in these AUS variables can provide earlier evidence of disease progression before bowel perforation.12 27 35 53 55

Table 3

Stage of NEC66 74 with correlating presentation on AUS25 27 35 and abdominal radiograph

Spontaneous intestinal perforations

Current literature on preterm and term NEC suggests a similar imaging presentation on AR. Paralytic ileus in the setting of sepsis is the most important differential diagnosis for non-perforated NEC, and SIP in perforated NEC. NEC may be distinguished from SIP, because SIP can have pneumoperitoneum without PI and PVG.56 SIP in extremely low birthweight infants (≤1000 g) typically presents with a gasless abdomen with echogenic free fluid within the first 2 weeks of life. Echogenic free fluid has a 100% sensitivity and 89% specificity, with 58% positive predictive value and 100% negative predictive value for SIP.22

Surgical NEC

The prognosis for patients with NEC worsens after bowel perforation.57 The only universally accepted indication for a surgical procedure in NEC is the presence of perforation, which may manifest as free air and/or complex fluid in the peritoneal cavity. Free air is traditionally diagnosed at AR performed with a horizontal beam but may also be detected with AUS (figures 1–6). Perforation may also lead to an accumulation of complex peritoneal fluid alone without free air.13 23 Pneumoperitoneum, bowel necrosis, complex abdominal fluid and intestinal obstruction on imaging are common indications for surgery in neonates with NEC.12 23 55 AR can diagnose ascites, but AUS is much better at differentiating between simple ascites that does not require surgery and complex ascites which is a surgical indication.13 23 24 26 Garbi-Goutel et al 58 found that AUS is a reliable modality for the prognostic monitoring of preterm NEC, whereas AR findings were not associated with complications. AUS visualisation of intramural gas was significantly associated (OR 11.8; 95% CI 1.5 to 95.8) with intestinal obstruction in NEC. In AUS, thickening of the bowel wall (OR 2.8; 95% CI 1.1 to 7.2), free air in the peritoneum (OR 8.0; 95% CI 1.4 to 44.2), complex intraperitoneal fluid (OR 3.5; 95% CI 1.3 to 9.4) and PVG (OR 3.9; 95% CI 1.2 to 12.9) were significantly associated with poor outcomes needing surgery and/or mortality.58

Superiority of AUS over AR in NEC diagnosis

Traditionally, the gold standard for imaging evaluation of NEC has been AR. Bell’s original classification system, as well as its modified version by Walsh and Kliegman, is still widely applied. This may be why the current first-line imaging modality for NEC diagnosis continues to be AR. AR is affordable, readily accessible in most hospitals and has dominated the imaging evaluation of NEC.17 AR interpretation is very difficult and AR suffers from worse interexpert variability compared with AUS. There is also a significant overlap of radiographic findings between NEC and other ANIDs.17 29 59–63 It is difficult to distinguish NEC from SIP as intestinal gas patterns are non-specific and small amounts of ascites (present in the early NEC) are very hard to see on radiograph.27 35 AR offers limited insight into disease progression. Free air visible on AR is a late finding indicating intestinal perforation.23 24 Every 6 hours, AR monitoring for NEC also exposes vulnerable neonates to ionising radiation, which may potentially increase lifetime cancer mortality risk up to 4.3 to 20-fold.64 Early in 2009, the Duke Abdominal Assessment Scale (DAAS) was proposed to improve agreement and reproducibly of NEC radiographical reporting.61 Even with DAAS, Markiet et al found that there was still very high interexpert variability when interpreting and reporting on radiological signs and hence could not validate DAAS as a reliable tool to facilitate reporting of AR findings in NEC.63

AUS is superior to AR for identifying PI and PVG19 and hence it has been incorporated in the ‘Two out of three’ model (table 1).65 66 These criteria could replace Bell’s criteria and the Vermont Oxford case definition providing a specific case definition for preterm NEC. The advantages of deploying AUS (table 3) include earlier visualisation of pathognomonic signs of NEC (PI and PVG),14 19 26 55 real-time assessment of peristalsis, vascular perfusion, bowel wall thickening and abdominal fluid. Comprehensive reviews by Bohnhorst14 and an insightful practical radiographical report by Faingold et al 35 are worth mentioning. Recently, Chen et al 12 published the first meta-analysis and systemic review of comparative trials of AR over AUS. It demonstrated that AUS had low sensitivity and high specificity for definite NEC which is crucial for decision-making for a surgeon. Chen et al 12 found several statistically significant radiographic and sonographic parameters were associated with the prognosis of patients with NEC. The study also proved AUS performance was significantly NEC (p=0.014) superior to AR in detecting NEC, with the area under the curve of the receiver operating characteristic (AUROC) of AR 0.745 (95% CI 0.629 to 0.812), being significantly lower than the AUS logistic model (AUROC: 0.857, 95% CI 0.802 to 0.946).12

Future considerations

We believe that incorporating AUS in diagnosis of NEC can lead to earlier diagnosis of preterm NEC and hence improve outcomes.7 This will involve a significant collaborative effort by neonatologists, surgeons and radiologists to devise standard imaging protocols and reporting for AUS for NEC and other ANIDs. There are a few barriers to widespread adoption that need to be overcome: lack of widespread knowledge about the procedure, lack of availability of qualified personnel to perform the procedure reliably in smaller practices and lack of data directly relating AUS to improved patient outcomes. These barriers can be addressed in the following ways. First, this paper helps to improve knowledge by concisely summarising when AUS can be used for neonatal bowel disease and how using AUS can help guide clinical management. Second, expanding the availability of AUS to allow bedside use would drive increased availability and thus adoption. Development of standard training for bedside AUS use under guidance of radiologists would help accomplish this. Third, increased standardised AUS implementation will then provide meaningful data that can be used for prospective research into impact of AUS findings on outcomes and prognosis of NEC and other ANIDs. Navigating the causality chain from diagnostic imaging interventions to patient outcomes is often difficult because so many diagnostic and therapeutic decisions are made between imaging and outcomes. However, the links in this causal chain can be more easily interrogated if there are standard data along all parts of the chain. Once good data are available on what imaging biomarkers influence clinical interventions and outcomes, a positive feedback loop could form that improves our understanding of imaging biomarker thresholds for NEC interventions.

Recent advances in AI could speed up this positive feedback loop of insights and improvements. The key input into these algorithms is large volumes of accurately labelled standardised data. Large data sets would be available from aggregation of international neonatal network data.67 However, these data are lacking due to large variation in diagnostic definitions, imaging protocols and data processing tools to aggregate images.1 13 If the medical community treating neonates can standardise our protocols and language, it would create large data sets amenable to AI predictive approaches. Much of our medical training is on an individual level (eg, attending physician to trainee mentorship), and the idiosyncrasies of individual practice will only be overcome with significant support from perinatal organisations, opinion leaders and leading researchers. Standard imaging protocols where a well-defined set of images is always acquired will help facilitate this powerful research, because this would provide standard input into AI algorithms. AI can create powerful predictive models that could bridge the gap between imaging biomarkers and clinical outcomes. Another advantage to using AI is that many clinical questions can be answered using the same data set and the predictive models can be continuously improved based on new data. AI is making observational trial design more relevant and powerful. Previously, researchers could not control for population variation in large cohorts and thus had to use randomised controlled trials of carefully selected patient populations to answer specific clinical questions (usually one at a time). AI has the ability to control for many variances and confounding variables in large samples using multivariable analysis. AI can help overcome bias by offering objective extraction of insights.9 68 AI approaches in AUS research in ANIDs can help define benefits, risks and real-life utilisation.69

AUS must be integrated into the assessment and management of NEC, similar to the neonatologist-performed echocardiography for patent ductus arteriosus.70 Collaboration with paediatric radiology teams and development of training packages and pathways will help standardise use to facilitate better diagnosis and monitoring. Ongoing efforts to generate consensus in terms of definitions and differential features are critical to generate cleaner data sets. Hopefully, data and insights generated from this collective effort can aid in further research into NEC and other ANIDs and more importantly in improving outcomes for infants.

Search strategy and selection criteria

Narrative review question: Should abdominal ultrasound, like abdominal radiograph, be recommended as part of standard care for early diagnosis and management of necrotising enterocolitis?

A literature search of Embase, SCOPUS, PubMed, CINAHL and Science Direct was conducted using a search strategy including the MeSH headings searching ‘necrotizing enterocolitis’, ‘sonography’, ‘abdominal ultrasound’, ‘AUS’, ‘computer-aided detection’, ‘CAD’ and/or ‘NEC’. Additionally, an ancestry search was performed on review articles and selected articles of interest. Reference lists of the main articles were reviewed and added if judged relevant. Publication  notifications  on new published research matching keywords where reviewed ongoing and added to existing evidence.

Only peer-reviewed articles published in English between 2008 and 2018 were considered in the primary search, and articles in multiple formats were included like reviews, systematic reviews, retrospective and prospective studies, surveys and research reports. Case reports and case series involving fewer than five patients were not considered.

A multidisciplinary team (neonatologist, radiologist, PhD candidate researcher , data scientist, medical librarian and machiene learning experts) advised on the review question and article structure after the initial literature summary from the PhD student thesis draft. The structure focused on the review question and information needed to progress current practice. Agreement between authors focused on the review question, the strength of the evidence and relevance to the structure.


We thank John Losasso, Deputy Librarian Sir Thomas Browne Library Norfolk and Norwich University Hospital for assisting in the initial literature search strategy including formulating MeSH headings. Also Prof Paul Clarke, Neonatal Researcher of Norfolk and Norwich University Hospital, for reviewing an earlier version of our manuscript and his research insights in NEC.



  • Contributors JvD: main literature search, evidence acquisition, writing of first draft, design of figures and formatting. MK: clinical insights, current NEC research insights, review of article, writing, review and formatting. SSC: clinical insight, acquisition of radiology images for table 2, additional evidence acquisition and current neonatal radiology research insights. SS: project supervisor, review article, AI insight and CAD technology insight. HA: research methodology and review of manuscript.

  • Funding JvD was awarded an educational grant by NEC UK ( that partially supported her PhD research, 2018–2019.

  • Competing interests SSC reports grants and other from Jazz Pharmaceuticals, outside the submitted work. JvD received a consultation fee from Danone to independently compile a report and present at a stakeholder meeting.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Patient consent for publication Not required.

Linked Articles

  • Fantoms
    Ben J Stenson