Article Text

Download PDFPDF

Usefulness of abdominal ultrasound in diagnosing necrotising enterocolitis
  1. Bettina Bohnhorst
  1. Correspondence to Dr Bettina Bohnhorst, Department of Pediatric Pulmonology, Allergology and Neonatology, Hannover Medical School, Carl-Neuberg-Straße 1, Hannover D-30625, Germany; bohnhorst.bettina{at}


Necrotising enterocolitis is a serious disorder in preterm infants with a mortality of up to 60%. Therefore, early and precise diagnosis and rapid initiation of proper treatment are essential. Clinically suspected diagnosis is usually confirmed by typical findings on plain abdominal radiograph, for example, pneumatosis intestinalis, portal venous gas and, in case of intestinal perforation, pneumoperitoneum. Recently, there has been growing evidence that with real-time ultrasound, intramural air and portal gas can be better detected than with x-ray. Furthermore, ultrasound is able to assess the bowel wall directly and detect bowel wall thickening or thinning, reduced peristalsis or disturbed bowel wall perfusion. Intra-abdominal fluid, both intraluminal or extraluminal is also visible. However, data regarding the diagnostic validity and prognostic value of abdominal ultrasound are limited and often focused on a single finding rather than a combination of them. Additionally, until now, ultrasound findings seem to have little influence on therapeutic decisions. Therefore, the value of abdominal ultrasound in the diagnosis of necrotising enterocolitis has to be determined by further studies until its use can be generally recommended.

  • necrotizing enterocolitis
  • abdominal ultrasound

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.


In 1888, Paltauf described five newborns who died a few days after birth, soon after developing abdominal distension and cyanosis. Postmortem examination revealed inflammation and one or multiple perforation(s) of the colon, that is, the typical pathological findings in necrotising enterocolitis.1 In the following decades only scattered reports of such cases appeared, until the term ‘necrotising enterocolitis’ (NEC) was introduced for this disease entity in the 1950s, and larger case series were reported thereafter.2 ,3 Despite important advances in neonatology, NEC still remains a devastating disorder affecting up to 10% of infants with a gestational age <28 weeks. Mortality rises up to 60% in the most immature ones.4

The aetiology of NEC is yet incompletely understood, although fundamental factors in the pathogenesis, such as prematurity, enteral feeding, intestinal colonisation by bacteria and bowel ischaemia are known. NEC mainly affects the ileum and the colon. Its clinical presentation includes gastrointestinal symptoms like feeding intolerance, abdominal distension, bloody stools and, in more advanced cases, abdominal wall erythema on one hand, and systemic signs like increased apnoea, lethargy, instability of rectal temperature and septic shock on the other.5 The clinically suspected diagnosis is usually confirmed by abdominal radiography (AR). Since the 1950s, pneumatosis intestinalis (PI) and portal venous gas (PVG) were considered to be the typical radiographic findings of NEC, and in 1960, Miller described the famous ‘football sign’ resulting from pneumoperitoneum due to perforation of the intestine.6–8

In 1978, Bell9 developed his well-known staging system of NEC based on clinical symptoms and findings in AR, which allows a grading of therapeutic interventions and facilitated a standardisation of treatment. Therapy mainly consists of withholding enteral feeding, decompression of the intestinum by use of a nasogastric tube, antibiotics and, depending on the progress of the disease, surgical intervention. Bell's original classification system, as well as its modified version by Walsh and Kliegman,10 are still widely applied, and this might be the reason why the current standard imaging modality for diagnosing NEC continues to be plain AR, although first reports on ultrasound use for diagnosing NEC date from the early 1980s.11–13

AR can depict bowel distension, to some extent bowel wall thickness, PI, PVG and free abdominal air. All these are easy to depict by ultrasound as well. Ultrasound, however, provides important additional information regarding bowel wall viability and free abdominal fluid, which might be helpful in diagnosis and management of NEC. Therefore, this review will focus on the most important sonographic findings in the course of NEC and current knowledge concerning their diagnostic and prognostic value.

Which transducers should be used? 

Due to the fine texture of the anatomical structures which have to be examined in the near field, high-resolution transducers with 8–15 MHz frequency are desirable. For evaluating the bowel, use of a linear-array transducer is preferable, while screening for PVG, free fluid and free abdominal air can possibly be done with a small convex-array transducer. Gray-scale ultrasound evaluation should be used to assess the bowel wall echotexture, to measure bowel wall thickness, to look for peristalsis, PI, PVG, free abdominal air and free abdominal fluid. Colour Doppler (CD) ultrasound is needed to evaluate blood flow in the superior mesenteric artery (SMA) and bowel wall perfusion (BWP).

Appearance of normal bowel in the neonate

The ordinary small bowel is smooth, with thin walls and obvious peristalsis. It contains fluid, gas and stool with various echogenicities (figure 1). Data on normal bowel wall thickness in preterm neonates are rare. According to personal experience, the thickness of bowel wall in preterm infants ranges between 1.3 and 2 mm. Faingold et al studied 30 neonates with mean gestational age of 37 weeks (range 24–41), and mean corrected age of 40 weeks (range 31–47), and found that bowel wall thickness ranged from 1.1 to 2.6 mm (mean 1.72 mm). Mural flow was detected in 26 out of the 30 healthy neonates with a velocity of 0.086 cm/s. According to Faingold, BWP is normal if 1–9 CD signal dots per cm2 occur (mean 3.8).14 Regarding the colon, three segments are easy to identify: the rectum behind the bladder, the colon descendens in the left flank and the colon ascendens in the subhepatic area (figure 2).

Figure 1

Appearance of ordinary small bowel in a preterm infant with a gestational age of 24 2/7 weeks, at the time of examination 6 weeks old, with a body weight of 1475 g. Grey arrow: normal bowel wall texture with different layers, the hypoechogenic layer represent the muscularis propriae. White double arrow: measurement of thickness of the bowel wall=1.6 mm; black arrows: intraluminal air with posterior acoustic shadow.

Figure 2

Normal colon ascendens in the subhepatic area in a preterm infant with a gestational age of 34 6/7 weeks and a birth weight of 2500 g on the second day of life. The colon contains hypoechogenic meconium, punctuated with echodense gas bubbles. White arrow: liver; grey arrows: bowel wall with haustra.

Which ultrasound examinations are reasonable in the course of NEC?

The following sections outline the ultrasound examinations which are meaningful to be performed in the course of NEC. Readers interested in more details are referred to the excellent review article written by Epelman et al.15

Bowel wall thickness, peristalsis and BWP

Thickening of bowel wall during NEC occurs due to mucosal haemorrhage and oedema, and its finding is proposed to reflect the earliest pathologic feature in NEC prior to the formation of PI, and was first described in ultrasound by Patel et al16 in an infant prior to other specific imaging findings consistent with NEC. In neonates, a bowel wall thickness of >2 mm should be considered as suspicious. Thickening of the bowel wall usually is accompanied by an increase in echogenicity. On the other hand, a thickness below 1.0 mm indicates an abnormal thinning resulting from ischaemia or necrosis.14 Measurements should be taken in relaxed bowel segments.

Peristaltic contractions of the small bowel normally appear more than 10 times/min.14

By contrast with AR, BWP can be depicted by using CD ultrasound. To evaluate BWP, the lowest possible pulse repetition rate and the highest Doppler gain settings without producing blooming artefacts should be used. BWP is believed to be present if CD signals are reproducible or confirmed with pulsed Doppler waveforms. Increased BWP, originating from bowel inflammation, is characterised by various specific flow patterns: ring-shaped CD signals result from enhanced circumferential flow around the entire bowel wall; a Y-shaped pattern is noticed due to additional enhanced CD Doppler flow in distal mesenteric and subserosal vessels, and in longitudinal scans the ‘zebra pattern’ may be observed, which consists of multiple, parallel CD lines and reflects hyperaemia of the mucosal folds of the small intestine.14

Absent BWP can be assumed when no CD signal is detected at the slowest possible velocity (0.029 m/s in the study of Faingold).

Intramural gas

Intramural gas consists of gas bubbles along the subserosal and submucosal layers, and is generally called PI. Intramural gas emerges due to the passage of intraluminar air, delivered by bacterial fermentation of intestinal contents, into the injured bowel wall.17 On ultrasound, it appears as highly echogenic dots in the bowel wall (figure 3). The amount of intramural gas may vary between only a few foci and involvement of the whole circumference of the bowel, that is, the ‘circle sign’,18 thus making the normal structure of the bowel wall unrecognisable. Animal studies confirmed the correlation of this sonographic finding with histopathological findings of NEC.19 PI is most commonly seen in the distal small bowel and the colon. The challenge is to distinguish PI from intraluminal air, that is, ‘pseudo-PI’, being particularly difficult in non-dependent parts of the gut, or if the PI does affect the circumference only partially. Gentle compression of the abdomen might help in such cases, since this moves intraluminal gas bubbles, while PI will not change its position.18

Figure 3

Term neonate with double-outlet-right ventricle suffering from necrotising enterocolitis after cardiac surgery. Some small bowel loops are seen surrounded by echofree fluid (grey arrows). The upper left loop presented with intramural gas (white arrow).

Portal venous gas

The accidental detection of PVG by ultrasound in a previously asymptomatic preterm infant who developed clinical and radiological signs of NEC thereafter, and died 24 h later, was first described by Malin et al.11 During the following years, some case reports and case series were published, confirming the relevance of that finding.12 ,20–22 PVG is believed to originate from the absorption of intramural gas into the intestinal venous system travelling into the portal vein thereafter, thus PVG is as early a finding as PI.

For evaluation of PVG, the portal vein should be adjusted in a transverse or longitudinal section of the liver and can easily be identified due to its undulating flow pattern. PVG appears as highly echogenic round particles with a diameter of approximately 1 mm, resulting from microbubbles. Due to their large acoustic impedance these can easily be differentiated from the low echogenic flat background noise of the ordinary blood stream (figure 4). Usually, the microbubbles move quite fast through the portal vein, but sometimes they migrate comparatively slowly. The pattern might be intermittent and the number of microbubbles ranges from few to numerous. The microbubbles might be trapped in the small branches of the portal vein, and then appear as dendriform granular echogenicities in the liver parenchyma (figure 5). In the usually supine patient, this pattern is observed particularly in the non-dependent, ventral parts of the liver.12

Figure 4

Same patient as in figure 3. Portal venous gas is seen in terms of microbubbles in the truncal of the portal vein (white arrow) and as dendriform echogenicities in the smaller branches of the portal vein in the liver parenchyma (grey arrows).

Figure 5

Extended portal venous gas in the smaller branches of the portal vein appearing as dendriform granular echogenicities in the liver parenchyma in a preterm infant with a gestational age of 26 4/7 weeks and a chronological age of 7 weeks, body weight 1200 g.

Free abdominal gas

Free abdominal gas results from bowel perforation. Therefore, in the evaluation of NEC, it is a sign of a fatal course (Bell stage IIIb). With ultrasound, free abdominal gas is detected by using a transverse section of the right upper abdomen showing the liver just below the diaphragm. Without compression of the abdominal wall, free air will present as a bright echogenicity between the abdominal wall and the anterior surface of the liver. Applying a moderate compression by the scanner, free air is displaced, and the liver parenchyma is depicted next to the abdominal wall. A small amount of free air could appear as single hyperechoic foci between anterior abdominal wall and liver. In case of free abdominal fluid, free air might rise to the surface, shown as echogenic foci throughout the fluid.15

Free abdominal fluid

Free abdominal fluid is frequently detected close to the bladder, but might occur in the entire abdomen between the bowel loops. In neonates, a small amount of intraperitoneal echo-free fluid is a common finding. The detection of a localised fluid formation or of echogenic material between the bowel loops, however, is highly suspicious of bowel perforation.23 The detection of an intraperitoneal mass with a hypoechoic periphery and an echogenic center is highly suspicious of an intra-abdominal abscess.13

Nevertheless, it has to be remembered that all these findings are not specific for NEC but might also indicate bowel perforation due to meconiumileus or spontaneous intestinal perforation.

Doppler sonography of the SMA

To assess the flow of the SMA, the transducer should be held in a longitudinal section just below the xiphoid. The mean value of the SMA diameter in healthy neonates is 3.2 mm (range 2.7–3.7 mm), the mean peak velocity is 57±3.1 cm/s.24 In preterm infants, the mean peak flow velocity varies between 34 and 56±6 cm/s depending on the amount of daily milk intake. Enteral feeding significantly raises the peak systolic and average flow velocities in the SMA in both preterm and term neonates with a maximum at 45 min after the feeding (mean values 88–97±11 cm/s).25 ,26

Diagnostic and prognostic validity of ultrasound findings in the diagnosis of NEC

Studies dealing with ultrasound findings in the diagnosis of NEC are often retrospective cohort studies. An overview of them including their characteristic features is provided in table 1. Altogether, these studies included 190 infants with confirmed NEC, 259 infants with suspected NEC and 711 controls.

Table 1

Overview on studies dealing with the usefulness of ultrasound in diagnosing NEC in neonates

PVG was the first pathognomic finding in NEC depicted by ultrasound and is considered in most studies. The leadoff studies done in this field suggested that PVG may be detected earlier by ultrasound than by AR.12 ,20–22 This may be the reason why PVG detected by ultrasound is not necessarily accompanied by such a poor prognosis as formerly presumed based on AR findings.15 Nevertheless, recent larger studies showed that PVG detected by ultrasound in NEC of all stages has a sensitivity of only 16–45%.27–29 In NEC confirmed by intraoperative findings as gold standard, the sensitivity is remarkably higher and reaches 82%.30 In any case, PVG in ultrasound does not differentiate between NEC stages.28

The specificity of PVG in ultrasound reaches 90–98%, since apart from suspected or confirmed NEC there is only one other neonatal disease which is associated with PVG in ultrasound, called ‘benign pneumatosis coli’.12 ,27–31

Sensitivity and specificity of PVG in AR have not yet been calculated, but detection rates of PVG by AR vary from 2.6% to 58%, thus suggesting low sensitivity.14 ,29 ,30 ,32–34

PI is detected by ultrasound in 13–100% of infants with NEC of all stages, the corresponding value for AR ranges from 20% to 95%.14 ,20 ,22 ,27 ,29 ,30 ,33 ,34–36 Recently, PI was detected by AR impressively more often than by ultrasound in cases of advanced NEC (74% vs 13%).29 Our own study on patients with surgically confirmed NEC revealed a sensitivity of 75% and a specificity of 91% for detection of PI and/or PVG in AR in diagnosing NEC.30 Therefore, data do not yet confirm a clear advantage of ultrasound over AR in depicting PI in advanced NEC.

In early-stage NEC, detection of PI using ultrasound might be more accurate. In one study, PI was detectable by ultrasound in all 40 neonates with the clinical diagnosis of early-stage NEC (Bell stage I) with bowel distension, but no evidence of PI on AR; 40% showed scattered echogenic dots, and 60% dense granular echogenicities of the bowel wall. PVG was absent in all patients. In those with more advanced PI, reintroduction of enteral feedings, based on clinical estimation, was significantly delayed.35 The authors concluded that bowel ultrasound might be helpful for the early diagnosis of NEC, although the definition of a gold standard to confirm the diagnosis is missing.

Studies regarding depiction of viability of bowel wall in NEC of all stages showed that bowel wall thickening, presence of PI and absence of peristalsis were not specific for necrosis. On the other hand, all patients with severe NEC (Bell stage III) presented with absent bowel perfusion in ultrasound, matching bowel necrosis in all of them (as confirmed by laparotomy or the presence of perforation later in the course), giving a sensitivity of 100%. By comparison, the sensitivity of free air on AR (indicating bowel necrosis with perforation) was only 40%.14 Moreover, thinning of the bowel wall (<1 mm) was highly suggestive for severe ischaemia and seen in 8 of 12 neonates with bowel necrosis. The authors concluded that ultrasound is more accurate than AR in depicting bowel necrosis in NEC. However, exact examination of the entire bowel is a time consuming procedure, which took approximately 25 min in the above mentioned study.

Another retrospective study revealed that increased wall echogenicity, bowel wall thickening or thinning, and free echogenic fluid in ultrasound were significantly correlated with an adverse outcome.36 When three of the seven sonographic features (PVG, PI, increased wall echogenicity, bowel wall thickening or thinning, absent perfusion, free echogenic fluid) were present, there was a sensitivity of 0.82 (95% CI 0.60 to 0.95) and a specificity of 0.78 (95% CI 0.52 to 0.94) for poor outcome. The authors emphasised that ultrasound was very helpful in cases where a diagnosis of NEC had been established clinically, but patients were not responding to medical therapy and had non-specific AR findings. In these cases, ultrasound seemed to be important for determining necrotic bowel prior to perforation.

The latter statement is supported by two other studies. In one of them, echogenic ascites was noted by ultrasound in 12 out of 23 infants with advanced NEC who did not show free air in AR. All those 12 infants proved at surgery to have suffered from bowel necrosis and perforation.37 In the other study with nine infants having advanced NEC, echogenic free fluid was observed, in only five of them, free air could be depicted with AR. All of them required surgical intervention due to deterioration.29

All studies dealing with sonographic examinations of SMA blood flow velocities discovered significantly elevated flow velocities (up to 120 cm/s) in infants suspicious of NEC.38–41 It is speculated that the elevated flow velocities arise from vasoconstriction of the ischaemic bowel. Anyway, the great variation of the values does not allow for a precise distinction between suspected and confirmed cases, and the duration of the elevated SMA velocity was not specified. One study addressing ultrasound findings in an experimental rabbit model of NEC suggested that the increased flow velocity might only be a transient finding lasting just a few hours and may thus easily be overlooked.42


The obvious advantages of ultrasound over AR are that it can be performed repeatedly and at the bedside any time, avoids radiation exposure and gives a broader picture of the dynamic process of NEC. To assess the intestine in preterm infants properly, however, high-quality ultrasound scanners with high-megahertz linear-array transducers are indispensible.

In uncomplicated NEC, ultrasound probably has limited diagnostic impact, but in cases in which the radiological findings are not in keeping with the clinical course, or if the clinical condition is worsening without evidence of pneumoperitoneum on AR, ultrasound might be helpful in diagnosing NEC.15

In practice, the neonatologist should use ultrasound

  • For early detection of PVG, which is pathognomonic of NEC and, due to its high acoustic impedance, can easily and quickly be noticed even by less experienced investigators

  • To look for free abdominal fluid, as the presence of echogenic free fluid is highly suspicious of bowel perforation

  • To assess BWP, because in the detection of necrotic bowel absent BWP in ultrasound is more sensitive and specific than AR.

In particular, the last two of these ultrasound findings might help in the decision making of surgical intervention. However, evaluation of the bowel is a challenging, time-consuming procedure, which might be a disadvantage in a critically ill infant. In addition, large amounts of bowel gas make evaluation of bowel wall impossible, the right appreciation is highly dependent on the experience of the examiner, and as yet no study has assessed interobserver and intraobserver variability in NEC.

So far, the validity and the prognostic value of sonographic findings in the course of NEC have been poorly studied. Data on how sonographic findings affect therapeutic management are still missing, and it has not yet been established when and how often ultrasound should be performed in the course of NEC.

Consequently, plain AR and ultrasound should be regarded as complementary imaging modalities for NEC, and no imaging modality can substitute for a close clinical examination.


The author is grateful to Doris Franke, MD, for helpful discussions.



  • Competing interests None.

  • Provenance and peer review Commissioned; externally peer reviewed.