The development of techniques such as ultrasound, computed tomography (CT), radionuclide scintigraphy and magnetic resonance imaging has greatly enhanced our diagnostic capabilities in children. Many conditions still require conventional radiographic examinations and contrast studies to make a diagnosis. The chosen imaging modality will be the one that provides most diagnostic information in the shortest time while minimizing the discomfort and radiation dose to the child. In this latter respect ultrasound and magnetic resonance imaging have the advantage that they do not use ionizing radiation.

Contrast studies of the gastrointestinal tract are usually performed with barium, but when aspiration or intestinal perforation are likely to be present the newer low osmolarity water-soluble contrast media are safer. The use of Gastrografin® is rarely indicated, except possibly in the treatment of meconium ileus. Many congenital abnormalities, such as diaphragmatic hernia, duodenal atresia, omphalocele, and congenital hydronephrosis can be diagnosed antenatally by ultrasound, and the obstetrician and paediatric surgeon forewarned. If appropriate the mother can then be transferred prior to delivery to a paediatric surgical unit. It is beyond the scope of this section to consider antenatal diagnosis and paediatric surgical conditions occurring outside the abdomen.

The diagnosis of oesophageal atresia is usually suspected clinically and can be confirmed on a chest radiograph, which will show a soft catheter passed into the oesophagus coiling back on itself. Further confirmation can be obtained by injecting a small quantity of air down this catheter: this will distend the proximal oesophageal pouch. Contrast medium should not be injected as this may lead to aspiration.

A tracheo-oesophageal fistula is present in 85 per cent of infants with oesophageal atresia, and an abdominal radiograph will show gas within the intestine: the abdomen is gasless in those without a fistula. Tracheo-oesophageal fistula can occur in isolation: contrast studies of the oesophagus may be necessary to demonstrate the fistula, which often runs upwards from the oesophagus to the trachea in the lower part of the neck.

Imaging studies are also necessary to demonstrate coexisting abnormalities, which are present in about 50 per cent of patients. These include vertebral and renal abnormalities, duodenal and anal atresia, and congenital heart disease.

Since ultrasound uses non-ionizing radiation, it has now replaced barium studies in the diagnosis of pyloric stenosis in the small percentage of babies in whom the clinical diagnosis is in doubt. The interpretation of ultrasound is operator-dependent, and if such expertise is not available a barium study should be performed.

Ultrasound clearly demonstrates the hypertrophied muscle, and the length and thickness of the pyloric muscle can be measured to confirm the diagnosis (Fig. 1(a)) 168.

Barium studies demonstrate a narrow elongated pyloric canal with indentations on the gastric antrum and duodenal bulb from the thickened muscle (Fig. 1(b)) 169.

Duodenal obstruction may arise from a number of causes (Table 1) 117; the radiographic appearances depend on whether or not the obstruction is complete. If it is complete, as in duodenal atresia, the classical ‘double bubble’ appearance of air and fluid in the distended stomach and proximal duodenum is seen (Fig. 2) 170. If the baby has recently vomited or if a nasogastric tube has been passed the abdomen may appear almost gasless. In this situation the obstruction can be seen if 15 to 20 ml of air are injected through the nasogastric tube. Plain radiographs may be virtually normal in patients with partial obstruction, and contrast studies may be necessary to demonstrate abnormalities such as a duodenal web or bowel malrotation.

During the embryonic period the duodenojejunal loop rotates 270° around the superior mesenteric artery axis in an anticlockwise direction. The caecocolic loop, which initially lies inferiorly to the superior mesenteric artery, also rotates 270° in an anticlockwise direction. Finally the caecum and ascending colon become fixed to the posterior peritoneum. If this process is interrupted at any point then malrotation or non-rotation results.

Intestinal obstruction, with the possibility of vascular compromise, is due to either an associated volvulus or extrinsic compression from peritoneal Ladd's bands. Abdominal radiographs may either suggest duodenal or small bowel obstruction. The diagnosis of malrotation is made by an upper gastrointestinal tract contrast study. This demonstrates the duodenojejunal flexure and ligament of Treitz to be in an abnormally low position or in the right side of the abdomen rather than in their normal position, which is in the left side of the abdomen on a level with the duodenal bulb. Although the position of the caecum is also abnormal, its position is variable in the neonate, and contrast studies of the upper rather than the lower gastrointestinal tract are therefore more reliable.

If a volvulus is present then the affected segment of bowel will appear twisted, whereas Ladd's bands give an extrinsic impression on the bowel outline. Rarely, malrotation presents less acutely, with cyclical abdominal pain and vomiting or malabsorption.

Abdominal radiographs show dilated loops of small bowel with air–fluid levels on erect and decubitus views. It can be difficult to differentiate between small and large bowel obstruction in young infants, because the normal differentiating features of valvulae conniventes of the small bowel and haustrae of the colon are not apparent in this age group. The anatomical position of the dilated loops of bowel may be helpful but it may be necessary to perform a contrast study of the colon simply to define the level of obstruction. It is also important to differentiate between an ileus and obstruction. Infants often develop ileus secondary to septicaemia and metabolic disturbances.

Occasionally a ‘bubbly’ appearance is present in the right side of the abdomen; this is due to meconium mixed with air and may be seen in meconium ileus, ileal atresia, or Hirschsprung's disease. If perforation occurs during the intrauterine period the spilt meconium produces a sterile peritonitis, with the formation of peritoneal calcification or, less often, rim calcification in a ‘meconium pseudocyst’. Ascites may be present.

This condition, seen almost exclusively in babies with cystic fibrosis, causes small bowel obstruction due to thick tenacious meconium in the distal ileum. This prevents the normal bowel contents from entering the colon, which, as a result, is often a ‘microcolon’. Ileal atresia is usually associated with a colon of normal calibre, as it is frequently caused by a vascular accident late on in fetal life.

The abdominal radiograph shows dilated small bowel loops, but is often remarkable for the lack of fluid levels due to the very sticky meconium within the bowel. Fifty per cent of babies with meconium ileus have other abnormalities, such as intestinal atresia or volvulus, and this should be suspected if many fluid levels are present on the radiograph.

The diagnosis of meconium ileus is confirmed by a water-soluble contrast study of the large bowel with reflux of contrast into the distal ileum. This will demonstrate ‘balls’ of inspissated meconium in the terminal ileum (Fig. 3) 171.

Some success has been reported in the non-operative treatment of meconium ileus. Hyperosmolar water-soluble contrast material is refluxed amongst the impacted meconium and may relieve the obstruction both through its lubricating effect and the fact that it pulls fluid into the bowel lumen. Reported success rates for this procedure vary but a rate of 50 per cent can be achieved. The procedure is, however, only appropriate in uncomplicated cases, which account for less than half the total number. The baby has to be very closely monitored as the hyperosmolar effect of the contrast media may lead to considerable fluid and electrolyte imbalance.

In infants over the age of 3 months the most common causes of small bowel obstruction are intussusception and obstructed herniae. The appearance of intussusception on plain radiographs vary from a virtually normal radiograph to a picture of marked small bowel obstruction. A soft tissue mass of the intussusception itself will be visible in about one-half of the patients, or there will be an absence of the normal colonic gas pattern in the right side of the abdomen (Fig. 4) 172. The diagnosis can be confirmed by either ultrasound or contrast examination of the colon.

Once the diagnosis has been confirmed a decision has to be taken whether to attempt non-operative reduction with either barium or gas or whether to proceed directly to surgery. This requires close co-operation between the surgeon and the radiologist. The surgeon must carefully assess the child: signs of peritonitis and bowel ischaemia are contraindications to non-operative reduction, as is radiographic evidence of perforation.

Non-operative treatment has for many years relied on hydrostatic pressure from a barium or water-soluble contrast enema to reduce the intussusception. Recently there has been renewed interest in the use of gas, with suggestions that better rates of reduction can be achieved; in addition, the procedure is cleaner and quicker. Reported series suggest a successful reduction rate using barium of between 50 and 80 per cent, whereas most studies using gas report rather better figures of 75 to 95 per cent. Factors which suggest that the procedure is less likely to be successful are a history of longer than 24 h, the presence of small bowel obstruction, and an intussusception in the distal colon. The incidence of perforation of the bowel during non-operative reduction is between 0.4 and 1.2 per cent. If gas is being used then perforation may lead to respiratory embarrassment due to gross pneumoperitoneum. The spillage of barium and faecal material into the peritoneum was thought to be associated with a very high morbidity: recent studies suggest that an immediate laparotomy reduces the long term sequelae to a minimum. The rate of recurrence of the intussusception following non-operative reduction is similar to that following surgery (4–10 per cent).

The causes of large bowel obstruction are listed in Table 3 119. As previously discussed the differentiation between large and small bowel obstruction may be difficult using plain radiographs alone. It may be necessary to perform a contrast study of the colon to define the level of obstruction as well as to differentiate obstruction from ileus.

The majority of infants with Hirschsprung's disease present with large bowel obstruction in the first week of life. The remainder usually present in the first 3 years with constipation. The definitive diagnosis is made by rectal biopsy. Plain abdominal radiographs suggest a distal large bowel obstruction or, in the rare instance of total aganglionosis of the colon, small bowel obstruction. Contrast studies using low osmolality water-soluble contrast will define a transition zone in 70 per cent of affected infants (Fig. 5) 173. This transition zone demarcates the proximal normally innervated and dilated bowel from the distal abnormally innervated bowel which is of normal or reduced calibre.

This condition occurs most commonly in premature infants and infants of diabetic mothers. The colon is normally innervated but plain radiographs suggest large bowel obstruction. Contrast studies demonstrate plugs of meconium obstructing the distal colon. The condition is probably due to functional immaturity of the colon and is self limiting. The act of performing a contrast study will dislodge the meconium.

These include ectopic anus, imperforate anus, rectal atresia, and anal or rectal stenosis. In these conditions, imaging defines the relationship of the atretic segment to the levator ani pelvic sling and the integrity of sphincters and their nerve supply. Both CT and magnetic resonance imaging can be used for this purpose. Fistulae can be demonstrated by contrast studies of the bladder, urethra, and vagina, and associated malformations such as renal ectopia, hydronephrosis, spinal and cardiac anomalies, and oesophageal or duodenal atresia can also be visualized.

Prior to complex imaging of the pelvic musculature and renal and gastrointestinal tracts, plain radiographs will show a degree of large bowel obstruction. Spinal abnormalities will also be apparent and if a fistula is present gas may be visible within the bladder (Fig. 6) 174. A ‘cross table’ radiograph, with the baby prone and with its pelvis raised, is sometimes used to assess the distance between the most distal loop of bowel containing gas and the perineum. This may be misleading as the bowel proximal to the atresia may be plugged by meconium and not visible on the radiograph. This film should be delayed until the baby is about a day old to allow gas to reach the distal bowel.

The diagnosis of necrotizing enterocolitis is both a clinical and radiological one, as radiographic abnormalities may not always be apparent. Radiographic abnormalities include intestinal ileus and distension, and bowel wall thickening. Intramural gas leads to lucencies in the bowel wall which may be linear, circular, or ‘bubbly’ (Fig. 7) 175. Air may pass from the bowel wall into the mesenteric venous system and become apparent in the portal vein. If perforation occurs, free air will be visible in the peritoneal cavity and abscesses may form. Free air is best seen on decubitus radiographs but the first signs, including visualization of both sides of the bowel wall, the falciform ligament, and umbilical artery remnants, may be detected on a supine film. Long term sequelae are stricture formation and, rarely, enterocolonic fistulae. These can be demonstrated by contrast studies of the bowel.

In the immediate neonatal period most abdominal masses are renal in origin, due either to cystic disease or to hydronephrosis. Other masses occurring at birth are omental, mesenteric or duplication cysts, choledochal cysts, ovarian cysts, and adrenal haemorrhage. Tumours such as Wilms' tumour and neuroblastoma are more common around the age of 2 or 3 years; liver tumours, which are much less common, can occur at any age. The role of imaging is primarily to define the organ of origin of the mass, to define its nature and, if the mass is a tumour, whether there is evidence of spread. Ultrasound is the initial investigation of choice. This may be supplemented by CT or magnetic resonance imaging where appropriate.

Hydronephrosis may be due to obstruction, vesicoureteric reflux or the prune belly syndrome. Obstruction most commonly occurs at the pelviureteric junction. Ultrasound determines whether hydronephrosis is present, whether it is unilateral or bilateral, whether the ureters are dilated, and whether the bladder and posterior urethra are normal. It gives no information about function, however: the intravenous urogram provides some information about function, but this is best assessed by radionuclide diuretic renography using technetium-99m either labelled to diethylene-triamine-penta-acetate (DTPA) or mercaptoacetyltriglycine (MAG 3) renography.

There are many forms of renal cystic disease in infancy. The most common cause of an abdominal mass is multicystic dysplastic kidney, which must be differentiated from gross hydronephrosis as renal cysts and dilated calices can be confused. Multicystic dysplastic kidney is usually unilateral, and a number of non-communicating cysts of varying size are seen. An anomaly of the contralateral kidney, such as pelviureteric obstruction, is occasionally seen. A technetium-99m labelled dimercaptosuccinic acid (DMSA) scintigram will demonstrate a non-functioning kidney. A micturating cystourethrogram should also be performed to assess whether vesicoureteric reflux is occurring on the contralateral side. Infantile polycystic kidney disease is almost invariably bilateral. The ultrasound appearances of large highly echogenic kidneys are characteristic, although the actual cysts are usually small and therefore not seen.

Wilms' tumour is the most common urinary tract tumour of childhood, and together with neuroblastoma, it is the most common solid organ tumour outside the central nervous system. Initial imaging with ultrasound or urography demonstrates an intrarenal solid mass which may expand within the kidney or protrude from the surface (Fig. 8) 176. Areas of cystic necrosis may be present. Calcification is seen in about 5 per cent of patients on plain radiographs and in 10 per cent of those examined with CT. Chest radiography, ultrasound, and CT are used for tumour staging. Close attention is paid to the presence of pulmonary or hepatic metastases, lymph node involvement, and involvement of the contralateral kidney, which occurs in 5 per cent of cases.

Neuroblastoma arises in the adrenal glands or anywhere in the sympathetic chain. The imaging protocol depends on the site of origin: about 70 per cent of these tumours occur within the abdomen and half of these originate in the adrenal gland. Within the abdomen, ultrasound demonstrates a solid extrarenal mass which may displace the kidney. Calcification is seen in 50 per cent of plain radiographs and in up to 80 per cent of CT studies. CT is also used to assess local spread, including spread into the spinal canal, and metastatic disease. Scintigraphy is required to detect skeletal metastases. Iodine-131 or iodine-123- m-iodobenzylguanidine (MIBG) may be used to detect the primary tumour and any metastases, as well as monitoring response to therapy. The role of magnetic resonance imaging in both neuroblastoma and Wilms' tumour is still being evaluated.

Imaging studies in children with urinary tract infection allow the early detection of conditions such as hydronephrosis and vesicoureteric reflux, which may lead to irreversible renal damage. If damage has occurred, the severity can be assessed.

As previously stated, hydronephrosis can be evaluated by ultrasound, intravenous urography, and renal scintigraphy. Vesicoureteric reflux is assessed by a micturating cystourethrogram (Fig. 9) 177. This may also demonstrate urethral abnormalities such as posterior urethral valves, ectopic ureters, and ureteroceles. Less commonly, reflex can be demonstrated by direct or indirect radionuclide cystography.

Renal scarring is best demonstrated by technetium-99m labelled dimercaptosuccinic acid (DMSA) scintigraphy (Fig. 10) 178. This also permits quantification of split renal function. Ultrasound is unreliable in the detection of mild or even moderate degrees of renal scarring.

In the young infant (under 1 year of age) an ultrasound scan to detect hydronephrosis and a micturating cystourethrogram to detect reflux are performed. Between the age of 1 and 5 years there is debate as to the most appropriate investigations. One approach is to perform an ultrasound scan and a DMSA scintigram initially, followed by a cystogram if either show scarring or a dilated urinary tract, suggestive of reflux. After the age of 5 years an ultrasound scan and a plain abdominal radiograph, to detect renal calcification and spinal anomalies, are probably all that is required. In all age groups, if an abnormality is detected, then the appropriate imaging protocol will be followed depending on whether obstruction, reflux, or scarring are suspected.

The diagnosis of biliary atresia must be made quickly so that surgery can be performed in the first 2 months of life. Ultrasound is performed initially to exclude other causes of biliary obstruction such as choledochal cyst, although the two may coexist. The gallbladder is not usually visualized in patients with biliary atresia, but examination of the liver is otherwise normal. Following ultrasound examination biliary excretion is assessed using either [¹²³I]- bromosulphthalein or [&sup9;&sup9;Tc&supm;] iminodiacetic acid compounds. Reported sensitivity in differentiating atresia from hepatitis varies between 85 and 100 per cent. The final diagnosis and differentiation may depend on liver biopsy and operative cholangiography.

This may occasionally be detected on plain radiographs as a fluid filled or air filled mass containing debris, or a calcified enterolith may be seen. Contrast studies of the small intestine may fill the diverticulum but the standard method of detection relies on technetium scintigraphy. Provided that the diverticulum contains ectopic gastric mucosa then the scintigram should be positive.

Aaronson IA, Cremin BJ. Cysts of the kidney. In: Clinical Paediatric Uroradiology. Edinburgh: Churchill Livingstone, 1984: 76–99.

Aaronson IA, Cremin BJ. Tumours of the urinary tract. In: Clinical Paediatric Uroradiology. Edinburgh: Churchill Livingstone, 1984: 307–34.

Bell RS, Graham CB, Stevenson JK. Roentgenologic and clinical manifestations of neonatal necrotising enterocolitis. Am J Roentgenol, 1971; 112: 123–34.

Berdon WE, Slovis TL, Campbell JB, Baker DH, Haller JO. Neonatal small left colon syndrome: its relationship to aganglionosis and meconium plug syndrome. Radiology, 1977; 125: 457–62.

Bisset GS, Kirks DR. Intussusception in infants and children: diagnosis and therapy. Radiology, 1988; 168: 141–5.

Boechat MI, Kangarloo H. MR Imaging of the abdomen in children. Am J Roentgenol, 1989; 152: 1245–50.

Boechat MI, et al. Primary liver tumours in children: comparison of CT and MR imaging. Radiology, 1988; 169: 727–32.

Bower RJ, Sieber WK, Kiesewetter WB. Alimentary tract duplications in children. Ann Surg, 1978; 188: 669–74.

Carty H, Pilling, DW, Majury C. Iodine-123 BSP scanning in neonatal jaundice. Ann Radiol, 1986; 29: 647–50.

David R, et al. The many faces of neuroblastoma. Radiographics, 1989; 9: 859–82.

Gans SL. Classification of ano-rectal anomalies: a critical analysis. J Pediatr Surg, 1970; 5: 511–13.

Gu L, et al. Intussusception reduction in children by rectal insufflation of air. Am J Roentgenol, 1988; 150: 1345–8.

Hibi S, Todo S, Imashuku S, Miyazaki T. Iodine-131 meta-iodobenzylguanidine scintigraphy in patients with neuroblastoma. Pediatr Radiol, 1987; 17: 308–13.

Hope JW, Borns PF, Berg PK. Roentgenologic manifestations of Hirschsprung's disease in infancy. Am J Roentgenol, 1965; 95: 217–29.

Houston CS, Wittenborg MH. Roentgen evaluation of anomalies of rotation and fixation of the bowel in children. Radiology, 1965; 84: 1–17.

Kluth D. Atlas of oesophageal atresia. J Pediatr Surg, 1976; 11: 901–19.

Kohda E, Fujioka M, Ikawa H, Yokoyama J. Congenital anorectal anomaly: CT evaluation. Radiology, 1985; 157: 349–52.

Leonidas JC, Berdon WE, Baker DH, Santulli TV. Meconium ileus and its complications, a reappraisal of plain film roentgen diagnostic criteria. Am J Roentgenol, 1970; 108: 598–609.

Martin GI, Kutner FR, Moser L. Diagnosis of Meckel's diverticulum by radio isotope scanning. Pediatrics, 1976; 57: 11–12.

Nussbaum AR, Sanders RS, Hartman DS, Dudgeon DL, Parmley TH. Neonatal ovarian cysts: sonographic–pathologic correlation. Radiology, 1988; 168: 817–21.

Rescorla FJ, Grosfeld JL, West KJ, Vane DW. Changing patterns of treatment and survival in neonates with meconium ileus. Arch Surg, 1989; 124: 837–40.

Ros PR, Olmsted WW, Moser RP, Dachman AH, Hjermstad BH, Sobin LH. Mesenteric and omental cysts: histologic classification with imaging correlation. Radiology, 1987; 164: 327–32.

Rosenfield NS, et al. Hirschsprung disease: accuracy of the barium enema examination. Radiology, 1984; 150: 393–400.

Sato Y, et al. Congenital anorectal anomalies: MR Imaging. Radiology, 1988; 168: 157–62.

Snyder WH, Chaffin L. Malrotation of the intestine. In: Mustard WT, Ravitch MM, Snyder WH, Welch KJ, Benson CD, eds. Pediatric Surgery. Chicago: Year Book Publishers, 1969: 808–17.

Westra SJ, de Groot CJ, Smits NJ, Staalman CR. Hypertrophic pyloric stenosis: Use of the pyloric volume measurement in early US diagnosis. Radiology, 1989; 172: 615–19.

Working group of the research unit, Royal College of Physicians. Guidelines for the management of acute urinary tract infection in childhood. J R Coll Phys, 1991; 25: 36–42.