Note: This copy is for your personal non-commercial use only.

To order presentation-ready
copies for distribution to your colleagues or clients, contact us at www.rsna.org/rsnarights.

ORIGINAL RESEARCH
Biliary Atresia: US Diagnosis1

# PEDIATRIC IMAGING
Terry M. Humphrey, MHSC, DMU, DCR
Purpose: To evaluate prospectively the sensitivity of ultrasonogra-
Mark D. Stringer, MS, FRCS, FRCPCH
phy (US) in the diagnosis of biliary atresia (BA), with
surgery as the reference standard.

Materials and After institutional ethical approval and with informed pa-
Methods: rental consent, 90 consecutive fasting infants with conju-
gated hyperbilirubinemia underwent detailed US studies
performed by a single operator with a 7.5-MHz curvilinear
transducer and a 13.5-MHz linear-array transducer. The
following features were prospectively recorded: gallblad-
der morphology, triangular cord sign, presence of a com-
mon bile duct, liver size and echotexture, splenic appear-
ance, and vascular anatomy. The operator was blinded to
results of other investigations. Sensitivity, specificity, and
positive and negative predictive values were calculated for
each US variable. BA and non-BA groups were compared
by means of the Fisher exact test for categorical variables
and an unpaired t test for continuous variables.

Results: Thirty infants (13 male, 17 female) had surgically con-

firmed BA, and 60 (35 male, 25 female) had other docu-
mented causes of neonatal jaundice; the mean ages at US
assessment were 48.5 and 52.4 days, respectively (P !
.5). Eight US features showed a significant difference be-
tween BA and non-BA groups (P " .001, Fisher exact
test). The features with the greatest individual sensitivity
and specificity, respectively, in the diagnosis of BA were
triangular cord sign (73% and 100%), abnormal gallblad-
der wall (91% and 95%) and shape (70% and 100%), and
an absent common bile duct (93% and 92%). The hepatic
artery diameter was significantly larger in infants with BA
than in those without BA (mean # standard deviation, 2.2
mm # 0.59 vs 1.6 mm # 0.40, P " .001), but portal vein
diameters were not significantly different. By means of all
these US features, 88 of 90 infants were correctly classified
as having or not having BA, for an overall accuracy of 98%.

Conclusion: BA can be distinguished with US from other causes of

conjugated hyperbilirubinemia in 98% of infants if multiple
US features are carefully evaluated.
1
From the Department of Radiology (T.M.H.) and Chil-
dren’s Liver & GI Unit (M.D.S.), St James’s University ! RSNA, 2007
Hospital, Beckett Street, Leeds LS9 7TF, England. Re-
ceived June 17, 2006; revision requested August 21;
revision received September 27; accepted November 1;
final version accepted January 19, 2007. Address corre-
spondence to T.M.H. (e-mail: teresa.humphrey@leedsth
.nhs.uk).

" RSNA, 2007

Radiology: Volume 244: Number 3—September 2007 845

PEDIATRIC IMAGING: US Diagnosis of Biliary Atresia
T
he evaluation of infants suspected
of having biliary atresia (BA) is
challenging because no single pre-
operative investigation enables the diag-
nosis to be made with certainty. Liver
biochemistry assessment, biliary radio-
nuclide excretion scanning, magnetic
resonance cholangiography, endoscopic
retrograde cholangiography, percutane-
ous needle liver biopsy, and laparos-
copy can all be helpful, but their results
are not individually diagnostic. The di-
agnosis of BA is usually suggested after
results of several of these investigations
(typically including liver biopsy) have
been reviewed and the diagnosis is con-
firmed at laparotomy.
Numerous ultrasonographic (US)
features have been described as useful
pointers to the diagnosis of BA. Abnor-
malities in the shape and wall of the
gallbladder have, for experienced oper-
ators, yielded sensitivities and specifici-
ties of more than 90% in the diagnosis
of BA (1). The triangular cord (TC)
sign, a focal area of increased echoge-
nicity anterior to the bifurcation of the
portal vein representing the fibrotic
remnant of the extrahepatic biliary tree
in BA, has been considered an impor-
tant diagnostic feature by some (2,3).
Other groups (4) have suggested that a
combination of three gallbladder fea-
tures—namely, length less than 19 mm,
an irregular wall, and an indistinct mu-
cosal lining (the so-called gallbladder
ghost triad)—is diagnostic.
In addition to these US features, BA
may be associated with other congenital
structural anomalies that are detectable
with US (5). For example, in the BA
splenic malformation syndrome, as de-
fined by Davenport et al (6), BA can be
Advances in Knowledge
# An accuracy of 98% can be
achieved in the US diagnosis of
biliary atresia (BA).
# We have shown for the first time,
to our knowledge, that the caliber
of the hepatic artery is signifi-
cantly increased in infants with
BA compared with the caliber in
infants with other causes of conju-
gated jaundice.
846
associated with polysplenia, situs inver-
sus, an interrupted inferior vena cava,
and other cardiovascular anomalies.
Furthermore, infants with BA typically
have a degree of hepatic fibrosis or cir-
rhosis, and the majority have some evi-
dence of portal hypertension as early as
8 weeks of age (7); these pathologic
changes may sometimes be evident at
US. Finally, hypertrophy of the hepatic
artery has been reported in infants with
BA (8). The purpose of our study was to
evaluate prospectively the sensitivity of
US in the diagnosis of BA, with surgery
as the reference standard.
Materials and Methods
Study Group
Between February 2002 and November
2005, consecutive infants with conju-
gated hyperbilirubinemia of unknown
cause were prospectively evaluated at
our supraregional unit, which is one of
three pediatric liver units in the United
Kingdom. The study was approved by
our Institutional Clinical Research Eth-
ics Committee, and informed parental
consent was obtained. Infants with con-
jugated hyperbilirubinemia of unknown
cause undergo a series of investigations
to establish the diagnosis. In those sus-
pected of having BA, the investigations
include a detailed biochemical and met-
abolic workup, an abdominal US scan, a
radioisotope biliary excretion scan, and
a percutaneous needle biopsy of the
liver. In all cases of BA, the diagnosis
was confirmed at laparotomy, the refer-
ence standard.
US Imaging
All infants underwent a detailed abdom-
inal US examination performed by a sin-
gle operator (T.M.H., with 10 years of
experience with pediatric US) who was
blinded to the results of other investiga-
tions, such as hepatobiliary scintigra-
Implication for Patient Care
# Accurate diagnosis of biliary atre-
sia with US may reduce the need
for invasive procedures such as
percutaneous needle liver biopsy.
Humphrey and Stringer
phy, liver function testing, and histo-
logic findings after liver biopsy. The
equipment used was a US system (Sono-
line Elegra; Siemens, Washington, DC)
with a 7.5-MHz multifrequency curvi-
linear transducer and a 13.5-MHz mul-
tifrequency linear-array transducer. In-
fants were not fed for 4 hours before
their examinations because it was thought
that the gallbladder could be assessed
more easily when it was not contracted.
None of the infants were sedated.
The gallbladder was interrogated
with the 13.5-MHz linear-array trans-
ducer to assess its shape and the regu-
larity of its wall. An irregular wall and a
distortion of the gallbladder’s usual
shape were considered abnormal (Fig
1). In some instances, irregularity of the
gallbladder wall may be subtle (Fig 1c),
and it is therefore important that the
gallbladder is examined with a high-
frequency probe. The gallbladder length
was measured. The high-frequency
transducer was also used to look for the
TC sign; when present, this sign can be
found by scanning in a transverse
oblique plane and identifying a focal
area of increased echogenicity anterior
to the bifurcation of the portal vein (Fig
2). If the common bile duct (CBD) was
visible, its caliber was recorded. Color
Doppler flow imaging was used to help
identify the CBD. Liver size and echo-
texture were assessed; the liver was
considered to be enlarged if the right
lobe extended below the right kidney
Published online
10.1148/radiol.2443061051
Radiology 2007; 244:845– 851
Abbreviations:
BA $ biliary atresia
CBD $ common bile duct
TC $ triangular cord
Author contributions:
Guarantors of integrity of entire study, T.M.H., M.D.S.;
study concepts/study design or data acquisition or data
analysis/interpretation, T.M.H., M.D.S.; manuscript draft-
ing or manuscript revision for important intellectual con-
tent, T.M.H., M.D.S.; manuscript final version approval,
T.M.H., M.D.S.; literature research, T.M.H., M.D.S.; clini-
cal studies, T.M.H., M.D.S.; statistical analysis, T.M.H.,
M.D.S.; and manuscript editing, T.M.H., M.D.S.
Authors stated no financial relationship to disclose.
Radiology: Volume 244: Number 3—September 2007
PEDIATRIC IMAGING: US Diagnosis of Biliary Atresia Humphrey and Stringer

Figure 1
(9). Features of portal hypertension,
such as ascites, reversed blood flow in Figure 1: Longitudinal sonograms obtained with
the portal vein, portal vein caliber, 13.5-MHz transducer. (a) Normal gallbladder (arrow)
splenic varices, and splenomegaly, were in 5-week-old infant without BA has normal shape
recorded, and the caliber of the hepatic and regular wall. Calipers and dotted line indicate
artery was noted. Splenomegaly was de- gallbladder length measurement. Bowel (arrowhead)
fined as when the spleen had a length of is seen posterior to the gallbladder. (b) Small, abnor-
more than 6 cm (10). Any features sug- mal gallbladder (arrow) in 8-week-old infant with BA
gestive of the BA splenic malformation has irregular wall and abnormal shape. (c) Abnormal
syndrome (6), such as polysplenia and gallbladder (arrow) in 10-week-old infant with BA.
an interrupted retrohepatic vena cava, Note the more subtle irregularity of the wall (arrow-
were also noted. heads) of this 21-mm-long gallbladder.

Statistical Analysis
Sensitivity, specificity, and positive and
negative predictive values were calcu-
lated for each US variable. The Fisher
exact test was used to determine
whether the frequency of each variable
differed significantly between the BA
and non-BA groups. An unpaired t test
was used to compare normally distrib-
uted continuous variables, such as por-
tal vein and hepatic artery diameters.
The Bonferroni method was used to ad-
just for multiple comparisons. Statisti-
cally significant differences were de-
fined as those with P values less than
.05, and analyses were performed with
statistical software (SPSS, version 12;
SPSS, Chicago, Ill).

Results
A total of 90 infants were evaluated—30
with BA and 60 with other causes of
conjugated hyperbilirubinemia (Fig 3). Figure 2
Clinical characteristics were compared
for the two patient groups at the time of
US scanning; there was no significant
difference (P ! .5) in age between the
groups (Table 1).

BA Group
In all infants with BA (n $ 30), the
diagnosis was confirmed at surgery and
at subsequent histologic examination;
29 infants had type 3 BA (no residual
patent extrahepatic bile ducts), and one
infant had type 1 BA (CBD atresia).
Four had features of the BA splenic mal-
formation syndrome (6) (see below).
Preoperative evaluation included a ra-
Figure 2: TC sign. (a) Transverse sonogram shows TC sign (arrow) in 7-week-old infant with BA. Area of
dioisotope (technetium 99m [99mTc]
increased echogenicity (as indicated by calipers) is seen anterior to bifurcation of portal vein (arrowheads). (b) Sag-
HIDA) biliary excretion scan in 29 in-
ittal sonogram in 10-week-old infant with BA shows TC sign (arrow) superior to portal vein (arrowhead).
fants, which showed no excretion in all

Radiology: Volume 244: Number 3—September 2007 847

PEDIATRIC IMAGING: US Diagnosis of Biliary Atresia Humphrey and Stringer

Figure 3
29, and percutaneous needle liver bi-
opsy in 21 infants. One patient in this
group died of congenital heart disease.
All others were still being followed up at
a median age of 14 months (range, 2– 46
months); 18 were free of jaundice and
were well after a Kasai portoenteros-
tomy, and 11 had undergone successful
transplantation.

Non-BA Group
A wide variety of diagnoses accounted
for conjugated hyperbilirubinemia in
this group of infants (n $ 60) (Table 2).
A radioisotope (99mTc HIDA) biliary ex-
cretion scan was performed in 53 in-
fants (no excretion was found in 23 in-
fants), and a percutaneous needle liver
biopsy was performed in 27. Five in-
fants in this group eventually required
Figure 3: Flow diagram shows US features and outcome in 90 consecutive infants with conjugated hyper-
laparotomy, surgical cholangiography,
bilirubinemia. BASM $ BA splenic malformation syndrome, GB $ gallbladder.
and wedge liver biopsy to exclude BA
definitively. Five infants underwent fol-
low-up elsewhere, and the other 55
were followed up locally for a median of
Table 1
16 months (range, 3– 44 months). In 44
Patient Characteristics at Time of US Assessment of 55 infants, jaundice completely re-
solved (as was confirmed with biochem-
Characteristic BA Group (n $ 30) Non-BA Group (n $ 60)
ical testing). Six infants had persistent
Age (d) 48.5 # 25.6 (18–143) 52.4 # 27 (7–126) jaundice (two each had progressive fa-
Male-to-female ratio 13:17 35:25 milial intrahepatic cholestasis, Alagille
Median gestation (wk) 40 (30–42) 37 (26–42) syndrome, and cystic fibrosis), four in-
Weight (kg) 3.94 # 0.77 (2.2–5.6) 3.11 # 0.92 (1.2–5.5) fants with complex multisystem disor-
Total bilirubin (%mol/L) 160 # 45 (107–303) 133 # 58 (27–386) ders died, and one infant with neonatal
Note.—Unless otherwise specified, data are means # standard deviations, with ranges in parentheses. sclerosing cholangitis underwent suc-
cessful liver transplantation. This group
of patients had a wide spectrum of dis-
orders, and detailed follow-up con-
Table 2 firmed that none had BA.

Causes of Conjugated Hyperbilirubinemia in 60 Infants without BA US Findings

Diagnosis No. of Infants There was a significant difference in the
frequency of each US feature between
Neonatal hepatitis syndrome 22
the BA and non-BA groups (Table 3). In
Prematurity with sepsis and parenteral nutrition 15
two comparisons, detection of the gall-
&1-Antitrypsin deficiency 4
bladder and polysplenia, differences
Alagille syndrome 4
Cystic fibrosis* 3
just failed to reach statistical signifi-
Cytomegalovirus hepatitis 2 cance after adjustment for multiple
Hypopituitarism 2 comparisons. When all US features
Neonatal sclerosing cholangitis 2 were used, 88 of 90 infants were cor-
Progressive familial intrahepatic cholestasis 2 rectly classified into the BA or non-BA
Nonsyndromic biliary hypoplasia 2 group, for an overall accuracy of 98%
Inspissated bile syndrome (with gallbladder agenesis) 1 (Table 4). Two infants were incorrectly
Choledochal cyst 1 classified as having BA; one infant with
histologic features of biliary hypoplasia
* Cystic fibrosis with or without prematurity, sepsis, and parenteral nutrition.
had a sonographically abnormal gall-

848 Radiology: Volume 244: Number 3—September 2007

PEDIATRIC IMAGING: US Diagnosis of Biliary Atresia Humphrey and Stringer

Table 3
bladder wall and no visible CBD, and
one premature, growth-retarded infant US Findings in 90 Infants with Conjugated Hyperbilirubinemia
with neonatal hepatitis syndrome had
Infants in BA Infants in Non-BA
no detectable gallbladder or CBD at US.
Finding* Group (n $ 30) Group (n $ 60) Total (n $ 90)
The mean gallbladder length was
significantly smaller in infants with BA Gallbladder†
(mean # standard deviation, 18.8 Detected 23 58 81
mm # 7.9 [n $ 23]) than in those with- Not seen 7 2 9
out BA (25.9 mm # 6.9 [n $ 58], P " Gallbladder length
.001) (Fig 4). The mean hepatic artery "19 mm 14 9 23
diameter was significantly larger in the !19 mm 9 49 58
Gallbladder shape
BA group than in the non-BA group (2.2
Normal 7 58 65
mm # 0.59 vs 1.6 mm # 0.40, respec-
Abnormal 16 0 16
tively; P " .001) (Fig 5). In contrast,
Gallbladder wall regularity
portal vein diameters were not signifi-
Normal 2 55 57
cantly different between the two groups
Abnormal 21 3 24
(BA group, 4.1 mm # 0.58 [n $ 29]; TC sign
non-BA group, 4.1 mm # 0.59 [n $ 60]; Present 22 0 22
P ! .5), and there were no infants with Absent 8 60 68
reversed blood flow in the portal vein, CBD
an occluded portal vein, or splenic vari- Detected 2 55 57
ces. Only one infant, a boy with BA, had Not seen 28 5 33
an abnormal, coarse liver echotexture. Liver size
In two infants with BA, the CBD was Normal 17 56 73
considered to be visible at US. One of these Enlarged 13 4 17
infants was found at surgery to have a Spleen size
patent distal CBD in continuity with the gall- Normal 19 59 78
bladder but no patent proximal extra- Abnormal 11 1 12
hepatic bile ducts (type 3 BA). Of the four Polysplenia‡
infants with the BA splenic malformation Present 3 0 3
syndrome, three had polysplenia and an in- Absent 27 60 87
terrupted retrohepatic vena cava at US. Inferior vena cava§
Interrupted 5 0 5
The fourth had an absent portal vein, and at
Normal 25 60 85
laparotomy he was found to have a congen-
ital portocaval fistula and a segmented Note.—P " .001 for difference between groups for each finding, except where noted. P values were calculated with the Fisher
spleen. One other infant had an interrupted exact test.
vena cava and BA but no other anomalies at * Findings were evaluated in all 90 infants, except gallbladder length, shape, and wall regularity, which were evaluated in 81
infants because the gallbladder was not visualized in nine infants.
laparotomy. †
P $ .006.
Three infants in the non-BA group had ‡
P $ .035.
cystic fibrosis; two of these infants had a §
P $ .003.
small gallbladder ("19 mm long) and two
had an irregular, thickened gallbladder wall
(Fig 6). In two infants in the non-BA group, limited number of specific US features. used the gallbladder ghost triad (gall-
the gallbladder could not be identified with Using a 7.5-MHz probe in jaundiced in- bladder length " 19 mm, lack of a
US. One infant with no other US evidence fants, Farrant et al (1) determined that smooth mucosal lining with an indistinct
of BA had inspissated bile plug syndrome an absent gallbladder or one with an wall, and irregular or lobular contour)
with agenesis of the gallbladder that was irregular wall or abnormal shape had a in the diagnosis of BA. US examinations
subsequently confirmed at laparotomy. The sensitivity, specificity, and accuracy of were performed with a 5–12-MHz lin-
other also had an absent CBD and was one 90%, 92.4%, and 91.9%, respectively, ear-array transducer in fasting infants.
of the two infants with incorrectly classified in the diagnosis of BA. Diagnostic accu- The gallbladder ghost triad was present
disease described above. racy improved to 95.6% in a subsequent in 30 of 31 infants with BA, with no
study (11) of 158 infants (35 of whom false-positive findings.
had BA) when a 13-MHz linear-array Choi et al (3) evaluated the TC sign
Discussion probe was used. Patients were not fast- in 41 infants and concluded that it was a
Several groups have previously at- ing, and the presence of a TC sign was sensitive marker of BA. BA was con-
tempted to diagnose BA by means of a not recorded. Tan Kendrick et al (4) firmed in 12 of 13 jaundiced infants with

Radiology: Volume 244: Number 3—September 2007 849

PEDIATRIC IMAGING: US Diagnosis of Biliary Atresia Humphrey and Stringer

a positive sign, but one of the 28 infants fants, 25 of whom had BA, and Tan evitable when the presence or absence
with an absent TC sign was mistakenly Kendrick et al (4) identified an identical of the CBD is used as a criterion for
considered to have neonatal hepatitis sensitivity in 31 infants with BA. diagnosing BA, because a patent distal
and was later found to have BA. Other US features of BA have been CBD in continuity with the gallbladder
Whereas some authors (12,13) have evaluated. Azuma et al (16) recorded can be found in 10%–20% of affected in-
promoted the TC sign as a sensitive and the presence or absence of the CBD in fants with type 3 BA (17). Ho et al (8)
specific US marker of BA, others 30 infants, as seen with a 5- or 7-MHz noted that the hepatic artery in children
(4,14,15) have been more cautious. probe. In the 23 infants with confirmed with BA appeared hypertrophied, and
Thus, Park et al (14) found that the TC BA, nonvisualization of the CBD had a others (4) have also commented on the
sign had a sensitivity of 84% and a spec- sensitivity of 83% but a specificity of prominence of the hepatic artery in in-
ificity of 98% in their series of 79 in- only 71%. False-negative results are in- fants with BA.
After a detailed review of previous
Table 4 reports, we evaluated these and other
US features prospectively in a consecu-
Specific US Features as Predictors of BA tive series of infants suspected of having
Sensitivity Specificity Positive Predictive Negative Predictive BA to determine the accuracy of ab-
US Feature (%)* (%)* Value (%) Value (%) dominal US in diagnosis. Although the
frequency of each US feature was signif-
Absent gallbladder 23 (11, 43) 97 (87, 99) 78 72
icantly different between the BA and
Gallbladder " 19 mm 61 (39, 80) 84 (72, 92) 61 84
Abnormal gallbladder shape 70 (47, 86) 100 (92, 100) 100 89
non-BA groups, the variable sensitivity
Irregular gallbladder wall 91 (70, 98) 95 (85, 99) 88 96 and specificity of individual features
TC sign 73 (54, 87) 100 (93, 100) 100 88 noted in the literature were confirmed.
Absent CBD 93 (76, 99) 92 (81, 97) 85 96 Using the criteria proposed by Farrant
Enlarged liver 43 (26, 62) 93 (83, 98) 76 77 et al (11) (irregular gallbladder wall
Abnormal spleen size 37 (21, 56) 98 (89, 100) 92 76 and shape), we obtained an overall
Polysplenia 10 (3, 28) 100 (93, 100) 100 69 diagnostic accuracy of 95%. We agree
Interrupted inferior vena cava 17 (6, 35) 100 (93, 100) 100 71 with Tan Kendrick et al (4) that the
TC sign is a highly specific but less
* Data in parentheses are 95% confidence intervals.
sensitive marker of BA, but we could
not confirm the accuracy of their gall-
bladder ghost triad. Indeed, gallbladder
length was 19 mm or greater in nine of
Figure 4 Figure 5 23 infants (39%) with BA in our series.
In our study, visualization of the
CBD was associated with a sensitivity

Figure 6

Figure 4: Graph shows gallbladder length in Figure 5: Graph shows hepatic artery (HA) Figure 6: Longitudinal sonogram of abnormal
BA and non-BA groups. Mean gallbladder length diameter in BA and non-BA groups. Mean hepatic irregular gallbladder wall (arrow) in 6-week-old
(horizontal lines) was significantly smaller in in- artery diameter (horizontal lines) was significantly infant with cystic fibrosis and conjugated hyperbil-
fants with BA (P " .001). larger in patients with BA (P " .001). irubinemia. This infant did not have BA.

850 Radiology: Volume 244: Number 3—September 2007

PEDIATRIC IMAGING: US Diagnosis of Biliary Atresia
and specificity of more than 90%, but
the reliability of this feature is limited
for several reasons. The normal caliber
of the CBD in an infant is 0.74 mm # 0.4
(18), which makes detection technically
challenging. Moreover, confirming that
a feature is absent is often more de-
manding than ascertaining its presence.
Finally, a patent distal CBD may be
present in some infants with BA and
proximal extrahepatic bile duct occlusion.
We have shown for the first time, to
our knowledge, that the caliber of the he-
patic artery is significantly larger in in-
fants with BA, but the degree of overlap
with the non-BA group and the potential
likelihood of intraoperator variability mean
that individual measurements are not
particularly discriminative. However, a
prominent hepatic artery can be re-
garded as a supporting feature for the
diagnosis of BA. Why the extrahepatic
artery is larger in BA is unknown, but
the increased size may be secondary to
increased arterial resistance from hy-
pertrophy of the media of intrahepatic
arteries, which has been observed spe-
cifically in BA (19). Portal vein diame-
ters were not significantly different be-
tween infants with BA and those with
other causes of conjugated hyperbiliru-
binemia, indicating that the narrow por-
tal vein commonly seen in infants with
BA at liver transplantation (20) is ac-
quired rather than congenital.
An enlarged liver and splenomegaly
were both found to be individually sta-
tistically significant features in infants
with BA. However, these are relatively
nonspecific findings, and a diagnosis of
BA could not be made solely on their
basis. As with hepatic artery caliber,
the presence of these “soft markers”
can support the diagnosis of BA.
A possible limitation of our study
was that all US examinations were per-
formed by a single operator. Operator
experience is a critical component in the
detection and interpretation of US
findings, and similar results may be
achieved only if infants with conjugated
hyperbilirubinemia are assessed by ded-
icated experienced personnel in a refer-
ral center that deals with similarly large
numbers of infants with conjugated hy-
perbilirubinemia. We have not yet at-
Radiology: Volume 244: Number 3—September 2007
tempted to determine the interobserver
variability of our findings in the US eval-
uation of BA.
The key conclusion of our study is
that with use of a combination of all US
features, 88 of 90 infants were correctly
classified in the BA or non-BA group, for
an overall accuracy of 98%. Impor-
tantly, no infant with BA was incorrectly
classified in the non-BA group. Our
study suggests that in most patients BA
can be accurately diagnosed by means
of US, avoiding the need for potentially
harmful and expensive additional diag-
nostic procedures in this group of pa-
tients. If a TC sign is clearly visualized,
the patient can proceed to minilapa-
rotomy and surgical cholangiography.
The likelihood of seeing a TC sign in a
jaundiced infant who does not have BA
is extremely small; in a total of seven
published series (3,4,12–15,21) com-
prising 416 infants with conjugated
jaundice, only three (0.7%) had a false-
positive TC sign. In four of these series,
no false-positive TC sign was recorded,
which concurs with our experience to date.
An abnormal gallbladder is also an
important positive pointer to BA, but
this is less reliable as an isolated finding
in the absence of other US features of
BA, as we also observed it in infants
with jaundice associated with cystic fi-
brosis. Infants with no US evidence of
BA will still require further investigation
to establish the cause of their conju-
gated hyperbilirubinemia. An accurate
diagnosis of BA is possible if multiple US
features are carefully analyzed by an ex-
perienced operator.
Acknowledgment: We thank Vicki Allgar, PhD,
for her assistance with the statistical analysis.
References
1. Farrant P, Meire HB, Mieli-Vergani G. Ultra-
sound features of the gall bladder in infants
presenting with conjugated hyperbilirubinae-
mia. Br J Radiol 2000;73:1154 –1158.
2. Choi SO, Park WH, Lee HJ, Woo SK. ‘Trian-
gular cord’: a sonographic finding applicable in
the diagnosis of biliary atresia. J Pediatr Surg
1996;31:363–366.
3. Choi SO, Park WH, Lee HJ. Ultrasonographic
“triangular cord”: the most definitive finding
for non-invasive diagnosis of extrahepatic bili-
ary atresia. Eur J Pediatr Surg 1998;8:12–16.
4. Tan Kendrick AP, Phua KB, Ooi BC, Tan CE.
Biliary atresia: making the diagnosis by the
Humphrey and Stringer
gallbladder ghost triad. Pediatr Radiol 2003;
33:311–315.
5. Tanano H, Hasegawa T, Kawahara H, Sasaki
T, Okada A. Biliary atresia associated with
congenital structural anomalies. J Pediatr
Surg 1999;34:1687–1690.
6. Davenport M, Savage M, Mowat AP, Howard
ER. The biliary atresia splenic malformation
syndrome. Surgery 1993;113:662– 668.
7. Howard ER. Biliary atresia: etiology, manage-
ment and complications. In: Howard ER,
Stringer MD, Colombani PM, eds. Surgery of
the liver, bile ducts and pancreas in children.
2nd ed. London, England: Arnold, 2002; 103–
132.
8. Ho CW, Shioda K, Shirasaki K, et al. The
pathogenesis of biliary atresia: a morphologi-
cal study of the hepatobiliary system and the
hepatic artery. J Pediatr Gastroenterol Nutr
1993;16:53– 60.
9. Siegel MJ. Liver and biliary tract. In: Siegel
MJ, ed. Pediatric sonography. New York, NY:
Raven, 1991; 115.
10. Rosenberg HK, Markowitz RI, Kolberg H,
Park C, Hubbard A, Bellah RD. Normal
splenic size in infants and children: sono-
graphic measurements. AJR Am J Roentgenol
1991;157:119 –121.
11. Farrant P, Meire HB, Mieli-Vergani G. Im-
proved diagnosis of extrahepatic biliary atre-
sia by high frequency ultrasound of the gall
bladder. Br J Radiol 2001;74:952–954.
12. Kotb MA, Kotb A, Sheba MF, et al. Evaluation
of the triangular cord sign in the diagnosis of
biliary atresia. Pediatrics 2001;108:416 – 420.
13. Visrutaratna P, Wongsawasdi L, Lerttumnon-
gtum P, et al. Triangular cord sign and ultra-
sound features of the gall bladder in infants
with biliary atresia. Australas Radiol 2003;47:
252–256.
14. Park WH, Choi SO, Lee HJ. The ultrasono-
graphic ‘triangular cord’ coupled with gall
bladder images in the diagnostic prediction of
biliary atresia from infantile intrahepatic cho-
lestasis. J Pediatr Surg 1999;34:1706 –1710.
15. Lee HJ, Lee SM, Park WH, Choi SO. Objec-
tive criteria of triangular cord sign in biliary
atresia on US scans. Radiology 2003;229:395–
400.
16. Azuma T, Nakamura T, Nakahira M, et al.
Preoperative ultrasonographic diagnosis of
biliary atresia: with reference to the presence
or absence of the extrahepatic bile duct. Pedi-
atr Surg Int 2003;19:475– 477.
17. Kobayashi H, Stringer MD. Biliary atresia. Se-
min Neonatol 2003;8:383–391.
18. Hernanz-Schulman M, Ambrosino MM, Free-
man PC, Quinn CB. Common bile duct in
children: sonographic dimensions. Radiology
1995;195:193–195.
19. dos Santos JL, da Silveira TR, da Silva VD,
Cerski CT, Wagner MB. Medial thickening of
hepatic artery branches in biliary atresia: a
morphometric study. J Pediatr Surg 2005;40:
637– 642.
20. Ravindra KV, Guthrie JA, Woodley H, et al.
Preoperative vascular imaging in pediatric
liver transplantation. J Pediatr Surg 2005;40:
643– 647.
21. Kanegawa K, Akasaka Y, Kitamura E, et al.
Sonographic diagnosis of biliary atresia in pedi-
atric patients using the “triangular cord” sign
versus gallbladder length and contraction. AJR
Am J Roentgenol 2003;181:1387–1390.
851