A proper greyscale B-mode image of the respective vessel is essential. For PW-DS measurements, switching to and fro the 2D image to CDS (duplex mode = simultaneous greyscale and CDS) is helpful in improving angle correction and for exact positioning of the gate inside the vessel. At present, the triplex mode (simultaneous greyscale imaging, CDS and PW-DS) is mostly too slow to be practicable or of additional help. In most instances, it degrades the quality of the Doppler spectrum. Technical evolution may in future overcome this restriction. In situations difficult to survey, e.g. with multiple vessels in a close location, artefacts due to heart beating or vessels with slow flow close to vessels with fast flow, using PW-DS without switching to CDS is sometimes preferable. On CDS and particularly power Doppler, vessels tend to become blurred (Fig. 6.3) so that these images cannot be used for the measurement of the vessel diameters, e.g. for F _Vol calculations or measurement of stenosis diameter. As mentioned previously, vessel diameters should only be measured in the greyscale US.

Possible anatomic variations such as an accessory HA should be ruled out. A dedicated effort to visualise the coeliac trunk, the splenoportal confluence, the superior mesenteric vein and artery as well as the inferior vena cava (IVC) should be undertaken. The investigation of the main splenic vein is mandatory in all conditions with suspicion of flow alterations. The entire vascular anatomy and the respective flow patterns of all supplying liver vessels should always be visualised and analysed.

The liver anatomy can be described and subdivided in two different manners: morphological (classical anatomic description) or functional (angiographic and surgical approach). The classical morphological anatomy is based on the external appearance of the liver and does not take the internal vessel and biliary duct branching into account, which is obviously essential for a surgical procedure such as hemihepatectomy (Fig. 6.14a, b).

The anatomic division of the liver into two lobes is along the falciform ligament, which runs along the diaphragmatic liver surface. It divides the anatomic left lobe (segment II and III) and the anatomic right lobe (segments I, IV–VIII).

The anatomic approach described differs enormously from the functional classification according to the classification of Couinaud (Fig. 6.15), where only the segments V–VIII belong to the right lobe. The left lobe is fed by the left PV and includes the segments I–IV. In the Couinaud classification, the liver is divided into eight functionally independent segments, each with an own vascular inflow and outflow as well as an own biliary drainage. Each segment can be resected without damaging the remaining segments. The resection lines parallel the course of the liver vein (LV), preserving the centrally located branches of the PV, HA and bile ducts. The segmental classification according to Couinaud therefore is more useful for surgical procedures, which are orientated on the vascular supply and therefore preferred in the clinical setting. Therefore, DS descriptions should preferably be based on this functional classification. The main PV stem branching into a right and left intrahepatic branch divides the liver into lower and upper segments; thereafter, the main right and left PV stems branch superiorly and inferiorly into the centre of each segment (Fig. 6.15). The middle LV divides the functional right from the left liver lobe. This imaginary plane is passing through the gallbladder anteriorly and the IVC posteriorly and is named the Cantlie line. The left LV is positioned between the segments II/III and IVa + b and divides the left lobe into a medial and a lateral part. The middle LV runs between segments IVa + b and V/VIII and divides the liver into the right and left functional lobe. The right LV runs between segments V/VIII and VI/VII and divides the right lobe into anterior and posterior segments. The classification of Couinaud from 1957 is predominately used in Europe and Asia, whereas the similar Bismuth’s classification is popular in the United States (Smithuis, Radiology Assistant). The differences between these two classifications are mainly that Couinaud divided the liver into eight and Bismuth into four sectors: right lateral (segments VI and VII), right paramedian (segments V and VIII), left paramedian (segments III and IV) and left lateral liver sector (segment II).

The PV is responsible for approximately three quarters of the liver perfusion and supplies about 50–60 % of the liver’s oxygen requirements under normal circumstances. Especially the hepatic sinusoids are supplied by portal venous branches, whereas the bile ducts and the connective tissue of the portal triads are predominantly perfused by the HA. Due to the dual blood supply, the native liver is quite resistant to ischaemia and collaterals may develop rapidly. In case of HA stenosis or thrombosis, the PV flow increases and usually preserves normal liver function in a native liver (Killi 1999; Hoyer 2005).

On greyscale US, the main PV stem can be seen in an upper abdominal oblique to transverse cross section in the midline. A right lateral intercostal access is also suitable, especially if the main PV is obscured by gas on the median approach (Fig. 6.16). For visualising the PV branching in the main right and left branch, a section tilted cranially and rotated anticlockwise for approximately 90° is useful. The PV is normally much larger than the HA, and there is no significant tapering of the PV diameter when entering the liver. There may be a dilatation of the left PV branch, the sinus venosus (synonyms: venous recess, recessus Rex or umbilico-portal confluence), a physiological remnant of the umbilical venous drainage, which can be of interest when a surgical mesenterico–portal Rex shunt is necessary in children with PV thrombosis and absent liver cirrhosis for bypassing the superior mesenteric vein to the umbilical portion of the left intrahepatic PV (Stenger et al. 1999) (Fig. 6.17a). Another indication for evaluating the venous recess is possible thrombosis after umbilical vein catheterisation in neonatal intensive care patients (Fig. 6.17b, c).

In the foetal circulation (Fig. 6.18), the oxygenated blood from the placenta comes through the umbilical vein, which gives a branch to the PV and continues afterwards as venous duct of Arantii to the ICV. As a remnant of the foetal circulation, the former umbilical vein degenerates under normal circumstances into the ligamentum teres hepatis, which runs in the free rim of the falciform ligament from the venous recess of the liver to the umbilicus (Table 6.1). The former venous duct (Arantii) can be visualised as venous ligament from the left PV branch to the IVC. It may be seen open in neonatology (Vade et al. 2001) and regain patency in cases of portal hypertension and portosystemic shunting (Fig. 6.19a–c). Both ligaments may persist with a patent lumen and some shunt flow until a maximum of four weeks postnatally (Fig. 6.19). If they stay open more longer or even show high (shunt) flow (Fig. 6.19d, e), an underlying pathological liver condition must be considered.

The size of the PV varies with age: in 0–5-year-old infants, the diameter is 3.5 ± 2 mm, in 6–12-year-old children 6.3 ± 2 mm and in >12-year-old children 7 ± 2.6 mm (Kao 1996; Vergesslich 1987). A diameter less than 3.5 mm is suggested to be predictive for PV stenosis after liver transplantation (Suzuki et al. 2008). However, no reliable age-dependent cut-off values exist for the PV diameter in the diagnosis of portal hypertension in paediatric patients. In approximately 20 % of the population, there are variations of PV branching pattern (Couinaud 1999; Fraser-Hill et al. 1990). Intrahepatically, the branches of the PV, the HA and the bile ducts neighbour side-by-side (“trias of Glisson” or periportal triangle). In preterm babies and neonates, the left liver lobe may be physiologically large.

CDS helps to identify the PV, possible anatomic anomalies (Fig. 6.20) and its patency and will give first information on flow direction (Coley 2004; Corness et al. 2006). Normally, the PV flow is hepatopetal (synonym: antegrade), exhibits an antegrade band-like flow spectrum and usually has slight respiratory modulations in children in contrast to adults (Fig. 6.21), who more commonly exhibit an unmodulated flow. Flow patterns may be altered by respiration and by atrial contractions (Fig. 6.22a, b). A retrograde trans-sinusoidal transmission of atrial pulsations (Görg et al. 2002) and transmission of pulsations of neighbouring vessels such as the IVC, splanchnic arteries and especially the HA may also alter flow patterns. A pathological pulsatile flow in the PV is predominately seen in severe right heart failure (Fig. 6.22a, b), arterio-portal fistulas and portocaval shunts (Görg et al. 2002; Westra et al. 1995).

DS spectral tracing enables a more detailed analysis after angle correction. Typically, flow velocities (and sometimes F _Vol) are measured in the main PV and vary not only with age, but also with splanchnic activity, increasing after meals. Therefore DS values are ideally acquired in a standardised fasted child. For practical reasons, the fasting state in infants and small children is usually confined to special indications on demand or research. Some authors (Berzigotti and Piscaglia 2011; 2012) recommend the crossing point of the PV with the HA as the ideal reproducible section to measure the PV flow (Fig. 6.23). However, particularly in children with liver cirrhosis and reduced PV flow as well as increased arterial flow to the liver, the PV flow is likely to show major concomitant pulsations from the HA in this section, thus altering the PV flow spectrum; furthermore, it may be difficult to obtain an angle <60° in this section. Patients after liver transplantation usually do not have this clear crossing point any longer.

The greyscale US access to the main HA is the same as for the PV. Additionally one should try to visualise the coeliac trunk in median axial and sagittal sections in order to pick up potential variations such as an additional HA originating from the superior mesenteric or the gastric artery. The HA is usually relatively small and sometimes is only clearly identifiable using CDS. Colour Doppler allows the differentiation of the HA from the common bile duct running just posterior-laterally or from an intrahepatic bile duct portion. Anatomic variations are common, the most common being the accessory HA (Killi 1999). An increased diameter and a tortuous appearance of the HA can be found in patients with hepatitis, liver cirrhosis and liver tumours. HA aneurysms are rare in children.

The HA supplies a rather low resistive organ thus exhibiting an antegrade (hepatopetal) systolic-diastolic flow. The typical DS measurements in the HA are V _s, V _ed and RI. Optimal angle correction is essential for meaningful DS measurements. Measurements should be taken at the hilar region and, if indicated, in the main branches of the right and left liver lobe.

At least two to three similar consecutive pulse phases should be traced (if possible), to ensure a correct measurement. Exact location of the sample volume within the centre of the vessel is important for correct measurements.

The RI is usually below 0.75. Very low RI < 0.45 may indicate a stenosis of the artery proximally (Fig. 6.24). Additional reasons for a low RI may be hyperperfusion due to inflammation or peripheral shunts (e.g. haemangioendothelioma, arteriovenous or arterio-portal shunts; Fig. 6.25a–b). A low RI may appear secondary to systemic circulation alterations in patients on ECMO or heart assist. The flow velocities and RI decrease from the hilus to the peripheral regions.

Flow velocities and the RI vary with age and, to a lesser extent, with the degree of splanchnic activation. In fasted patients velocities are lower, and after feeding the diastolic flow increases. After meals the velocities and RI in the HA increase by 30 % (Numata et al. 1999), though this phenomenon is less obvious than in the superior mesenteric artery.

F _Vol measurements in the HA are problematic due to the small diameter especially in young children with a normal size of the hepatic artery – they get more accurate in older children or when the HA is enlarged. As the diameter of the HA cannot be measured exactly, volume flow in the HA cannot be measured in infancy and childhood.

The valveless main three LVs can best be displayed in a transverse-oblique section tilted cranially from a right midline transducer position (Fig. 6.26b). As the situs and the access may vary, other approaches such as a more lateral or a more sagittal view may become necessary for assessing a vein, particularly if there are anatomic variations such as an accessory vein or an unusual branching pattern (Fig. 6.27a). Another approach is the measurement of the left LV in a midline sagittal section. The branching inside the liver is sharp-angled; in cases with fatty liver degeneration, it may become blunt-angled (Fig. 6.28a, b). In congestion of the liver, the LV branches are broadened and can be seen up to the periphery. In chronic congestion they may exhibit an echogenic delineation of the vessel borders (Fig. 6.29a), which is normally not present. In liver cirrhosis the LV may alter the straight into a bowed and irregular course and may even not be detectable any longer by greyscale US (Fig. 6.30a, b). In this case, CDS enables detection of the location of the LV and reveals information on the flow direction and directional modulation.

The caudate lobe has an own vein, which drains directly into the IVC. This is of special importance in the case of liver vein thrombosis in Budd-Chiari syndrome. The medium LV divides the angiographic left and right liver lobe. Occasionally there may be more than 3 LVs (Fig. 6.27a).

The detailed flow analysis is performed by spectral Doppler analysis, which should be measured close to but not at the inflow into the ICV. The flow in the LV becomes physiologically laminar when measured further away into the liver periphery. If possible, the measurements should be taken at the end of expiration or preferably during normal superficial respiration. Observation of flow modulation is essential: the LV exhibits a strongly undulating bidirectional triphasic flow pattern influenced by the cardiac and the respiratory cycle (Fig. 6.26b) resembling the atrial pressure similar to the jugular venous pulse. The liver veins have a mainly hepatofugal flow consisting of two large waves due to atrial filling in the atrial diastole and ventricular systole and a pronunciation by the tricuspid valve opening. The hepatopetal portion is due to right atrial contraction and tricuspid valve closure in atrial systole and ventricular diastole (Fig. 6.31). These directional changes can be appreciated by the changing colour coding on CDS and – more obvious – by the spectral flow trace (Fig. 6.32). The modulation is usually more prominent in the left and/or the middle than in the right LV. Mono-, bi- and triphasic flows are discriminated (Fig. 6.31). In severe tricuspid regurgitation, the flow will be undulating (Fig. 6.33a–c). In liver cirrhosis the LV flow is becoming monophasic (Zang et al. 2011) due to the decrease in tissue compliance. LV may drain directly into the right atrium without passing the IVC (Fig 6.34), especially after liver transplantation. No changes of the LV flow are depictable by DS with food intake (Chavhan et al. 2008).

There are only little data on normal values of the flow parameters in liver vessels, particularly in neonates, infants and young children. In older children, normal values are similar to those in adults. In general, PV and HA flow velocities and volumes are significantly lower in neonates than in elder children. From neonatology to adolescence, flow velocities gradually increase with age. RI is relatively age independent. This statement is not completely true for neonates – RI values are physiologically higher in neonates than later in life, before they relatively quickly adjust to adult values. This is partially due to the transition of foetal circulation to postnatal circulation. In the first days of life, the still patent ductus Arteriosus Botalli causes a higher RI.

Portal hypertension (PH), one of the main conditions that indicate liver DS (Table 6.2), can be diagnosed with high accuracy. The main causes and sonographic features are listed in Tables 6.4 and 6.5. The most common causes are pre-existing systemic or liver diseases (e.g. biliary atresia with consecutive biliary cirrhosis, cystic fibrosis, liver fibrosis in autosomal recessive polycystic kidney disease or nephronophthisis, liver cirrhosis in M. Wilson, α-1 antitrypsin deficiency, tyrosinaemia, M. Niemann-Pick) (Grunert et al. 1989; Tüney et al. 1998). Additionally, acquired PV thrombosis, e.g. after umbilical vein catheterisation, compression by tumours, thrombosis of the LV and congenital PV malformations (the Abernethy malformation complex), may cause portal hypertension.

Basically, greyscale US already delivers a range of findings that suggest the diagnosis liver cirrhosis:

The sensitivity for the US diagnosis of low-grade cirrhosis is rather low (up to70%); however, for advanced disease and PH, the sensitivity is much higher (90–95 %) (Nürnberg et al. 2006; Berzigotti and Piscaglia 2012). Findings suggestive for PH are as follows:

On CDS the depiction of PV colour signals may be poor due to the low flow velocities, which may even be missing, undulating or eventually reversed (Fig. 6.36a). In contrast, the HA usually exhibits prominent high-flow signals, due to the compensatory arterial hyperperfusion. Additionally collateral venous vessels, as listed above, may be observed. In the course of PV thrombosis, cavernous transformation of the occluded PV may be visualised as a network of irregular, ectatic, usually physiologically insufficient and torturous atypical low-flow veins in the liver hilus around the course of the PV (Fig. 6.39a–c).