Secondary Logo

Journal Logo

Original Articles: Hepatology and Nutrition

Impact of Virtual Imaging Procedures on Treatment Strategies in Children With Hepatic Vascular Malformations

Fuchs, Jörg*; Warmann, Steven W*; Sieverding, Ludger; Haber, Hans P; Schäfer, Jürgen§; Seitz, Guido*; Hofbeck, Michael; Bourquain, Holger; Peitgen, Heinz O

Author Information
Journal of Pediatric Gastroenterology and Nutrition: January 2010 - Volume 50 - Issue 1 - p 67-73
doi: 10.1097/MPG.0b013e3181a87187
  • Free


Hepatic surgery in infants and children is commonly performed for liver tumors, biliary disorders, abdominal trauma, and congenital malformations. Congenital malformations also include hepatic vascular malformations (HVM). For these, the treatment outcome depends on a correct diagnostic evaluation and treatment strategy; however, no treatment standards exist for many cases of HVM in children. Decisions concerning therapeutical approaches must be made individually for most of the patients (1–3). A major objective of the primary workup in children with HVM is to exactly clarify the topographic anatomy not only of all vascular structures but also of the segmental anatomy of the liver, the biliary tree, and the surrounding structures.

The curative approach in children with portal vein thrombosis includes 2 main objectives: the restoration of the intrahepatic portal venous flow to ultimately prevent hepatic encephalopathy (11–13) and the decrease of hypertension within the mesenteric venous system.

Both aspects can be achieved through the mesenterico-left portal vein bypass, also known as the Rex shunt procedure. The diagnostic challenge lies in the reduced diameter and blood flow of the postthrombotic portal branches and in the regular existence of portalvenous transformation. Therefore, it is not always doubtlessly assessable whether the evaluated vessels are suitable. Ultrasound may be insufficient because of the hypoplastic vessels with inconstant flow pattern. Also, it is not always possible to distinguish between portal vessels and collaterals in the porta hepatis using sonographic techniques. Furthermore, visualization of the topography may be necessary for planning the shunt.

The software assistants MeVis LiverAnalyzer (formerly HepaVision2) and MeVis LiverExploer (formerly InterventionPlanner, Mevis, Bremen, Germany) are used to process multislice CT or MRI (magnetic resonance imaging) data to create and visualize a patient's individual 3-dimensional model of anatomical structures as well as for virtual surgical planning (4). These tools were developed and have been established for analysis and planning of adult living donor liver transplantation, major hepatic surgery, or hepatic interventions in adults. Analyses include segmentation and volumetry of organs, planning of different resection proposals, and risk analyses for different safety margins or the different resection proposals (5). There are only a few experiences concerning the application of these analyzed software assistants in children (6); furthermore, only rare comparable data exist concerning pediatric surgical planning with similar tools (7–10). The aim of the present study was to investigate the impact of the MeVis LiverAnalyzer and MeVis LiverExplorer on the selection of treatment strategies of complex hepatic HVM in pediatric patients.


From June 2003 to June 2007, 12 children experiencing HVM were evaluated for treatment (for patients' data, see Table 1). Nine had a portal venous thrombosis and were assessed for the feasibility of a Rex shunt procedure. One child presented with a patent venous duct of Arantius, 1 had an intrahepatic portocaval shunt, and 1 experienced liver diffuse hemangioma. The mean age of the patients at investigation was 68 months (range 5.5–168 months).

Patients' data

Informed consent was obtained from the parents of each of the patients. The study protocol conforms to the ethical guidelines of the 1975 Declaration of Helsinki.

Computed Tomography

All of the children received a biphasic multislice helical CT scan (Volume Zoom Sensation 16, Siemens, Erlangen, Germany) with a 0.5-s gantry rotation. Anesthesia was used in children younger than 3 years of age. The scanning parameters were 120 kVp, 30 to 85 mA adapted to the body weight, 16 mm × 0.75 mm collimation, 12 mm table feed per rotation, slice thickness 0.75 mm, increment 0.5 mm. Contrast medium (2 mL/kg body weight, Imeron 400, Bracco-Byk Gulden, Konstanz, Germany) was applied through a 22- or 24-gauge needle via an antecubital vein using a power injector (Stellant, Medrad, Germany) at 1.2 to 3.0 mL/second. All of the patients were investigated using bolus tracking (threshold at 120–130 HU). The CDTIvol ranged from 4.06 to 10.02.

Three-dimensional Visualization and Virtual Surgical Procedures

The computer-aided evaluation of imaging data using the software assistants MeVis LiverAnalyzer and MeVis LiverExplorer has been previously described in detail (4). For visualization the liver was extracted from the CT data followed by a segmentation and structural analysis of the vascular systems including the portal vein, hepatic veins, and hepatic arteries. Liver segmentation was performed semiautomatically with the Live-Wire algorithm and compared with the segmentation as proposed by Couinaud. Patient individual vascular territories are calculated based on these analysis results. Finally, the 3-dimensional visualization was correlated with 2-dimensional CT slices, and intraoperative views were presented, such as cutting surfaces together with major transversal vessels. For analyses of vascular systems, related data were extracted and portal veins, hepatic veins, and arteries were visualized separately. Special investigations were carried out according to related clinical questions (eg, distance between mesenteric confluens and left portal vein in patients with portal vein thrombosis).


Virtual procedures using CT data excellently demonstrated the topography and spatial relations of various hepatic structures resulting in continuous, simultaneous, and individual imaging. Anatomical variants such as atypical arteries were clearly identified and could, consecutively, be taken into consideration for treatment planning. Virtual images could be rotated in every direction, allowing studies from various perspectives and all anatomical planes. In particular, the topography of the porta hepatis could be visualized continuously with an accuracy that in our view was superior to conventional diagnostic tools. The virtual imaging was confirmed in all of the patients who underwent surgery. The virtual procedures thus provided information that was often not obtained through conventional imaging techniques. There were no complications associated with the additional procedures.

Intrahepatic Portocaval Shunt

The patient with intrahepatic portocaval shunt was initially misjudged elsewhere as experiencing an Abernathy malformation. After referral to our institution, we suspected a portocaval shunt (Fig. 1A). Accordingly, the analysis clearly depicted the intrahepatic connection between right portal branch and right hepatic vein (Fig. 1B). Through this, the correct diagnosis was established. Further selective analyses of the vascular structures with the software assistants showed an additional aberrant (inferior) right hepatic vein and a regular distribution of the portal branches (Fig. 1C and D). At this stage, the selective angiography was indicated. In this investigation the virtual data were confirmed and the patient underwent interventional occlusion of the shunt with an Amplatzer device within the same session.

Intrahepatic portocaval shunt (patient no. 10). (A) Transverse ultrasound scan demonstrating the inferior vena cava (vci) and the shunt. (B) Example for anteroposterior view of virtual reconstruction: shunt from the right portal branch to the right hepatic vein. (C) Selective segmentation of the portal venous branches (P2: segment 2 portal vein, etc). (D) Transverse ultrasound scan demonstrating a restored flow within the right portal vein (blue vessel) after interventional shunt occlusion using an Amplatzer vascular plug (white arrow).

Portal Venous Thrombosis

We analyzed 9 children with portal venous thrombosis for feasibility of the Rex shunt procedure. Four children underwent umbilical catheterization as newborns. In 5, the reasons for thrombosis were not apparent and thus were most likely congenital. Four of them required instant treatment for life-threatening bleeding from the upper gastrointestinal tract; the 5 other patients were without clinical symptoms.

In 4 patients presenting without clinical symptoms but with gastroesophageal varices, a Rex shunt procedure was not feasible (Fig. 2). Further shunt operations were left optional if these patients became symptomatic. Imaging data showed no differences between symptomatic and asymptomatic patients.

Results of patients with extrahepatic portal vein thrombosis evaluated for the Rex shunt procedure. (A) Example of reconstruction showing portal venous transformation (patient no. 1); impossibility to identify the left intrahepatic portal branches. (B) Example of the visualization of the lienal and renal veins, allowing evaluation of Warren shunt in a patient not suitable for the Rex shunt procedure (patient no. 6). (C) Magnification of (B), measured distance between the 2 vessels in preparation of a Warren shunt. (D) Intraoperative view of the Warren shunt (patient no. 6): lienal vein (black arrow) and renal vein (white arrow).

Virtual imaging procedures made important contributions to analyzing children with portal venous thrombosis. The simultaneous delineation of the hepatic arteries, hepatic veins, and portal veins led to incomparable imaging results. Through rotation of the images, more precise analyses and intervention planning could be performed. These possibilities were not achievable using standard imaging procedures. Also, it was possible to analyze whether there was a communication between the left and the right portal vein, which is often not well assessable through ultrasound (Fig. 3).

Patient with portal venous thrombosis feasible for Rex shunt (patient no. 8). (A) Virtual reconstruction of CT data showing the segmental branches 3 and 4 of the left portal vein (p3 and p4), the superior mesenteric vein (v.m.s.), the inferior mesenteric vein (v.m.i.), the lienal vein (v.l.), the gastric vein (v.g.), and the distance between the recessus Rex and the venous confluens (34.3 mm). (B) Intraoperative view after dissection showing the venous confluens (right white arrow), the recessus Rex (left white arrow), and segment 4 of the left portal vein (blue arrow). (C) Intraoperative view with completed Rex shunt after reperfusion. (D) Postoperative Doppler ultrasound scan: Rex shunt (shunt) and left portal vein (li vp).

Patent Ductus Venosus

Patent ductus venosus (PDV) is a rare etiology in children, which may be associated with severe clinical symptoms including cardiac dysfunction, liver dysfunction, hypoxia, encephalopathy, or pulmonary hypertension (13). Patent ductus venosuses have varying sonographic appearances and their detection can lead to inadvertent diagnoses (14). Hemodynamic alterations result in decreased intrahepatic portal blood flow and hypoplasia of affected vessels. The existence and compliance of hypoplastic vessels to restore an intrahepatic portal venous flow contribute to the therapeutic decision whether the duct is to be partially or completely occluded. This can be performed interventionally by implantation of coils or vascular occluders, or surgically by vascular banding, depending on the clinical conditions of the patient as well as the size and flow patterns of the duct (15,16). Consecutively, the initial visualization of intrahepatic portal branches remains an important noninvasive step in patients with PDV.

In our case we received detailed analyses of the PDV topography, the intrahepatic portal branches, and the arterial structures of the liver hilum (Fig. 4A). The ultrasonographic evaluation could be interpreted correspondingly (Fig. 4B). The ductus had an immense diameter (10.8 mm) and blood flow (peak velocity 20.4 cm/second) and it was therefore decided to treat this patient surgically with a banding of the duct (Fig. 4C). Postoperative ultrasound revealed restored flow in intrahepatic portal venous branches with sufficient perfusion (Fig. 4D).

Example of a persistent ductus venosus Arantii (patient no. 9). (A) Three-dimensional reconstruction of the duct (inferior anteroposterior view) together with hepatic arteries. (B) Transverse ultrasound scan showing left intrahepatic portal branch without perfusion (white arrows, DV: ductus venosus, VCI: inferior vena cava). (C) Intraoperative view of the ductus venosus after banding (white arrow). (D) Transverse ultrasound scan showing restored perfusion within left intrahepatic portal branch (white arrows) 10 days after surgery (diameter measurement: 2.2 mm).

Liver Diffuse Hemangioma

One patient with disseminated hemangioendotheliomatosis of both liver lobes was evaluated for the correct treatment approach. However, in this patient virtual imaging procedures did not produce sufficient data mainly because of immensely increased blood velocities (300 cm/second within the common hepatic artery). Insufficient enhancement of the liver parenchyma and the lesions did not allow us to distinguish between affected and normal liver parenchyma and, therefore, did not permit the requested analysis. After the diagnostic workup the patient was not considered suitable for a surgical or an interventional approach. The child provisionally received a conservative treatment with corticoids and is under evaluation for liver transplantation.


General Aspects

The most common reasons for hepatic imaging in children include trauma, suspected mass, evaluation before liver transplantation, monitoring after liver transplantation, jaundice, or liver dysfunction (17). Connatal hepatic vascular malformations are rare and often need individual diagnostic workup, treatment planning, and therapy. The analyzed software assistants were developed for planning of major hepatic resections and living donor liver transplantations (4,5). The acquired data provide individual textbook-like imaging of the patient's anatomy. Experiences analyzing children with these software tools are limited (6). The assessment of pediatric tumors also revealed substantial additional information regarding the vascular supply and associated anomalies, which were without clinical relevance but important for the surgical procedures in these patients. The feasibility of the approach was without limitations, and the software assistants could be applied in the same manner as in adults. The thickness of reconstructed slices achieved using standard helical CT scan was far below 1 mm and allowed an exact visualization of anatomical structures in infants and children. It also produced a higher spatial resolution than standard MRI techniques. Thus, the age of the patients did not put a limit on the procedure.

The main advantages of the analyzed software assistants resulted from the simultaneous delineation of different vascular structures. The original data are usually acquired in different phases of the CT (arterial, venous, portal-venous) and are not compatible offhand. Also, the continuous presentation of the vascular structures is often not equally achievable in standard CT because of the limitations of the 2-dimensional imaging results. The virtual imaging techniques thus provided additional information and contributions to the team deciding on the treatment strategy. In small children the use of anesthesia was necessary. This seemed justified to obtain data of optimal quality. Comparable MRI-based techniques also require this approach. The use of anesthesia may be of concern in high-risk patients. However, clinical stability is precondition for any kind of valuable investigation.

Intrahepatic Portocaval Shunts and PDV

Both conditions belong to the group of portosystemic shunts (19). Regarding the intrahepatic portocaval shunts and the PDV we observed close correlations to the real findings as confirmed by digital subtraction angiography or intraoperatively. For these etiologies it is also essential to assess blood velocities within shunts and other vascular systems to choose the therapeutic strategy. Therefore, the traditional diagnostic tools such as ultrasound scan are still necessary. Also, it may be desirable to obtain information on the pressure status within the different vessels, which makes angiography in some cases unavoidable. During this procedure it is also possible to temporarily occlude the shunts and obtain detailed information on the effects on the hepatic and systemic hemodynamics. This may be important especially with regard to intrahepatic portal venous branches and their ability to adapt to the blood flow after occlusion of the shunts. Consequently, it can be decided whether a complete or partial occlusion is possible, and whether this can be achieved in the same session interventionally or by surgery afterward. However, in some cases it may be already obvious after virtual analyses that the patients' conditions are not suitable for an interventional approach. In one of our patients with PDV, this was the case because of the diameter of the duct and the flow parameters as determined by ultrasound. The virtual data thus prevented a more invasive and time-consuming diagnostic workup of the patients without loss of information before successful treatment (18).

In cases of intrahepatic portocaval shunts and PDV the analyses can be considered additional tools that may deliver important information that can either not be achieved otherwise or require much more invasive methods. The result can be a more stringent and more rapid diagnostic workup for the patients.

Portal Venous Thrombosis

Umbilical catheterization is a major factor contributing to the development of portal venous thrombosis. The Rex shunt procedure provides a curative approach for children with portal venous thrombosis because it restores portal perfusion and reduces portal hypertension (11,12). In this operation a shunt (usually the left internal jugular vein) is interposed between the mesenteric venous confluence and the recessus Rex at the extrahepatic left portal venous branch. The feasibility of this intervention is only given when the diameter of the left portal venous branch is sufficient. Furthermore, the confluence between lienal vein and superior mesenteric vein must be patent. These parameters are commonly determined using ultrasound procedures or MR angiography. In some of our patients, virtual imaging procedures were limited in displaying the portal venous vessels sufficiently. A precise evaluation toward the Rex shunt was possible (Fig. 3A). We observed 2 factors responsible for this result.

The first reason was the reduced flow and velocity pattern within distal portal branches. Because of reduced perfusion the vessels distally of the thrombosis did not show sufficient contrast medium enhancement resulting in poor segmentation results and, therefore, suboptimal reconstructions.

The second reason was the portal venous transformation, which usually exists in patients with thrombosis of the common portal vein. Collateral vessels were not well distinguishable from regular portal branches because only the morphology can be visualized.

The preparation of the Rex shunt procedure requires not only qualitative analysis of the vessels but also a quantification of blood flow within vessels of interest. For this essential quantification the ultrasound scan was indispensable in patients with portal venous thrombosis (Fig. 4).

We experienced comparable limitations using virtual planning in some patients. In some of these cases the virtual data were not sufficient to distinguish between portal branches and collaterals; also in some cases, reduced contrast in the postthrombotic vessels did not allow identification.

A major advantage of the software assistants lies in the additional information that is not achievable otherwise. In some children with portal venous thrombosis, a therapeutic approach must be taken because they present with secondary complications, especially upper gastrointestinal bleeding from esophageal varices. In these cases the children received a Warren shunt at our institution. The exact planning for this intervention could be performed based on the previously achieved data that revealed the renal vein and the lienal vein simultaneously and also allowed identification of topographical specifications as well as distance measurements. In addition, the visualization of collaterals remains important, because these vessels must be occluded in some patients during the Warren shunt procedure to achieve sufficient shunt perfusion. In earlier cases, the decision concerning the shunt procedure often had to be made intraoperatively. Here the imaging tools had a relevant impact on the therapeutical strategy.

Liver-diffuse Hemangioma

This condition is regarded as real tumor (19). The above described limitations made virtual procedures not suitable for this pathology. Liver-diffuse hemangiomas are increasingly treated interventionally (19). During catheterization it is possible to temporarily occlude feeding arteries and thus achieve an improved contrast in the process. Herein lies a relevant advantage of the interventional approach. The use of virtual imaging procedures in unilobar lesions remains to be determined.

Taken together, MeVis LiverAnalyzer and MeVis LiverExplorer provided important contributions to the analyses of children experiencing HVM, making them desirable tools for the standard workup of these children. On the basis of the observations in children with portal venous thrombosis and liver-diffuse hemangioma, we experienced limitations of the software assistants in cases of extremely reduced or increased blood flow and velocity, respectively. A relevant variation of the liver territories was observed, which confirmed previous reports of limitations of the classification proposed by Couinaud (10,20). The use of the analyzed software assistants was found to be an essential attribution to conventional diagnostic tools. Additional analyses using quantitative methods such as duplex Doppler ultrasound scan or magnetic resonance angiography remain essential for determination of blood flow and velocities (21–27). However, more invasive methods can in some cases be avoided using MeVis LiverAnalyzer and MeVis LiverExplorer.

A standardization of the presented tools seems desirable. For this purpose, studies of larger and homogenous patient populations are necessary to establish targets for inter- and intraobserver validation.

A further issue that must be taken into account is the additional application of radiation, which makes MRI-based data in the same quality desirable. In principle, the presented software tools could also be used for MRI-based data. However, at present, the quality of CT-based results in children is considered superior because the higher spatial resolution of CT compared with MRI allows more detailed reconstructions. Other studies also describe the superiority of CT-gathered data, especially in analyzing intrahepatic shunts and small HVM (28). Furthermore, MRI/angio-MRI currently results in inhomogeneous single intensities, making such complex and highly resolved 3-dimensional models less accurate. Future high-resolution MRI techniques including 4-dimensional MRI may render this approach preferable to CT-based analyses (29,30). To date, however, the use of multislice CT scan is justified in our view for obtaining virtual data because of the eminent contribution of the results to determining therapeutic approaches in distinct and complex cases of HVM in children. The presented tool of 3-dimensional reconstruction based on CT data is strongly recommended in selected cases of HVM in combination with duplex ultrasound.


1. Burrows PE, Dubois J, Kassarjian A. Pediatric hepatic vascular anomalies. Pediatr Radiol 2001; 31:533–545.
2. Boon LM, Burrows PE, Paltiel HJ, et al. Hepatic vascular anomalies in infancy: a twenty-seven-year experience. J Pediatr 1996; 129:346–354.
3. Voboyril R, Weberova J, Dvorak J. Anomalies in liver arterial blood supply and their importance for hepatic surgery. Acta Med 2002; 45:39–44.
4. Bourquain H, Schenk A, Link F, et al. HepaVision2: a software assistant for preoperative planning in living-related liver transplantation and oncologic liver surgery. In: Lemke HU, editor. Computer Assisted Radiology and Surgery (CARS). Paris: Elsevier; 2002. pp. 341–346.
5. Wald C, Bourquain H. Role of new three-dimensional image analysis techniques in planning of live donor liver transplantation, liver resection, and intervention. J Gastrointest Surg 2006; 10:161–165.
6. Fuchs J, Warmann SW, Szavay P, et al. Three-dimensional visualization and virtual simulation of resections in pediatric solid tumors. J Pediatr Surg 2005; 40:364–370.
7. Gunther P, Schenk JP, Wunsch R, et al. Abdominal tumors in children: 3-D visualisation and surgical planning. Eur J Pediatr Surg 2004; 14:316–321.
8. Smouha EE, Chen D, Li B, et al. Computer-aided virtual surgery for congenital aural atresia. Otol Neurotol 2001; 22:178–182.
9. Günther P, Ley S, Tröger J, et al. 3D perfusion mapping and virtual surgical planning in the treatment of pediatric embryonal abdominal tumors. Eur J Pediatr Surg 2008; 18:7–12.
10. Schenk A, Prause G, Peitgen HO. Efficient semiautomatic segmentation of 3D objects in medical images. In: Seminars on Medical Image Computing and Computer-assisted Interverntion (MICCAI). Vol 1935. New York: Springer; 2000: 186–95.
11. Couinaud C. Liver anatomy: portal (and suprahepatic) or biliary segmentation. Dig Surg 1999; 16:459–467.
12. Fuchs J, Warmann S, Kardorff R, et al. Mesenterico-left portal vein bypass in children with congenital portal vein thrombosis—a unique curative approach. J Pediatr Gastroenterol Nutr 2003; 36:213–216.
13. Superina R, Bambini DA, Lokar J, et al. Correction of extrahepatic portal vein thrombosis by the mesenteric to left portal vein bypass. Ann Surg 2006; 243:515–521.
14. Uchino T, Matsuda I, Endo F. The long-term prognosis of congenital portosystemic venous shunt. J Pediatr 1999; 135:254–256.
15. Vade A, Lim-Dunham J, Iqbal N. Imaging of the ductus venosus in neonates: from patency to closure. J Ultrasound Med 2001; 20:681–687.
16. Yoshimoto Y, Shimizu R, Saeki T, et al. Patent ductus venosus in children: a case report and review of the literature. J Pediatr Surg 2004; 39:E1–E5.
17. Araki T, Kamada M, Okamoto Araki T, et al. Coil embolization of a patent ductus venosus in a 52-day-old girl with congenital heart disease. Ann Thorac Surg 2003; 75:273–275.
18. Donnelly LF, Bisset GS 3rd. Pediatric hepatic imaging. Radiol Clin North Am 1998; 36:413–427.
19. Stringer MD. The clinical anatomy of congenital portosystemic venous shunts. Clin Anat 2008; 21:147–157.
20. Takehara Y, Mori K, Edagawa T, et al. Presumed hypoplastic intrahepatic portal system due to patent ductus venosus: importance of direct occlusion test of ductus venosus under open laparotomy. Pediatr Int 2004; 46:484–486.
21. Christison-Lagay ER, Burrows PE, Alomari A, et al. Hepatic hemangiomas: subtype classification and development of a clinical practice algorithm and registry. J Pediatr Surg 2007; 42:62–67.
22. Warmann S, Bertram H, Kardorff R, et al. Interventional treatment of infantile hepatic hemangioendothelioma. J Pediatr Surg 2003; 38:1177–1181.
23. Fischer L, Cardenas C, Thorn M, et al. Limits of Couinaud's liver segment classification: a quantitative computer-based three-dimensional analysis. Comput Assist Tomogr 2002; 26:962–967.
24. Ravindra KV, Guthrie JA, Woodley H, et al. Preoperative vascular imaging in pediatric liver transplantation. J Pediatr Surg 2005; 40:643–647.
25. Jequier S, Jequier JC, Hanquinet S, et al. Hepatic vein Doppler studies: variability of flow pattern in normal children. Pediatr Radiol 2002; 32:49–55.
26. Naik KS, Ward J, Irving HC, et al. Comparison of dynamic contrast enhanced MRI and Doppler ultrasound in the pre-operative assessment of the portal venous system. Br J Radiol 1997; 70:43–49.
27. Valls E, Ceres L, Urbaneja A, et al. Color Doppler sonography in the diagnosis of neonatal intrahepatic portosystemic shunts. J Clin Ultrasound 2000; 28:42–46.
28. Siddiki H, Doherty MG, Fletcher JG, et al. Abdominald findings in hereditary teleangiectasia: pictorial essay on 2D and 3D findings with isotropic multiphase CT. Radiographics 2008; 28:171–184.
29. Li G, Citrin D, Camphausen K, et al. Advances in 4D medical imaging and 4D radiation therapy. Technol Cancer Res Treat 2008; 7:67–81.
30. Cashen TA, Jeong H, Shah MK, et al. 4D radial contrast-enhanced MR angiography with sliding subtraction. Magn Reson Med 2007; 58:962–972.

hepatic vascular malformations; liver surgery; virtual surgery planning

© 2010 Lippincott Williams & Wilkins, Inc.