The Three-dimensional Reconstruction And Anatomy Of Celiac Artery And Portal Vein With CT Scanning Data

Background and purpose:Primary hepatic carcinoma is a common malignant tumor in China. The hepatic artery supplies 90% to 99% of their blood supply, while portal vein supplies the surrounding cancer cells. In the hepatoma resection, normal and abnormal tumor blood supply should be accuratly ligation, while the feeding arteries of carcinoma should be superselective catheterization during transcatheter hepatic arterial chemoembolization. The hepatic artery and portal vein are in Glisson system, while the portal vein is important in hepatic segment resection because its caliber is thick and branches are relatively stable. Therefore, to fully understand the anatomical variations of origin, courser of hepatic artery and portal vein have a very important significance.Pancreatic cancer is a common malignancy of digestive system and surgical resection is the only means of cure, but when a clear diagnosis of pancreatic cancer there is more involved to varying degrees of pancreatic artery. Therefore, a detailed understanding of pancreatic peripheral vessels especially the anatomical variations and involvement of pancreatic feeding artery is needed presurgery. Celiac artery is the main blood supply of upper abdominal organs, while the hepatic artery and most of the pancreatic feeding artery are the branches directly. With the continuous improvement of surgical techniques and expanding of interventional treatment in relevant upper abdominal organs, mastering the normal abdominal anatomy and variations of celiac artery system becomes the important guarantee of the abdominal surgery and interventional treatmentThere are methods for study vascular structures in previous studies. Autopsy, being the oldest research methods, has been confined by corpse source and distortion inevitably through various processing. Intraoperative vivisection which is influenced by operation time, scope and manipulation, can not carry out extensively, as well as DSA, which is involved with complication caused by operator experience, catheter diameter, contrast media dosage and so on. Multi-row spiral computed tomography (MSCT), thanks to isotropic spatial resolution, the extremely fast CT data acquisition and the consequent optimization of enhancement, has made MSCT angiography (MSCTA) which is non-invasive invaluable in the evaluation of vascular anatomy, and in particular the splanchnic vessels.Three-dimensional (3D) reconstruction technique, being an important research direction of computer image processing, had been applied extensively in anatomic study of skeleton, vascular, gastrointestinal tract and so on. The starting material of 3D reconstruction includes digital human database which is associated with tremendous data volume and longer time when reconstruction although its high accuracy, magnetic resonance angiography (MRA) which is influenced by breath, heartbeat, enterokinesia and so on when applied in abdominal region, and MSCTA which clear display the normal anatomy and variation of abdominal vascular and its branch and is benefit for the segmentation and 3D reconstruction of target.MSCTA were evaluated and interactively processed on dedicated 3D reconstruction workstation in previous literature. The model reconstructed by workstation can not output, which limit apply in virtual operations research. Therefore, in these studies, based on 64-row helical CT scanning data, utilizing computer-assisted 3D reconstruction soft ware reconstructed celiac artery and portal venous system, and homologized anatomic data was obtained from models through observation and measurement. These results will provide 3D models of various types of anatomical variations for surgery planning and simulation, offer correlate living anatomical data for intervention, supply digitalization atlas for anatomic teaching and enrich anatomic teaching method. This topic was supported from the team project of natural scientific foundation in Guangdong province.Materials and methods:1. Clinical data120,170,42 and 64 cases, which were examined by MSCT in Zhujiang hospital of Southern Medical University from January to October 2007 and complied with no pathologic state which may alter vascular structure, were used to 3D reconstruct the celiac artery, hepatic artery, pancreatic feeding artery and portal venous system separately.2. CT scanning methodAll of the CT examinations were performed with a 64-row helical CT device (Philips Brilliance 64, Philips Mecial Systems, Eindhoven, The Netherlands). Examination parameters were 120 kV,300 mA,5-mm section thickness,5-mm reconstruction interval,0.984 pitch and 0.5 second rotation time. All patients were placed in the supine position and scanned from the diaphragm to the inferior pole of double kidney. After plain scan, the nonionic iodinated contrast agent at a concentration of 370mgI/ml, which was inserted into the cubital vein, was used at 1.5 ml/kg, with an infusion rate of 5 ml/s. Arterial phase and venous phase images were acquired 20-25s and 50-55s respectively after intravenous injection. The image whose format was DICOM 3.0, were transferred to the Mxview workstation and stored in portable hard risk.3.3D reconstruction and anatomyCeliac artery, pancreatic feeding artery and portal venous 3D model respectively were reconstructed by Mimics software, hepatic artery 3D models were reconstructed by MIPS software, and all models were observed and measured for revelant anatomic data. Statistical analyses were made with the Statistical Package for Social Sciences version 13.0 for Windows (SPSS Inc, Chicago, Illinois, USA). Various length of CA was indicated by the mean value±standard deviation (x±s). McNemar test was used for evaluation of the depiction rate of pancreatic feeding arteries in axial CT images and 3D models. P< 0.05 was accepted as statistically significant.Results:1. CA1.1 The 3D model displayed vivid three-dimensional structures. The depiction rate were as follows:100% of celiac, common hepatic, splenic and gastroduodenal arteries,78.3% of left gastric artery,98.3% of proper common artery,79.2% of left hepatic artery and 95% of right hepatic artery.1.2 The classical configuration of the CA was detected in 107 cases of our series (89.2%) and the length of CA was measured for (2.76±0.60) cm. The gastro-splenic trunk was detected in 6 cases (5%), the gastro-hepatic trunk was detected in 1 case (0.8%), the gastro-hepato-spleno-mesenteric trunk was detected in 2 cases (1.7%), the CA quadrifurcation was detected in 3 cases (2.5%), and the SA together with a short trunk which diverged into CHA and LGA was detected in 1 case (0.8%).1.3 The CA arose at the level between the 12th thoracic vertebra and 1 st lumber vertebra in 96.7% and its course showed many styles, mostly toward right in 83.3%.2. Hepatic artery2.1 Though the model reconstructed by MIPS represented the shape and structure of anatomical variations of hepatic arteries, its surface was not smooth with many irregular chips. After smoothed and modified by FreeForm Modeling System, the model presented with realistic profile of hepatic artery that allowed zooming, rotation and vivid 3D observation in full views.2.2 Among 170 cases of hepatic artery models, the normal anatomy of Michels typeⅠwas observed in 113 cases (66.5%), while the anomalies consisted of Michels typeⅡin 14 cases (8.2%), typeⅢin 7 cases (4.1%), typeⅣin 4 cases (2.4%), type V in 4 cases (2.4%) and typeⅥin 2 cases (1.2%). No cases of typeⅦandⅧwere detected. TypeⅨanⅩwere observed in six cases (3.5%) and one case (0.6%) separately. Unclassified variations were observed in 15 of the cases(8.8%).3. Pancreatic feeding artery3.1 In axial CT images and 3D models, depiction rates of ASPDA were 100%, 88.1% respectively; PSPDA 100%,85.7%; IPDA 97.1%,79.4%; AIPDA 92.9%, 59.5%; PIPDA 90.5%,45.2%; DPA 95.2%,81%; PMA 83.3%,52.4%; TPA 71.4%,47.6%. There was significant difference in depiction rates of PSPDA, IPDA, AIPDA, PIPDA, DPA, PMA and TPA between axial CT images and 3D models.3.2 IPDA whose frequency was 81%, were absent in 8 cases among 42cases. There were 15 cases had anatomic variation of origin, including 7cases DPA arose from SMA (16.7%) and 8 cases variations were different from each other.4. Portal vein4.1 In 3D models of portal venous system, the depiction rates of 4 grade branches were 100%, while 5 grade branches 90.6%, more than 6 grade branches 53.1%.4.2 The depiction quality of portal venous and its branches were as follows:Ⅰgrade of MPV had 10 cases (15.6%),Ⅱgrade 39 cases (60.9%),Ⅲgrade 15 cases (23.4%),Ⅳgrade no cases;Ⅰgrade of RPV which were absent in 11 cases had 21 cases (39.6%),Ⅱgrade 26 cases (49.1%),Ⅲgrade 6 cases (11.3%),Ⅳgrade no cases;Ⅰgrade of LPV had 25 cases (39.1%),Ⅱgrade 33 cases (51.6%),Ⅲgrade 6 cases (11.3%),Ⅳgrade no cases;Ⅰgrade of SMV had 6 cases (9.4%),Ⅱgrade 34 cases (53.1%),Ⅲgrade 22 cases (34.4%),Ⅳgrade 2 cases (3.1%);Ⅰgrade of SPV had 9 cases (14.1%),Ⅱgrade 32 cases (50%),Ⅲgrade 22 cases (34.4%),Ⅳgrade 1 case (1.6%).4.3 Common branching patterns of portal vein were observed in 50 cases (78.1%), while the anomalies consisted of typeⅠin 9 cases (14.1%), typeⅡin 2 cases (3.1%), typeⅢin 1 case (1.6%) and typeⅣin 2 cases (3.1%).Conclusions:The enhancement scanning data of 64-row helical CT clearly displayed normal anatomy and variation of abdominal vascular whose border was sharp to segment. The 3D models of CA and portal vein system reconstructed by Mimics and MIPS software represented the shape and structure of anatomical variations and can be measured for anatomic data. These results will provide 3D models of various types of anatomical variations for surgery planning and simulation, offer correlate living anatomical data for intervention, supply digitalization atlas for anatomic teaching and enrich anatomic teaching method.