| BackgroundThe development in hepatic clinical anatomy has greatly promoted the advances in hepatic surgery. The traditional research on hepatic duct system was on the basis of fixed specimens from cadavers, in which the ducts were not cast and thus sank flat The hepatic spatial conformation varies from one to another so that the data about the hepatic ducts were not accurate and reliable. The flying development in modern science and technology has built a concrete ground for the vital studies on the three-dimensional (3D) reconstruction of hepatic duct systems in recent years. The 3D reconstruction through hispeed CT angiography conducted by Raptopouls V, et al displayed the full views of portal and hepatic veins as well as the spatial conformational relations anatomically so that the portal and hepatic veins can be located and their diameters, obstructions and collateral circulation can be evaluated. Wigmore SJ et al reconstructed liver from the scanned images by hispeed CT and performed the virtual surgical incision of liver for the purse of assessing the risk of hepatic failure after hepatic incision and deciding the incision range. Therefore, the 3D reconstruction of hepatic duct system is of great importance in liver conformation, diagnosis and treatment of hepatic diseases. On the other hand, however, there are a lot of defects existing in the studies on hepatic duct system. First of all, the fixed but not cast livers are faint spatially and no doubt the obtained data of duct are different from those from vital liver. Secondly, the images obtained by CT and MRJ aretwo-dimensional and the conformations of ducts are not digitalized. At last, there are many issues like: What is the relation between the anatomical data from cast liver and non-cast liver? What are the differences between the anatomical data from 3D reconstructed liver and the traditional one? Is it possible for us to obtain the data of hepatic duct systems and further create a sub-dataset of hepatic duct system of a Chinese so as to reconstruct a 3D liver with complete ducts, perform virtual surgical incision of liver and even develop a software for clinical utility. Up till now, no systematic report in this field has been published. ObjectiveOur study is to create an infused and cast model of hepatic duct system, research on the imaging method for the 3D reconstruction of liver based on CT slice layer scanning after casting the hepatic duct system, display the typical conformations of hepatic ducts from the dataset of Virtual Chinese-Female Number 1 (VCH-F1), the 3D reconstruction and the surgical incision of digitalized virtual liver, reconstruct a digitalized virtual liver with infused complete hepatic duct system and finally study the clinical value of MRA combined with MRCP in the diagnosis of carcinoma of head of pancreas. Methods1 Followed by pre-preparation, the entire liver with complete porta hepatis was infused, fixed and cast with infusing stuff and developing agents and was placed in the modeling abdominal cavity for CT slice' layer scanning.2 The images of liver from the dataset of VCH-F1 were used for observation.3 The serial images of DSCF2511-2520 from the dataset of VCH-F1 were used to create a 3D image using 3D reconstruction insertion value algorithms. Based on the color and position at the images, the hepatic parenchyma and hepatic ducts were automatically displayed in different colors.4 The serial images of DSCF2511-2520 from the dataset of VCH-F1 were used to register the images of hepatic tissue sections and then to segment and obtain the image information concerning liver parenchyma, hepatic veins, biliary duct and cholecyst.5 The entire liver with complete porta hepatis was prepared, followed by infusing, fixing and casting the hepatic artery, portal veins, inferior vena cava, bile duct system using filling materials in different colors. Then it was embedded, frozen and slice-cut by a milling machine so as to obtain the image sub-dataset of liver serial sections.6 Digitalized virtual liver(1) The 3D reconstruction of cast liver specimen by slice layer CT scanning: 1) The 3D reconstruction at the 3D working station on Hispeed CT advantage Window's 2.0. 242 two-dimensional images from No. 001 liver specimen were used to perform 3D reconstruction. ? Maximum image projection (MIP): In MIP, CT angiography was use to display hepatic veins, portal veins and hepatic arterial system. (D SSD: In SSD, the 3D reconstruction of ducts was performed by maximum density projection, followed by displaying the systems of hepatic and portal veins with different concentrations of contrast medium and then obtaining their images using threshold segmentation. After coloring them artificially, the conformations of ducts in different colors were obtained.2)The 3D reconstruction assisted with computer: 242 two-dimensional images from No. 001 liver specimen were used to perform 3D reconstruction. Surface rendering was adopted, by which all the images were read in and smoothened using Gaussian smoothing algorithm, followed by obtaining the external margins of hepatic ducts and parenchyma based on pixel values of the images using isosurface extraction. After completing obtaining the surfaces twice as above, Gausian smoothing algorithm was used once again to ensure the smoothness, followed by writing the obtained data onthe surfaces into a VTK document. Finally, the 3D reconstruction was realized by GUI program based on VC++.(2) The sub-dataset of liver from VCH-F1 and the digitalized sub-dataset of virtual liver obtained through slice-cutting the frozen cast liver specimen at 0.2mm by a milling machine:1) Image registration: In our experiment, image registration was conducted by external point force combined with moment-to-force. The markers pre-embedded around the cadaver specimen or near the liver were set as registration points and the image registration was performed based on the relatively-fixed positions between the liver and those markers.2) Image segmentation(D Segmentation of liver sub-dataset from VCH-F1: The hepatic vein/portal vein segmentation algorithm based on image serial correlation was used. The two adjacent images in the serial sections had minor differences so that a certain duct appeared nearly at the same location or overlapped in the two adjacent images. Therefore, once an area of hepatic veins was marked in a single image, the overlapped area in the adjacent image was sure to be hepatic veins. In this way, the images of hepatic and portal veins were segmented after scanning all the images twice. For the extraction of liver image, the threshold segmentation was used.?Segmentation of liver : The arbitrary point of the image was labeled brown (100, 50, 50) if its color was not white (255, 255, 255), red (255, 0, 0), blue (0, 0, 255) or green (0, 255,0).?Segmentation of sub-dataset of infused and cast liver images slice-cut by the milling machine: It mainly aimed at the segmentation of hepatic and portal veins. The procedures included: A. Gaussian-Rapras algorithm was used to extract all the contours of both liver and ducts. B. The contours were expanded and refined toconnect their broken lines. C. The type of tissues incorporating in the contours wasjudged and identified by fully marking the portal veins with red color if they werereddish, fully marking the hepatic veins with blue color if they were bluish and finallymarking the liver parenchyma with brown color if they were neither reddish norbluish. Following the segmentation of all images, the image data of liver parenchyma,hepatic veins and inferior vena cava, portal veins, hepatic artery, bile duct andcholecyst were all extracted, respectively.?3D reconstruction: Based on Visualization Tool Kit (VTK), the multiple serialtwo-dimensional colored images of the cross sections were used as the data resourceaccording to the image characteristics of the liver tissue sections. The surfacedescription was used to demonstrate the reconstructed 3D liver model with hepaticsurface and internal conformations and to develop it into a 3D visualizeddemonstration system of liver based on Windows platform of PC.(3)The digitalized virtual liver with cast hepatic ducts (hepatic artery, hepatic veins,portal veins and bile duct)7 42 cases with carcinomas of head of pancreas were examined by MRI and MRI3D reconstruction system, which was contrasted with the results via surgicalexplorations and operative approaches.Results1 The hepatic artery, hepatic veins, portal veins and bile duct were all infused and cast, respectively. Taking cast hepatic artery as reference standard, the capillaries of hepatic artery on the hepatic surface were well displayed, which zigzag projected a bit from the hepatic envelop like fine silk, while the capillaries in the fundus of gallbladder took a net-like form.2 The sub-dataset of hepatic cross-sections(1) The sub-dataset of liver from VCH-F1 There totally were serial 875 images ofthe liver tissue, which sharply displayed the anatomical boundaries between liver and its surrounding conformations as well as the position relations anatomically between hepatic artery and abdominal aorta, inferior vena cava, diaphragm, stomach, duodenum and left kidney. Each image of cross sections showed complete duct conformations in liver, hepatic artery and veins showed to be red, bile duct and cholecyst brown and portal veins back at the majority of images.(2) The sub-dataset of cast hepatic ducts by hispeed CT slice layer scanning Each scanned image of cross section showed complete duct conformations in liver. At the center of the image, inferior vena cava, the right, middle and left veins of liver and even the branches of the portal veins at right and left liver. Some duct conformations were markedly reinforced, in contrast to the liver parenchyma and the portal veins could be seen clearly in the image of porta hepatis.(3) After slice cutting cast liver and its ducts by a milling machine, 910 serial cross-section images on liver were obtained and they were sharp and clear, with hepatic artery, portal veins, inferior vena cava/hepatic veins, bile duct system displayed in red, brownish red, black and blue.3 The 3D reconstruction of liver(l)The 3D reconstruction of infused and cast liver specimen by CT hispeed layerscanningA The reconstruction based on the 3D working station of Hispeed CT AdvantageWindow's 2.0: The 3D reconstructed images by MIP and SSD displayed the clearinternal ducts in liver, especially the duct twigs in the hepatic vein system and theportal vein system. We could make sure the dynamic features under differentconditions by observing the 3D dynamic imaging from front to rear, left to right or upto down. The 3D reconstructed images by SSD were well stereo and the ducts in themcould be displayed in different colors only through applying pseudo-colors to themaccording to their respective CT values obtained using silver powder. B The 3D reconstruction using image processing: The images of liver parenchyma, hepatic veins and portal veins were displayed respectively or in combination. Their 3D conformations were clearly observed by rotation and zooming. (2) The sub-dataset of liver from VCH-F1 and the 3D reconstructed virtual liver obtained from the digital sub-dataset by slice-cutting the frozen cast liver specimen with the milling machine: There totally were 875 images of the liver tissue, with 0.2 mm inter-space. Each cross section image would successively and completely demonstrate the liver and its corresponding duct system. The registered images were within the rectangular area of 1100X900, which ideally demonstrated the full view of the liver. According to different demands, the images of cholecyst, bile duct and hepatic parenchyma were respectively segmented and the corresponding images were accomplished. The serial 875 liver tissue images were performed with registration and segmentation, followed by the 3D reconstruction to accomplish the three-dimensionally reconstructed liver model. We set the portal vein permanently as 1, and when the pellucidit of hepatic parenchyma was 0 and the pellucidit of others was 1, the duct system of the liver could be demonstrated; when the pellucidit of hepatic vein was 1 and that of others was 0, the 3D model of the whole hepatic vein and portal vein was demonstrated; when the pellucidit of the hepatic artery, bile duct and cholecyst was set 1 and that of others was 0, the 3D demonstration of the portal vein, cholecyst, bile duct and the hepatic artery was shown; through rotation, the 3D histological image from each perspective could be achieved; when the pellucidit of the liver was set 1 and that of the hepatic vein, hepatic artery, bile duct, cholecyst and abdominal aorta was set 0, the hepatic parenchyma was demonstrated; when the pellucidit of the hepatic artery was set 1 and that of others was 0, the 3D structures of the portal vein and hepatic artery were demonstrated; when the pellucidit of thehepatic parenchyma was set 0.5 and that of other tissues was 1, both the hepatic parenchyma and the duct system of hepatic artery, portal vein, hepatic vein and bile duct were demonstrated.4 Clinically, 26 cases out of the 42 patients with carcinoma of head of pancreas were further found to oppress their adjacent blood vessels. Among them, 9 had invaded into the adjacent blood vessels and tissues and 3 had metastasis into distant organs. The diagnosis results accorded to surgical explorations in 35 cases, with the accuracy rate 89.7%. Surgical removal of pancreas and duodenum was performed in 28 cases, with removal rate 66.7% against the all 42 cases, internal drainage in 11 and interventional therapy in 3. Conclusions1 The infusion and casting of hepatic duct system is workable. MIP, SSD and computer-based images treatment, which are characterized by 3 dimensionally constructing images sharply, is ideal for the study of hepatic duct conformations.2 The accurate sub-dataset of live from VCH-F1 and the sub-dataset of hepatic serial cross-section images obtained by slice-cutting the infused, cast and frozen liver specimen with the milling machine have set the hepatic 3D reconstruction and digitalized virtual liver on a reliable data ground. VTK and value insertion algorithm, which can 3 dimensionally reconstruct the images of liver and its duct system to a clear and sharp extent, are ideal for 3D reconstructions of images.3 The 3D visualized liver demonstration system using the sub-dataset of live from VCH-F1 and the sub-dataset of hepatic serial cross-section images obtained by slice-cutting the infused, cast and frozen liver specimen with the milling machine has been satisfactorily developed based on PC-based VTK with windows platform, which is of great significance for further study on digitalized virtual liver, the establishment of Anatomy of Human Digitalized Liver and the promotion of hepatic clinicalanatomy.4 MRA combined by MRCP can definitely display the tumor metastasis andinvasion, especially its blood vessels' invasion into the adjacent tissues duringpreoperative diagnosis in patients with carcinoma of head of pancreas. The diagnosticresults by imageology are of great value for evaluation of pathological changes anddecision of clinical treatment.
|
PDF Full Text Request