| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Department of Radiology, Duke University Medical Center, Box 3808, Durham, NC
27710.
2 Department of Radiology, University of Wisconsin Medical Center, Madison, WI
53706.
3 Department of Surgery, Duke University Medical Center, Durham, NC 27710.
Received January 13, 2004;
accepted after revision May 6, 2004.
Address correspondence to E. K. Paulson
(pauls003{at}mc.duke.edu).
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. Sixty-three patients with colorectal metastases (n = 30), hepatocellular carcinomas (n = 13), giant hemangiomas (n = 5), and other lesions (n = 15) underwent dual-phase MDCT using either a LightSpeed QX/i 4-MDCT (n = 31) or LightSpeed QX/i Ultra 8-MDCT (n = 32) scanner. Contrast material (150 mL of Isovue 370 [iopamidol]) was injected at a rate of 5 mL/sec. The arterial phase images were rendered on a workstation to obtain 3D MDCT angiograms that were assessed by two reviewers who were blinded to the surgical findings. Arterial anatomy was categorized according to the Michels classification. The reviewers assessed the 2D and 3D images for quality, arterial branch visualization, and differences between the 4- and 8-MDCT images. In the 43 patients who underwent resection, imaging findings were correlated with intraoperative findings.
RESULTS. The anatomy of hepatic arteries in the 63 patients was classified as follows: Michels type I, 51 patients (80.9%); type III, four patients (6.3%); type V, five patients (7.9%); and types VII, VIII, and IX, one patient (1.6%) each. In 40 (93%) of 43 patients, the surgical findings concurred with MDCT findings. Three discrepancies were due to failure to identify small accessory left hepatic arteries. Branch visualization and image quality of the 2D images were superior to those of the 3D images. No significant difference was found between the 4- and 8-MDCT images in branch visualization and image quality.
CONCLUSION. Three-dimensional MDCT angiography is accurate for classification of hepatic arterial anatomy before hepatic resection. Although 2D data sets show small arteries to better advantage than 3D MDCT angiograms, the 3D MDCT angiograms provide a useful overview of hepatic anatomy.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Studies using single-detector CT and 4-MDCT technology have indicated that CT angiography of the liver is both feasible and accurate and may be a reasonable alternative to conventional angiography in patients undergoing evaluation for liver surgery [4–11]. To our knowledge, no studies have used 8-MDCT for this purpose nor have any studies determined whether the image quality and arterial anatomy are best viewed with a 2D or 3D approach.
The purpose of our study was to review our experience with MDCT angiography of the liver using dual-phase 4- or 8-MDCT, with particular focus on its utility in the delineation of hepatic arterial anatomy in patients who are scheduled to undergo hepatic resection. In addition, the 2D and 3D images were compared for overall quality and arterial branch visualization.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
All 63 patients underwent dual-phase MDCT on LightSpeed QX/i 4-MDCT (n = 31) or LightSpeed QX/i Ultra 8-MDCT (n = 32) scanners (GE Healthcare). Each study consisted of an unenhanced acquisition of the liver including the celiac axis and superior mesenteric artery. For the LightSpeed QX/i 4-MDCT scanner, the protocol for the unenhanced acquisition was 140 kVp, 130–230 mA, 5-mm thickness, pitch of 1.5, 15 mm per rotation, and 0.8-sec per rotation. For the LightSpeed QX/i Ultra 8-MDCT scanner, the protocol for the unenhanced acquisition was 140 kVp, 130–230 mA, 5-mm thickness, pitch of 1.35, 27 mm per rotation, and 0.5-sec per rotation.
Contrast material (150 mL of Isovue 370 [iopamidol, 370 mg I/mL], Bracco Diagnostics) was injected at a rate of 5 mL/sec via a mechanical power injector (PercuPump II, E-Z-EM; or Invision, MedRad). The timing of the arterial phase was determined with bolus tracking technology (SmartPrep, GE Healthcare) by placing the region of interest around the descending thoracic aorta at the dome of the diaphragm and measuring the change in aortic enhancement over time. Arterial phase scanning was initiated approximately 5 sec after identifying the onset of the upward slope of aortic enhancement, resulting in an average scanning delay of 20–25 sec for the arterial phase. A specific aortic threshold value was not used for the initiation of the arterial phase.
Arterial phase scanning was performed from the liver dome through renal hilum. For the LightSpeed QX/i 4-MDCT scanner, the arterial phase protocol was 140 kVp, 130–230 mA, 2.5-mm thickness, pitch of 1.5, 15 mm per rotation, and 0.8 sec per rotation. For the LightSpeed QX/i Ultra 8-MDCT scanner, the arterial phase protocol was 140 kVp, 130–230 mA, 1.25-mm thickness, pitch of 1.675, 16.75 mm per rotation, and 0.5 sec per rotation.
Portal venous phase imaging was conducted with a fixed scanning delay of 65 sec after contrast material administration, with scanning this time performed from the dome of the liver through the symphysis pubis. The portal venous phase protocol for the LightSpeed QX/i 4-MDCT scanner was 140 kVp, 130–230 mA, 5-mm thickness, pitch of 1.5, 15 mm per rotation, and 0.8 sec per rotation. The protocol for the LightSpeed QX/i Ultra 8-MDCT scanner was 140 kVp, 130–230 mA, 5-mm thickness, pitch of 1.675, 16.75 mm per rotation, and 0.5 sec per rotation. Oral contrast material was not administered as part of this protocol to reduce streak artifact and nonvascular contamination on the volume-rendered images. However, eight patients who inadvertently received oral contrast material were included in this study. The 2.5-mm arterial phase images from the 4-MDCT scanner were reconstructed at an interval of 1.0 mm. The 1.25-mm arterial phase images from the 8-MDCT scanner were reconstructed at an interval of 1.0 mm.
A workstation (Vitrea 2, Vital Images) was used to generate 3D models of the hepatic vascular anatomy. Each case was rendered by a board-certified radiologist with expertise in abdominal imaging. Cine loops were generated from both 2D axial source images and 3D volume-rendered images obtained during the arterial phase study. The 3D volume-rendered images were created using the segmentation tool while the threshold criteria were adjusted to optimize anatomic detail and minimize venous contamination. Specific prescribed threshold values were not applied. High-attenuation structures (e.g., bone, bowel containing oral contrast material, or surgical clips) were removed using the 3D sculpting tool. The estimated time required to perform these manipulations ranged from 5 to 30 min for each case.
Two board-certified radiologists with fellowship-training in abdominal imaging subsequently reviewed each case by consensus. The hepatic arterial anatomy was categorized according to the Michels classification [14] (Table 1). Both radiologists were blinded to the surgical findings in each case.
|
The intraoperative findings constituted the gold standard for the purpose of this study. Intraoperative findings included direct visualization of the hepatic vasculature, as well as the findings of the routinely performed intraoperative sonography (performed by the surgeon using a ProSound 5000 scanner, Aloka). Intraoperative sonography was used to confirm known masses and to detect metastatic deposits that may not have been detected on preoperative imaging. Intraoperative sonography was also used to assess the vessels around the tumors and in the porta hepatis. The surgeon who performed each of the surgeries knew the results of the MDCT angiography, posted selected MDCT images in the operating room, and routinely verified the MDCT angiographic findings or reported any discrepancies at the time of surgery.
To compare the 2D and 3D images, another fellowship-trained abdominal imager individually graded each study (both 2D and 3D) for overall quality on a 4-point scale (1, excellent; 2, good; 3, adequate; and 4, poor). Excellent quality (grade 1) referred to images with densely opacified, sharply marginated intrahepatic arteries that were continuous throughout the data set. Good quality (grade 2) referred to images with vivid opacification of intrahepatic arteries and few, if any, regions of discontinuity along the arterial path. Adequate quality (grade 3) referred to images that permitted adequate visualization of the main branches from the common hepatic artery but poor visualization of smaller intrahepatic branches and that had regions of vascular discontinuity. Poor quality (grade 4) referred to images on which vascular opacification was too faint for the reviewers to delineate the main branches arising from the common hepatic artery.
In addition, both the 2D and 3D images were evaluated for extent of visualization of arterial branches. For each set of images, the highest order branch (i.e., the smallest artery) visualized was recorded. The common hepatic artery was considered the first-order branch. The left and right hepatic arteries were considered second-order branches; the segmental branches, third-order branches; and the subsegmental branches, fourth-order branches. Fifth- and sixth-order branches were not scored.
Differences between arterial branch order and quality scores for the 2D and 3D images were assessed with the Wilcoxon's signed rank test. The differences between the 4- and 8-MDCT scanners for visualization of the branch order and image quality were compared with the Wilcoxon's rank sum test.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
On the basis of the review of the hepatic resection MDCT, the hepatic arterial anatomy was categorized as Michels type I in 51 patients (80.9%), type III in four patients (6.3%), type V in five patients (7.9%), and types VII, VIII, and IX in one patient (1.6%) each (Table 1 and Fig. 1).
|
The hepatic arterial anatomies of the 43 patients who eventually underwent hepatic resection were identified on the preoperative CT as type I in 34 patients (79%), type III in four patients (9.3%), type V in three patients (6.9%), and types VIII and IX in one patient (2.3%) each (Table 1). In these 43 patients, concordance between the MDCT angiographic and surgical findings was found in 40 (93%). In the remaining three patients (3/43 or 7%), small accessory left hepatic arteries were not identified on the preoperative MDCT angiography but were present at the time of surgery (Fig. 2A, 2B). As a result, these patients were incorrectly classified as having conventional or type I hepatic arterial anatomy when, in fact, they had type V hepatic arterial variant anatomy. In retrospect, these small accessory arteries were present on the 2D axial source images, although these vessels were not identified on the 3D volume-rendered images, probably because of their small size. In two of these cases, surgical management or placement of a chemotherapy pump was affected by the presence of the accessory left hepatic arteries. Two of these patients underwent CT using a 4-MDCT scanner; one underwent CT with an 8-MDCT scanner.
|
|
Overall, the visualization of peripheral branch arteries provided by the 2D images was far superior to that of the 3D images (Fig. 3A, 3B). Specifically, all 63 2D studies showed visualization of either third- or fourth-order branches, whereas only 42 (67%) of the 63 3D studies showed vessels of this caliber (Table 2). Additionally, the mean scores for branch order visualization was significantly higher with the 2D than with the 3D studies (p < 0.0001) (Table 3).
|
|
|
|
Furthermore, the overall quality of the 2D images was superior to that of the 3D images. For example, 21 (33%) of the 63 2D studies had a quality score of excellent, whereas only three (4.8%) of the 63 3D studies had a score of excellent (Table 2). In addition, the mean quality for the 2D images was rated as superior (i.e., lower scores indicating better quality) to that of the 3D images (Table 3). Differences in subjective measures of image quality were statistically significant (p < 0.0003).
The 4- and 8-MDCT scanners provided similar results in the visualization of the mean branch order (Table 3). However, for the 3D images, we found a trend toward superior branch visualization (mean score of 2.48 for 4-MDCT scanner vs mean score of 2.81 for 8-MDCT scanner, not statistically significant). No significant difference was found between the 4- and 8-MDCT scanners in overall quality (Table 3).
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Our study shows that hepatic CT angiography with MDCT technology is feasible and accurate in identifying specific hepatic arterial variants, a conclusion that is corroborated by several studies in which either conventional angiographic or surgical findings served as the reference standard [6, 8–11]. Takahashi et al. [6] and Sahani et al. [8] compared 4-MDCT angiography with conventional angiography in patients undergoing preoperative evaluation. The accuracy of MDCT angiography in identification of arterial variants in the two studies proved to be 97% and 98%, respectively, similar to our 93% accuracy. Likewise, in a group of patients undergoing surgical planning for liver transplantation, single-detector CT and 4-MDCT angiography accurately confirmed variant anatomy in 92% and 93% of patients, respectively [5, 9]. In contrast, Bogetti et al. [15] found that 4-MDCT angiography provided an accurate depiction of variant anatomy in only six (67%) of nine patients.
Our study confirms that variant anatomy of the celiac axis and its branches is common, occurring in 20% of our patients. The incidence of variant anatomy has been reported to be as high as 24–49% in large intraoperative and cadaveric series and as high as 21–40% in CT angiography series [9, 10, 14, 16–18]. In many thin patients with sparse fat in the porta hepatis, hepatic arterial variant anatomy can be determined by manual and visual inspection at the time of surgical exploration. However, in other patients, it may be difficult to visualize the hepatic arterial anatomy, such as in those patients with large amounts of lymphatic and fatty tissue in the duodenal hepatic ligament, those with hepatic tumors that infiltrate the porta hepatis, those who have had prior episodes of cholangitis, and those with primary biliary tumors. In these patients, preoperative knowledge of the arterial anatomy is particularly useful. Indeed, numerous surgical complications have been attributed to the inadvertent ligation or injury of variant hepatic arteries including hepatic ischemia, hepatic failure, biliary stricture, biliary leak with biloma, and hemorrhage [19].
A specific clinical situation in which knowledge of hepatic arterial variants is particularly important is presented by a patient with a large tumor that encroaches on the hilum of the liver. In such a patient, determining the specific branch point of the left and right hepatic arteries is essential as well as whether the right hepatic artery courses anterior or posterior relative to the common hepatic or common bile duct. Even in situations in which the right hepatic artery is involved by tumor, the left hepatic artery may be free of tumor if it has an early takeoff from the common hepatic artery. Accordingly, a right hepatectomy may be performed for a cure. A patient with a large replaced left hepatic artery may also be a candidate for curative resection in this setting.
In addition, knowledge of accessory and replaced arteries is important during the dissection required for a right hemihepatectomy. For example, in a patient with conventional anatomy, the left hepatic artery takes off proximally, well away from the main portal vein or common bile duct. In this instance, a standard dissection at the base of the medial segment is unlikely to result in incidental injury to the left hepatic artery. However, if the common hepatic arterial bifurcation (giving off the right and left hepatic artery) is distal and nearly intrahepatic, a standard hilar dissection performed for a right hemihepatectomy may injure the left hepatic artery, risking hepatonecrosis of the left liver remnant.
A resurgence of enthusiasm for hepatic artery pump placement followed the publication of the findings from a randomized trial of hepatic arterial infusion of chemotherapy versus systemic therapy in patients undergoing hepatic resection for colorectal metastases [20]. Although older randomized trials did not clearly show a benefit from hepatic arterial chemotherapy in patients with unresectable lesions, more recent studies [17, 20–23] suggest that patients with resectable lesions may benefit from adjuvant hepatic arterial chemotherapy in conjunction with systemic chemotherapy.
Although pump placement is well tolerated in most patients, complications may occur [17, 20–23]. Clearly, variant anatomy may contribute to hepatic artery pump complications including incomplete perfusion of the liver or liver remnant and extrahepatic perfusion, which may cause vessel thrombosis or misperfusion of chemotherapeutic agents (Fig. 4A, 4B). Many pump complications can be avoided through careful placement and adequate knowledge of the branching patterns of the celiac axis that can be accurately depicted on MDCT angiography [10]. For example, a gastroduodenal artery arising from the right hepatic artery or from a trifurcation point would necessitate ligation of one of the main hepatic arteries to ensure complete perfusion of the liver remnant. In such a case, if the left hepatic artery is not ligated to ensure cross-filling from the right hepatic lobe (which can usually be safely done without hepatonecrosis), the left hepatic lobe will not be perfused with the chemotherapeutic agent.
|
|
The limitations to this study should be acknowledged. We elected to use the intraoperative surgical findings, which included those of intraoperative sonography, as the gold standard. Although the operative findings are limited by the extent of the exposure of the porta hepatis at the time of resection, most hepatic arterial variants can be accurately characterized during surgery. The participating surgeon routinely identified the arterial anatomy and reported on any concordant or discordant vessels. An additional limitation to the study is that the 4- and 8-MDCT protocols were not identical, although they were similar in many respects. The protocols are those used in our routine clinical practice. Finally, this study was a retrospective, not prospective, review.
Our study found that the 2D axial source images were superior to the 3D images in branch order visualization and overall quality. The differences in arterial visualization likely reflect the postprocessing required to create MDCT angiograms. Specifically, creation of an MDCT angiogram requires selection of an appropriate threshold such that arteries are vivid, but surrounding parenchyma and early enhanced veins are excluded. An inevitable result of appropriate threshold selection is the exclusion of some areas of enhancement particularly from small-diameter or faintly opacified arteries. Evidence for this exclusion is that in many cases, small branch arteries were well depicted on 2D images but not visualized on the 3D images. Indeed in the three cases in which the arterial anatomies were misclassified, small accessory left hepatic artery branches were present on the 2D axial source images but not seen on the 3D MDCT angiograms. Other investigators have reported similar difficulty in identification of small-diameter accessory hepatic arteries on MDCT angiograms [5, 6, 9]. Nevertheless, the 3D MDCT angiograms provide a visually pleasing and almost instantaneous overview of the celiac artery and its branches. Many surgeons find MDCT angiograms helpful because they are familiar with conventional angiograms. Furthermore, the orientation of vessels on MDCT angiograms is analogous to that identified during dissection of the porta hepatis.
When evaluating patients before liver resection, we suggest a combined approach. The 3D CT angiograms provide a useful overview, nicely depict most variants, and illustrate the relationship of branch points to each other. The 2D images are useful for assessment of small-diameter arterial variants that may not be apparent on the 3D images. This combined approach has been suggested by others [5]. Although in our study, we restricted analysis to 2D axial source images and 3D MDCT angiograms, other display options, including maximal intensity projections and thick-slab multiplanar reformations, may prove useful.
To our surprise, the 8-MDCT images proved similar to the 4-MDCT images in arterial branch visualization and overall quality even though the slice thickness with the 8-MDCT technology was thinner (1.25 vs 2.5 mm). In fact, an in vitro phantom study comparing 4- and 8-MDCT scanners showed improved z-axis resolution with the 8-MDCT scanner [24]. The similarity encountered in this clinical study likely reflects the fact that the protocols, although not identical, were similar in most respects. Furthermore, we suspect that if a larger number of patients were studied, differences in image quality would become apparent. Likewise, we hypothesize that images created from 16-, 32-, or 64-MDCT scanners will prove superior to images created from 4-MDCT scanners, assuming that noise does not increase substantially.
In summary, 4- or 8-MDCT angiography is accurate for classification of hepatic arterial anatomy before hepatic resection. Although the 2D data sets show small arteries to better advantage than 3D MDCT angiograms, the 3D MDCT angiograms provide a useful overview of hepatic arterial anatomy.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  C. Heilmaier, R. Sutter, A. M. Lutz, B. Seifert, D. Weishaupt, B. Marincek, and J. K. Willmann Mapping of Hepatic Vascular Anatomy: Dynamic Contrast-enhanced Parallel MR Imaging Compared with 64 Detector Row CT Radiology, December 1, 2007; 245(3): 872 - 880. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. Ishigami, K. Yoshimitsu, H. Irie, T. Tajima, Y. Asayama, M. Hirakawa, and H. Honda Accessory Left Gastric Artery from Left Hepatic Artery Shown on MDCT and Conventional Angiography: Correlation with CT Hepatic Arteriography Am. J. Roentgenol., October 1, 2006; 187(4): 1002 - 1009. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. T. Johnson and E. K. Fishman IV Contrast Selection for MDCT: Current Thoughts and Practice Am. J. Roentgenol., February 1, 2006; 186(2): 406 - 415. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
