点击显示 收起
1 From the Department of Radiology, Osaka University Medical School, Osaka, Japan (S.T., T.M., M.T., T.K., M.H., Y.N., H.N.); and GE Yokogawa Medical Systems, Ltd, Tokyo, Japan (M.K.). Received January 17, 2001; revision requested February 20; revision received May 8; accepted June 20.
| ABSTRACT |
|---|
| Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
|---|
MATERIALS AND METHODS: In 62 patients, pre- and postcontrast triphasic helical CT were performed by using a multi–detector row CT scanner, with 2.5-mm detector row collimation, at a pitch of 6. The first and second arterial phases were performed during a single breath hold. One reader, blinded to the results of the angiography, reviewed the first arterial phase images on a cine display to assess hepatic arterial anatomy. Visualization of the portal vein and its branches was assessed by using second arterial and portal venous phase images.
RESULTS: Major arterial trunks (celiac, hepatic, superior mesenteric, and left gastric) were depicted in all cases. Visualization of small arteries was as follows: right and left hepatic, 62 (100%) of 62; middle hepatic, 52 (87%) of 60; cystic, 47 (90%) of 52; right gastric, 50 (89%) of 56; and right and left inferior phrenic, 57 (92%) and 55 (89%) of 62, respectively. Subsegmental or more peripheral branches of the portal vein were depicted in 83% of cases during the second arterial phase and in 96% during the portal phase. There was no difference in degree of visualization in these two phases.
CONCLUSION: Multi–detector row CT angiography was able to depict the hepatic vascular anatomy.
Index terms: Computed tomography (CT), angiography, 95.12916 • Computed tomography (CT), helical, 95.12915 • Computed tomography (CT), maximum intensity projection, 95.12919 • Hepatic arteries, CT, 95.12916 • Hepatic veins, CT, 95.12916
| INTRODUCTION |
|---|
| Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
|---|
Helical multi–detector row CT can be used to acquire multiple CT data sets with each rotation of the x-ray tube (14) and to scan through large anatomic areas 3 to 7 times faster than can single–detector row helical CT scanners. This system, therefore, enables us to obtain thinner section collimation with a shorter acquisition time than does a single-section helical CT scanner. Findings of a recent study (15) in which multi–detector row CT was used demonstrated that maximum tumor-to-liver contrast for hypervascular primary and metastatic neoplasms occurred during the second pass, when dual arterial phases were acquired during a single breath hold. In our institution, after the introduction of multi–detector row CT, triphasic dynamic studies for upper abdominal disease, which include two separate sets of CT images of the liver in the time generally regarded as the hepatic arterial dominant phase during a single breath hold, and portal venous phase images have been routinely performed. When reviewing these triphasic dynamic studies, we noted that the small branches of the hepatic arteries could often be depicted, without portal venous enhancement, during the first arterial phase. We also noted that the peripheral branches of the portal vein were depicted during the second arterial phase, without hepatic venous enhancement. We therefore postulated that if the hepatic vasculature could be demonstrated separately at multi–detector row CT, the vasculature could potentially be used for the planning of surgery or interventional procedures and replace conventional catheter angiography.
The purpose of this study was to determine by using multi–detector row CT, in a triphasic hepatic dynamic study that included single breath-hold dual-arterial phase acquisition, the accuracy and frequency of visualization of the small hepatic arterial and portal venous anatomy with angiographic correlation.
| MATERIALS AND METHODS |
|---|
| Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
|---|
The interval between CT and conventional angiography ranged from -48 to 92 days (mean, 15.2 days). No patients underwent surgical or interventional procedures between the CT examination and angiography. All the patients included in this study provided informed consent. This study followed the Declaration of Helsinki principles.
Among the 62 patients, cholecystectomy was performed in 10 patients, subtotal or total gastrectomy in six, splenectomy in four, and left hepatic lobectomy in two.
CT Imaging Technique
At our institution, triphasic helical CT imaging is the standard examination for the diagnosis and staging of pancreaticobiliary tumors and quadraphasic helical CT (triphasic helical scanning plus equilibrium phase scanning) for suspected malignant hepatic tumors. All helical CT scans were obtained by using a LightSpeed QX/i scanner (GE Medical Systems, Milwaukee, Wis).
First, precontrast helical scans were obtained at 120 KVp, 270 mA, a reconstructed section thickness of 5 mm, a detector row configuration of 5 mm, and a pitch of 3 (high-quality mode), from the dome of the liver to 20 cm below it. A 512 x 512 matrix was used for all studies.
Next, a 20-gauge plastic intravenous catheter (Supercath; Top, Tokyo, Japan) was placed in an antecubital vein. The line was then connected to a power injector (Autoenhance A-50; Nemotokyorindou, Tokyo, Japan). The scanning time delay was determined by using a test bolus (15 mL at 5 mL/sec) of contrast medium followed by a series of single-level CT scans at low dose (120 KVp, 10 mA). The scan location was 20 cm below the dome of the liver, and the monitoring scans were acquired every 2 seconds from 10 to 40 seconds. A cursor (10–15 mm2) was placed by one of the radiologists (S.T., M.T., T.K., M.H.) over the abdominal aorta at this level, and the time to peak aortic enhancement was used to determine the scanning delay for the first arterial phase images. The detector configuration was 4 x 2.5 mm in the interspaced high-speed mode in which four interspaced helical data sets were collected from eight 1.25-mm detector rows. The HS mode is equivalent to a pitch of 6 with the table speed set at 15 mm per rotation. One rotation of the x-ray tube was 0.8 second.
Then, all patients received low osmolarity contrast medium (Omnipaque; Daiichi Pharmaceutical, Tokyo, Japan; 300 mg of iodine per milliliter) at a rate of 5 mL/sec through a 20-gauge plastic intravenous catheter by using the power injector. The volume of contrast medium delivered was 2 mL per kilogram of body weight. Patient weights ranged from 34.5– 85.0 kg (mean, 58.2 kg). Therefore, the volume of contrast medium administered ranged from 69 to 170 mL (mean, 113 mL).
Scanning began at the dome of the liver (location determined by using a scout digital radiograph) and proceeded in a caudal direction for 10.5 seconds, covering a z-axis distance of 20 cm. These CT images constituted the first arterial phase. After an interscan delay of 5 seconds for table movement, scanning resumed from the dome of the liver in a caudal direction for 10.5 seconds. This constituted the second arterial phase. The total acquisition time was 26 seconds and was accomplished in a single breath hold. The average scanning delay for the first arterial phase was 19 seconds (range, 10–36 seconds), while the average delay for the second arterial phase was 34 seconds (range, 25–51 seconds). The portal venous phase was performed 20 seconds after the end of the second arterial phase. In cases with suspected malignant hepatic tumor, an equilibrium phase was also performed 120 seconds after the end of the portal venous phase, but these scans were not evaluated for this study.
The first and second arterial and the portal venous phase helical CT data were retrospectively reconstructed with a standard soft algorithm at 1.25-mm increments, with a 2.5-mm section thickness and a 30-cm field of view. The data were then transferred to a workstation (Advantage Windows 3.1; GE Medical Systems) and displayed as a cine loop.
Conventional Angiographic Technique
Digital subtraction angiography was performed by using two standard angiography units (Advantx; GE Medical Systems, or Integris V3000; Philips Medical System, Best, the Netherlands). All 62 patients underwent catheter angiography of the celiac and superior mesenteric artery, with a transfemorally inserted 4- or 5-F catheter. At each artery, 15–25 mL of iodinated contrast material was administered with an injection rate of 4–7 mL/sec. In most cases, selective arteriograms of the common hepatic and splenic arteries were obtained, but these images were not evaluated for the study.
Image Assessment
At the workstation, one radiologist (S.T.), who was blinded to the results of the angiography, assessed the data by viewing the cine loop. The observer also reconstructed the first arterial phase images by using an oblique thick-slab target maximum intensity projection (MIP) technique to obtain the CT arteriogram. When reconstructing the CT arteriogram, four images (transverse, reformatted sagittal, reformatted coronal, and target MIP images) were simultaneously revealed on the display of the workstation. This technique helped the observer to follow the course of the vessels. The radiologist then recorded the frequency of visualization of the major visceral arteries (celiac, hepatic, splenic, gastroduodenal, left gastric, and superior mesenteric). The frequency of visualization and the origin of the small arteries (inferior phrenic, right gastric, cystic, left, middle, and right hepatic) also were recorded. The arterial anatomy as described by Michels (16) was used as the reference for the estimation of the first arterial phase helical CT scans.
One radiologist (M.T.) assessed the celiac and superior mesenteric arteriogram to record the visualization and the origin of the major visceral arteries (celiac, hepatic, splenic, gastroduodenal, left gastric, and superior mesenteric) and the small arteries (inferior phrenic, right gastric, cystic, left, middle, and right hepatic). If a patient had undergone surgery in which an artery might have been removed, these arteries were excluded from evaluation; 10 cystic, four right gastric, two left hepatic, and two middle hepatic arteries were excluded. Although four patients had undergone splenectomy, proximal portions of the splenic arteries were depicted on angiograms in these patients. Therefore, all of the splenic arteries were evaluated. These results were compared with the results of angiography.
The degree of the portal venous enhancement at the first arterial phase was graded according to the following: absent enhancement; minimal enhancement of the trunk of the portal vein; adequate enhancement of the trunk of the portal vein, right and/or left portal vein, and second branch of the portal vein or more.
Both of the images of the second arterial and portal venous phase were reconstructed by a radiologist (S.T.) by using a target MIP technique to depict the portal vein and the hepatic vein. Target MIP images were reviewed by two radiologists (T.M., T.K.) at the same time. The degree of visualization of the portal vein and its intrahepatic branches was determined with a consensus agreement and recorded. The most distal branches demonstrated with target MIP images were graded by using the following scale: 0, portal vein not visualized; 1, visualization up to the trunk of the portal vein; 2, visualization up to right and/or left portal vein; 3, visualization up to segmental branches; 4, visualization up to subsegmental branches; 5, visualization of up to the first branches of subsegmental branches; and 6, visualization of second or more branches of subsegmental branches. Overall image quality was also evaluated according to the following: 1, nondiagnostic; 2, fair; 3, good; and 4, excellent. Statistical analyses for qualitative rating were made by using the Wilcoxon signed rank test. A P value of less than .05 was considered to indicate a statistically significant difference.
The CT attenuation numbers of the hepatic parenchyma and the portal vein were measured at the level of the porta hepatis on both the second arterial and portal venous phase images. A region of interest was carefully drawn by a radiologist (S.T.) to encompass as much of the lumen as possible for the portal vein (10–18 mm2). The CT number of the hepatic parenchyma was measured by using a region of interest (38–56 mm2) in areas devoid of focal changes of attenuation, such as cysts and vessels. Statistical analyses were made by using the paired t test. A two-tailed P value of less than .05 was considered to indicate a statistically significant difference.
| RESULTS |
|---|
| Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
|---|
|
|
|
|
|
|
Cholecystectomy was performed in 10 of the 62 cases, and only 52 cases were analyzed for visualization of the cystic artery. In these cases, the cystic artery was accurately detected at CT in 47 (90%) of 52 cases.
Subtotal or total gastrectomy was performed in six of 62 cases. The observer failed to detect the right gastric artery in three of the remaining 56 cases, while angiography failed to demonstrate the right gastric artery in four of them. Two of three right gastric arteries were not demonstrated at CT and were also not demonstrated at angiography. In three cases, the supraduodenal artery was misinterpreted as the right gastric artery. Although 50 (89%) of 56 right gastric arteries were detected at CT (Fig 4a), the origin of the right gastric artery was incorrectly assessed in seven cases. These seven cases included three right gastric arteries arising from the proper hepatic artery, two from the gastroduodenal artery, one from the common hepatic artery, and one from the middle hepatic artery. When the right gastric artery arose adjacent to the proper hepatic and gastroduodenal bifurcation, it was difficult to determine the correct origin of the right gastric artery.
|
|
|
|
|
Portal Venous Visualization
Subsegmental or more peripheral branches of the portal vein were seen in 55 (89%) of 62 cases during the second arterial phase and in 60 (97%) cases during the portal venous phase. On the other hand, Sub-subsegmental branches of the portal vein were more frequently demonstrated on second arterial phase (47 of 62 cases) than on portal venous phase (36 of 62 cases) (Fig 6a, 6b) images. However, the second arterial phase failed to demonstrate the segmental branches of the portal vein in seven cases and in two portal venous phase cases. In one of seven cases, in which the second arterial phase failed to demonstrate the segmental branches of the portal vein, a prior splenectomy was performed. In another case, transjugular portosystemic shunting was performed. In the remaining five cases there was no history of abdominal surgery. The mean score of most distal branches visualized on the target MIP images was 5.0 ± 1.2 (mean ± SD) for the second arterial phase and 4.7 ± 0.9 for the portal venous phase. The difference was statistically significant (P = .03; Wilcoxon signed rank test). However, the mean score of overall image quality of the target MIP images was 3.7 ± 0.6 for the second arterial phase and 3.2 ± 0.8 for the portal venous phase. The difference was not statistically significant (P = .82; Wilcoxon signed rank test).
|
|
|
|
| DISCUSSION |
|---|
| Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
|---|
After the introduction of multi-detector row helical CT, the usefulness of triphasic dynamic studies for liver, including dual-arterial phase acquisition during a single breath hold, for depicting hypervascular hepatic tumor has been reported (15). However, to our knowledge, no report has described the capability of multi–detector row CT for depicting hepatic vasculature. Our results showed that the hepatic arteries were well identified at the first arterial phase. There have been a few reports (12,13) that described the usefulness of CT angiography with a single helical CT scanner for evaluating the hepatic arterial anatomy. Patient populations of these studies were transplantation candidates, and the authors evaluated only the basic types of hepatic arterial blood supply (12,13). Since conventional catheter angiography had not been performed for use as an anatomic reference standard in all cases in these previous studies, it was impossible to determine the accuracy of CT for classifying the hepatic arterial blood supply patterns (12,13). In our study, by using conventional angiography as the standard, the basic types of hepatic arterial blood supply patterns are correctly assessed in 98% of cases. Only one accessory left hepatic artery arising from the left gastric artery was missed.
Little attention has been paid to the small branches of the hepatic artery, such as the middle hepatic artery, the cystic artery, and the right gastric artery at CT. However, it could be potentially important to evaluate these small arteries for interventional procedures, such as transarterial chemotherapy and transarterial embolization. On the basis that two-thirds (43 out of 62) of our patient population have hepatocellular carcinoma, most of them were candidates for transarterial chemotherapy or transarterial embolization. Foley et al (15) used a technique similar to ours for the evaluation of hypervascular hepatic neoplasms and reported that the second arterial phase was useful for lesion detection. Their results suggest that our protocol was also useful for depicting hypervascular hepatic neoplasms. So, our protocol may be useful for a detailed assessment of hepatic arterial anatomy as well as for detection of hypervascular lesions simultaneously.
Previous investigators (17–20) who have assessed the portal venous system by using biphasic helical CT have used a delay time of approximately 1 minute. Because of the limitation of the scanning speed of a single-section helical CT scanner, the second phase has to begin approximately 1 minute after the initiation of contrast medium injection, when a biphasic dynamic study has been performed.
However, our results show that the second arterial phase was better than the portal venous phase in demonstrating the intrahepatic branches of the portal vein in most cases. Kim et al (21) reported that the hepatic parenchyma started to enhance approximately 16 seconds after the CT number of the aorta reached 100 HU with an injection rate of 5 mL/sec. Therefore, the second arterial phase in our study corresponds in timing to the start of enhancement of the hepatic parenchyma. The portal venous phase in our study corresponds in timing to the portal venous phase of a biphasic hepatic helical scan with a single-section CT system. Since the enhancement of the portal vein decreases in the portal venous phase as compared with that in the second arterial phase, larger differences between the CT number of the portal vein and that of the hepatic parenchyma were obtained in the second arterial phase.
CT hepatic venography could be performed by using the images from the portal venous phase. Kamel et al (22) reported that multi–detector row CT was useful for depicting the hepatic venous anatomy. In our study, the target MIP images obtained during the portal venous phase also demonstrated, very well, the hepatic vein. However, as it is difficult to obtain an anatomic reference standard for the hepatic vein, we did not evaluate the hepatic vein in this study.
In conclusion, multi–detector row CT in a triphasic hepatic dynamic study, including a single-breath hold dual-arterial phase acquisition, was useful for the depiction of the small hepatic arterial and portal venous anatomy. We propose that the first arterial phase should be used for CT arteriography; the second arterial phase, for CT portography; and the portal venous phase, for CT venography.
| ACKNOWLEDGMENTS |
|---|
| REFERENCES |
|---|
| Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
|---|