DOI: 10.1148/radiol.2291020582
(Radiology 2003;229:81-90.)
© RSNA, 2003
Pancreatic Malignancy: Value of Arterial, Pancreatic, and Hepatic Phase Imaging with Multi–Detector Row CT1
Joel G. Fletcher, MD,
Maurits J. Wiersema, MD,
Michael A. Farrell, MD,
Jeff L. Fidler, MD,
Lawrence J. Burgart, MD,
Takashi Koyama, MD,
C. Daniel Johnson, MD,
David H. Stephens, MD,
Ellen M. Ward, MD and
W. Scott Harmsen, MS
1 From the Department of Radiology (J.G.F., M.A.F., J.L.F., T.K., C.D.J., D.H.S., E.M.W.), Division of Gastroenterology and Hepatology, Departments of Internal Medicine (M.J.W.) and Laboratory Medicine and Pathology (L.J.B.), and Division of Biostatistics (W.S.H.), Mayo Clinic Rochester, 200 First St SW, Mayo E-2 B, Rochester, MN 55905. From the 2001 RSNA scientific assembly. Received May 15, 2002; revision requested July 12; final revision received January 2, 2003; accepted February 24. Address correspondence to J.G.F. (e-mail: fletcher.joel@mayo.edu).
 |
ABSTRACT
|
|---|
PURPOSE: To assess the value of arterial, pancreatic, and hepatic phase imaging at multi–detector row computed tomography (CT) of the pancreas for pancreatic malignancy.
MATERIALS AND METHODS: Thirty-nine patients suspected of having resectable pancreatic adenocarcinoma underwent triple-phase multi–detector row CT. Images obtained during each phase were interpreted by one radiologist who evaluated presence of tumor, vascular invasion, and flow artifacts in the superior mesenteric vein and measured attenuation of tumor, normal pancreas, aorta, and superior mesenteric vein. Results were compared with histologic, follow-up, and correlative imaging findings.
RESULTS: Mean tumor-to-gland attenuation difference was greatest on images obtained in the pancreatic phase (42 HU) versus that on those obtained in the hepatic phase (35 HU) and in the arterial phase (25 HU). For tumor detection, sensitivity of the images obtained in pancreatic (0.97 [29 of 30]) and hepatic (0.93 [28 of 30]) phases was superior to that of those obtained in arterial phase (0.63 [19 of 30]) (P 
.008). For vascular invasion detection, sensitivity of images obtained in the hepatic phase (0.83) was better than that of those obtained in the pancreatic (0.58) and arterial (0.25) phases. Images obtained in the pancreatic phase demonstrated more flow artifacts and decreased attenuation in the superior mesenteric vein, compared with the artifacts revealed on images obtained in the hepatic phase.
CONCLUSION: Routine acquisition of images in the arterial phase is unnecessary for detection of pancreatic adenocarcinoma. Images of the pancreas obtained in the hepatic phase with multi–detector row CT most accurately display vascular invasion.
© RSNA, 2003
Index terms: Computed tomography (CT), multi–detector row, 77.1211, 77.12113 • Computed tomography (CT), phase imaging, 77.1211, 77.12113 • Pancreas, neoplasms, 77.32 • Pancreas, CT, 77.1211
|
INTRODUCTION
|
|---|
Biphasic computed tomography (CT) of the pancreas has been the accepted technique for the detection and staging of pancreatic ductal adenocarcinoma (1–3). Most authors advocate that the initial phase of the biphasic examination of the pancreas should be the pancreatic phase (1,2). Use of the pancreatic phase allows optimization of tumor conspicuity, with maximization of tumor-to-pancreas attenuation differences, and allows adequate mesenteric venous and arterial opacification for detection of vascular invasion (1). A few authors advocate that the initial phase of the biphasic examination should be an earlier arterial phase (3,4). These authors believe that images obtained during the arterial phase allow increased tumor conspicuity and good visualization of the peripancreatic arterial supply, demonstrate pancreatic enhancement, and are superior to delayed-phase images (3,4). Others found that images obtained in the arterial phase were unhelpful (5,6). General agreement exists that the second and final phase of the biphasic pancreatic CT examination, the hepatic (or portal) phase, allows optimization of the detection of liver metastases and provides a second look at the tumor.
Multi–detector row CT permits faster scanning of the abdomen and allows acquisition of images in all three phases of vascular enhancement, that is, the arterial, pancreatic, and hepatic phases (7). The scanning speed of multi–detector row CT allows the acquisition ofimages in these phases at a narrow section thickness and thus improves spatial resolution. Lesion contrast is potentially increased not only because of less volume averaging and motion but also a shorter scanning time for obtaining images during each vascular phase. Thus, the images obtained within each phase are more homogeneous with respect to parenchymal and vascular enhancement. These potential advantages may lead to improvements in the detection and staging of pancreatic cancer, but there are likely to be other unanticipated differences between multiphasic examinations performed with single– and multi–detector row CT scanners. For example, McNulty et al (7) discovered that tumor-to-pancreas attenuation differences on images obtained in the pancreatic phase were not as pronounced as those described by authors of earlier studies who performed single–detector row CT. It may be that the timing of the traditional vascular phases must be altered as a consequence of the faster scanning speed of multi–detector row CT.
We recently performed an evaluation in which we compared the performance of triple-phase multi–detector row CT, gadolinium-enhanced magnetic resonance (MR) imaging, and endoscopic ultrasonography (US) in patients with potentially resectable pancreatic adenocarcinoma (8). In that study, we did not examine the relative contribution of each phase of CT enhancement to decision making regarding clinical end points, such as detection of tumor and of vascular invasion. The routine acquisition and review of images obtained during multiple vascular phases is warranted only if each vascular phase contributes to the diagnosis of tumor presence and the staging. The acquisition and review of superfluous images obtained during additional phases leads to an increase in radiation dose to patients, in the radiologist’s interpretation time, and in machine costs, as well as to a proliferation of images. The purpose of our study was to assess the value of arterial, pancreatic, and hepatic phase imaging at multi–detector row CT of the pancreas for pancreatic malignancy.
|
MATERIALS AND METHODS
|
|---|
Patients
Triple-phase multi–detector row CT of the pancreas was performed as part of a study in which CT, MR imaging, and endoscopic US were compared in regard to the diagnosis and staging of potentially resectable pancreatic cancer, as described elsewhere (8). Informed consent was obtained in 43 patients prior to performance of triple-phase multi–detector row CT, according to a protocol approved by our institutional review board. Four of these patients erroneously underwent a biphasic examination of the pancreas and were excluded from our analysis. The mean age of the patients was 69 years (age range, 50–87 years), with 21 men and 18 women. Inclusion criteria were that the patient was suspected of having a pancreatic adenocarcinoma on the basis of findings at a prior imaging examination but that there was no evidence of unresectability and that the patient’s health status would not prevent surgery. Regional lymphadenopathy was not considered a surgical contraindication. Exclusion criteria included pregnancy, evidence of unresectability at prior scanning, age younger than 18 years, and comorbid conditions that precluded surgery. To ensure that only patients with potentially resectable tumors were included in our study population, findings of imaging examinations (38 CT, one US) performed at outside institutions were evaluated by a gastrointestinal radiologist at our institution (9). Of the 38 prior outside CT examinations, 37 (28 monophasic, nine biphasic) were performed with intravenously administered contrast material, with a mean section thickness of 6.5 mm ± 2.4 (SD). The scans were examined for evidence of unresectability, specifically noting the presence of vascular invasion (ie, celiac axis, superior mesenteric artery, or portal vein or superior mesenteric vein), distant metastases, and retroperitoneal or gastric invasion. Only patients without these signs of unresectability were included in the study.
Triple-Phase Multi–Detector Row CT Protocol
For triple-phase multi–detector row CT, we used a varying scale for the injection rate and timing of each phase of enhancement. Because the timing and intensity of maximal pancreatic and hepatic parenchymal enhancement and surrounding vascular structures is known to vary with the injection rate (10–12), we used a sliding scale that was proposed by Charnsangavej (13) for the timing of each vascular phase. A dedicated radiology nurse who placed the intravenous catheter in each patient determined the maximum possible injection rate on the basis of his or her experience with catheter size and venous access. The timing of each phase of enhancement was varied on the basis of this maximum injection rate, according to Table 1 (8,13).
Triple-phase multi–detector row CT was performed with CT scanners (LightSpeed Plus and LightSpeed QX/I; GE Medical Systems, Milwaukee, Wis). Patients drank 500 mL of water 20–30 minutes prior to scanning and an additional 250 mL immediately prior to scanning. Triple-phase multi–detector row CT of the pancreas was then performed with a section thickness and reconstruction interval of 2.5 mm each for each phase of enhancement. A 24-cm field of view, high-speed mode, 0.8-second tube rotation time, 15.0 mm per rotation table speed, 140 kVp, and 200 mA were used for acquisition of images during the arterial and pancreatic phases. A field of view to fit, high-quality mode, 0.8-second tube rotation time, 7.5 mm per rotation table speed, 140 kVp, and 190 mA were used for acquisition of images during the hepatic phase (8). With the assumption that the pancreas covers 9 cm in craniocaudal dimension and that the abdomen covers 16 cm, the acquisition of images in the arterial and pancreatic phases with this protocol requires 4.8 seconds for each phase and that of images in the hepatic phase requires 17 seconds (Fig 1).
View larger version (18K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 1. Graph shows timing of different phases in pancreatic CT protocols used by investigators. Assumptions include 9 cm of craniocaudal coverage for the pancreas (during the arterial and pancreatic phases) and 16 cm for the abdomen (during the hepatic phase). When a range of table speeds was available, the median table speed was used, with the assumption of a 4 mL/sec contrast material injection (*). In this study, during the hepatic phase only the pancreas (**) was imaged. Gray rectangles represent arterial phase; white rectangles, pancreatic phase; black rectangles, hepatic phase.
|
|
Image Interpretation
Because we sought to determine the clinical contribution of each phase, each phase of enhancement was interpreted by a different radiologist (J.G.F., M.A.F., J.L.F., T.K.). The radiologists had a similar level of expertise, with all having completed subspecialized fellowship training in gastrointestinal radiology, followed by 2–7 years of experience as an attending radiologist. In each patient, images obtained during the three phases of enhancement were interpreted by three of the four radiologists in a nonrandom fashion, with no radiologist interpreting images obtained in more than one phase of enhancement in any patient. Although an alternative method would have been to have one radiologist interpret images obtained during all three phases in the same patient, we thought that the idiosyncratic appearance of many of these small tumors would introduce recall bias, even if the interpretations of images obtained in each phase were separated by long periods of time. Radiologists were blinded to the history, the outside imaging examination findings, and the phases of enhancement that were not being interpreted, but they were aware of the inclusion criteria for our study.
The radiologists noted the presence of a pancreatic tumor. Masses not clearly neoplastic could be interpreted as indeterminate in nature. The radiologists also assessed the presence of vascular invasion (ie, of the celiac axis or the superior mesenteric artery or of the superior mesenteric vein or the portal vein). Although patients in our study had prior imaging examination findings that suggested a pancreatic mass without local vascular invasion, we hypothesized that thin-section multi–detector row CT for imaging of the pancreas may demonstrate evidence of vascular invasion not seen at these prior, largely screening, CT examinations. Criteria for arterial invasion constituted encasement, vessel margin irregularity, or tumor incursion into the periarterial fat plane with the tumor lying in juxtaposition to the vessel (14). Venous invasion was considered present when the tumor caused venous occlusion, flattening or narrowing, apposition with concavity toward the vessel lumen, or circumferential apposition greater than 180° (14–17). Each radiologist also noted tumor size, number of sections in which the tumor appeared, and biliary and pancreatic ductal involvement seen in each phase.
Image analysis and attenuation measurements were performed in each phase by the interpreting radiologist with a computer workstation (Advantage Windows 3.1 or 4.0; GE Medical Systems). However, for one patient with pancreatic adenocarcinoma, the images were inadvertently not placed on our picture archiving and communication system and were lost for retrieval. Hard-copy images were used for interpretation in this latter patient. Regions of interest were placed over the tumor, the normal pancreatic parenchyma, the aorta, and the superior mesenteric vein (at the level of the uncinate process), and measurements were determined. Readers were asked to subjectively assess whether the superior mesenteric vein was filled homogeneously or heterogeneously with contrast material (because flow artifacts were present). The radiologist who reviewed images obtained during the hepatic phase also determined measurements of regions of interest that were placed over the portal vein at the liver hilum, the hepatic parenchyma, and the largest hepatic vein near its confluence with the inferior vena cava. Circular regions of interest placed over all structures were created to be as large as possible, and the radiologists were specifically instructed to avoid measurements of regions of interest placed over any cystic regions within tumors. Tumor attenuation data and tumor-pancreatic attenuation differences were reported only for those patients who had pancreatic ductal adenocarcinoma.
Histologic and Cytologic Analysis
The diagnosis of tumor presence was determined with histologic or cytologic analysis for malignant tumors (32 patients). One 77-year-old patient with typical imaging findings of a benign serous cystadenoma at both MR imaging and endoscopic US underwent clinical follow-up at 24 months. The diagnosis of absence of tumor was determined on the basis of negative biopsy findings when a mass was identified (one patient) and on the basis of clinical follow-up findings at 12 months that included regression or stability at repeat imaging when a mass was not identified (five patients). The standard for vascular invasion was the histologic specimen when available (n = 15), the surgical assessment (n = 2), or the agreement with findings of a confirmatory test (MR imaging or endoscopic US, n = 15). A standard for tumor presence was available in all patients. A standard for vascular invasion was present in all but one patient, and this latter patient did not undergo resection (ie, because of liver metastases) and had CT findings that demonstrated vascular invasion but endoscopic US findings that did not. This patient was not included in the statistical analysis for vascular invasion. No effort was made to analyze distant metastases because we did not include the entire liver or abdomen on images obtained in the arterial and pancreatic phases.
Statistical Analysis
Normally distributed continuous data were presented as the mean ± SD. Comparison of attenuation differences on images obtained in different phases of contrast enhancement was performed with the paired Student t test. Sensitivity was defined as the number of true-positive imaging results divided by the sum of true-positive and false-negative results. Specificity was defined as the number of true-negative imaging results divided by the sum of the true-negative and false-positive results. Examinations with findings that were indeterminate for tumor presence were counted as false-negative when tumor was present and false-positive when tumor was absent. Accuracy was the sum of the true-positive and true-negative results divided by the number of all patients. Proportions were reported with 95% exact binomial CIs. The comparisons of sensitivity, specificity, and accuracy between any two phases within patients was performed with the sign test, an exact method for paired data. We assumed a priori that the performance of the four radiologists would be similar, given their level of training and experience. To test this assumption, we used the 
2 test to assess the equality of the sensitivity estimates together and of the specificity estimates separately among the four reviewing radiologists. We used this test to determine whether there was a significant difference in performance among the four radiologists for interpretation of images obtained during any phase of enhancement. Comparison of the proportions in patients with maximal contrast during the respective phases of enhancement was performed with the 2 test. Two-tailed P values of .05 or less were considered to indicate a significant difference, with Bonferroni correction used when multiple comparisons were performed.
|
RESULTS
|
|---|
There were 33 tumors in 39 patients: 30 were pancreatic ductal adenocarcinomas, one was an invasive ampullary carcinoma, one was cholangiocarcinoma of the distal common bile duct, and one was serous cystadenoma. Four patients had normal glands, and two had focal chronic pancreatitis. The mean tumor diameter was 2.6 cm ± 0.5, and the range was 1.7–4.0 cm. Table 2 shows the results of interpretation of images in each phase of enhancement separately for detection of a pancreatic neoplasm, with uncertain diagnoses being considered in error (ie, false-positive or false-negative examination results). For detection of only pancreatic ductal adenocarcinoma, the sensitivity during the arterial, pancreatic, and hepatic phases was 0.63 (19 of 30), 0.97 (29 of 30), and 0.93 (28 of 30). Sensitivity for detection of pancreatic adenocarcinoma during the pancreatic and hepatic phases was significantly greater than was the sensitivity during the arterial phase (P = .002, pancreatic phase; P = .008, hepatic phase).
Images obtained in both the pancreatic and hepatic phases failed to demonstrate an ampullary tumor in any patient. In one patient, a diagnosis of pancreatic adenocarcinoma was considered indeterminate on images obtained in the pancreatic phase, and in two patients, a diagnosis of this tumor was considered indeterminate on images obtained in the hepatic phase. These three interpretations occurred in different patients, so potentially there could have been no false-negative examinations for pancreatic adenocarcinoma if readers had been allowed to examine images obtained in both of these phases of vascular enhancement. There were more false-positive findings in examinations (ie, one in which tumor was thought to be present and three in which the diagnosis was considered indeterminate) performed in the pancreatic phase than in those performed in the other phases. The readers of images obtained in the arterial and hepatic phases correctly identified the absence of tumor in four of six patients.
The mean attenuation of the pancreas in all patients was greatest in the pancreatic phase at 107 HU ± 30 versus 65 HU ± 23 (P < .0001) in the arterial phase and 98 HU ± 29 (P = .05) in the hepatic phase. In 27 (71%) of 38 patients, the pancreatic enhancement was greatest in the pancreatic phase, whereas in 11 (29%), the pancreatic enhancement was greatest in the hepatic phase.
The mean pancreatic adenocarcinoma tumor attenuation increased steadily within each phase, as demonstrated in Figure 2, which graphically shows the relationship between tumor and gland enhancement. The mean absolute tumor-to-gland attenuation difference was greatest on images obtained in the pancreatic phase at 42 HU ± 21 (P = .003) and in the hepatic phase at 35 HU ± 25 (P = .009) when each of these values was compared with that in the arterial phase at 25 HU ± 22 (Fig 3). Maximal attenuation differences did not reliably occur on images obtained in the pancreatic phase. Nineteen patients had maximal tumor-to-gland enhancement on images obtained in the pancreatic phase; eight, on images obtained in the hepatic phase; one, on those obtained in the arterial phase; and one, on those obtained at equal maximal enhancement in the pancreatic and hepatic phases. Five patients in whom the tumor-to-gland attenuation difference was less than 10 HU on images obtained in the hepatic phase had mean tumor-to-gland attenuation differences of 43 HU on images obtained in the pancreatic phase. Despite the clear lack of an attenuation difference, all of these isoattenuating tumors were still detected by readers of images obtained during the hepatic phase because of other signs of tumor presence (eg, mass effect, ductal cutoff). There were no patients with tumor-to-gland attenuation differences of less than 10 HU on images obtained in the pancreatic phase, but five patients had attenuation differences that were between 10 and 16 HU. The number of patients with small tumor-to-gland attenuation differences on images obtained in the arterial phase was difficult to quantify because so many of the tumors were not depicted.
View larger version (140K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3a. Transverse CT images in 65-year-old man with pancreatic ductal adenocarcinoma in the pancreatic head, which was detected by readers on images obtained in arterial, pancreatic, and hepatic phases. (a) Image of tumor (arrow) obtained in arterial phase, with tumor-to-gland attenuation difference of 16 HU. (b) Image of tumor (arrow) obtained in pancreatic phase, with tumor-to-gland attenuation difference of 64 HU. (c) Image of tumor (arrow) obtained in hepatic phase, with tumor-to-gland attenuation difference of 46 HU. (d) Image obtained in pancreatic phase slightly cephalad to the tumor demonstrates flow artifact (arrow) in the superior mesenteric vein. (e) Image obtained in hepatic phase at same level as in d demonstrates homogeneous distribution of contrast material in the superior mesenteric vein (arrow), which is hyperattenuating, compared with the adjacent pancreas.
|
|
View larger version (138K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3b. Transverse CT images in 65-year-old man with pancreatic ductal adenocarcinoma in the pancreatic head, which was detected by readers on images obtained in arterial, pancreatic, and hepatic phases. (a) Image of tumor (arrow) obtained in arterial phase, with tumor-to-gland attenuation difference of 16 HU. (b) Image of tumor (arrow) obtained in pancreatic phase, with tumor-to-gland attenuation difference of 64 HU. (c) Image of tumor (arrow) obtained in hepatic phase, with tumor-to-gland attenuation difference of 46 HU. (d) Image obtained in pancreatic phase slightly cephalad to the tumor demonstrates flow artifact (arrow) in the superior mesenteric vein. (e) Image obtained in hepatic phase at same level as in d demonstrates homogeneous distribution of contrast material in the superior mesenteric vein (arrow), which is hyperattenuating, compared with the adjacent pancreas.
|
|
View larger version (140K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3c. Transverse CT images in 65-year-old man with pancreatic ductal adenocarcinoma in the pancreatic head, which was detected by readers on images obtained in arterial, pancreatic, and hepatic phases. (a) Image of tumor (arrow) obtained in arterial phase, with tumor-to-gland attenuation difference of 16 HU. (b) Image of tumor (arrow) obtained in pancreatic phase, with tumor-to-gland attenuation difference of 64 HU. (c) Image of tumor (arrow) obtained in hepatic phase, with tumor-to-gland attenuation difference of 46 HU. (d) Image obtained in pancreatic phase slightly cephalad to the tumor demonstrates flow artifact (arrow) in the superior mesenteric vein. (e) Image obtained in hepatic phase at same level as in d demonstrates homogeneous distribution of contrast material in the superior mesenteric vein (arrow), which is hyperattenuating, compared with the adjacent pancreas.
|
|
View larger version (141K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3d. Transverse CT images in 65-year-old man with pancreatic ductal adenocarcinoma in the pancreatic head, which was detected by readers on images obtained in arterial, pancreatic, and hepatic phases. (a) Image of tumor (arrow) obtained in arterial phase, with tumor-to-gland attenuation difference of 16 HU. (b) Image of tumor (arrow) obtained in pancreatic phase, with tumor-to-gland attenuation difference of 64 HU. (c) Image of tumor (arrow) obtained in hepatic phase, with tumor-to-gland attenuation difference of 46 HU. (d) Image obtained in pancreatic phase slightly cephalad to the tumor demonstrates flow artifact (arrow) in the superior mesenteric vein. (e) Image obtained in hepatic phase at same level as in d demonstrates homogeneous distribution of contrast material in the superior mesenteric vein (arrow), which is hyperattenuating, compared with the adjacent pancreas.
|
|
View larger version (145K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3e. Transverse CT images in 65-year-old man with pancreatic ductal adenocarcinoma in the pancreatic head, which was detected by readers on images obtained in arterial, pancreatic, and hepatic phases. (a) Image of tumor (arrow) obtained in arterial phase, with tumor-to-gland attenuation difference of 16 HU. (b) Image of tumor (arrow) obtained in pancreatic phase, with tumor-to-gland attenuation difference of 64 HU. (c) Image of tumor (arrow) obtained in hepatic phase, with tumor-to-gland attenuation difference of 46 HU. (d) Image obtained in pancreatic phase slightly cephalad to the tumor demonstrates flow artifact (arrow) in the superior mesenteric vein. (e) Image obtained in hepatic phase at same level as in d demonstrates homogeneous distribution of contrast material in the superior mesenteric vein (arrow), which is hyperattenuating, compared with the adjacent pancreas.
|
|
Readers also recorded the size and number of sections on which tumor or an indeterminate mass appeared. Although readers were able to collect these data for all patients with images obtained in the pancreatic and hepatic phases, readers of images obtained in the arterial phase did not record this information in six patients (ie, no tumor was identified or an indeterminate mass could not be accurately measured). When the readers examined pancreatic adenocarcinoma size as depicted during each phase of enhancement, there were no statistically significant differences between the phases or between the CT measurements and the size of the tumors that were resected. In those patients with arterial phase data regarding the number of sections that showed the tumor, the tumor or indeterminate mass was demonstrated in the arterial phase on a mean of 8.3 sections ± 3.4, compared with a mean of 9.4 sections ± 3.4 for the pancreatic phase and a mean of 7.9 sections ± 2.3 for the hepatic phase. In the six patients without definite imaging abnormalities on images obtained in the arterial phase, tumor was demonstrated on a mean of 7.3 sections ± 3.3 obtained in the pancreatic phase and on a mean of 5.7 sections ± 3.6 obtained in the hepatic phase. In 10 patients, images obtained in two or more phases equally displayed the tumor on the greatest number of sections. With images obtained in the pancreatic phase, tumor was demonstrated on more sections (20 [67%] of 30) than on images obtained in the arterial (eight [27%] of 30, P = .004) or the hepatic (12 [40%] of 30, P = .071) phases.
To assess the relative contribution of each phase to tumor staging, we measured the attenuation of the peripancreatic vessels, as has been determined previously, but we also compared the blinded interpretations for vascular invasion between radiologists who interpreted images obtained during the separate phases of enhancement. Table 3 shows the attenuation of the aorta and superior mesenteric vein in each phase of vascular enhancement. We measured attenuation in the aorta because we thought it reflected the attenuation of the superior mesenteric artery and of the arteries of the peripancreatic arcade. Table 3 shows that aortic enhancement is similar during the arterial and the pancreatic phases but diminishes in the hepatic phase. Enhancement in the superior mesenteric vein increases with each phase, in contradistinction to findings with single–detector row CT (1,2). In the majority of patients, higher attenuation was observed in the superior mesenteric vein on images obtained during the hepatic phase than was observed on those obtained during the pancreatic phase (24 [63%] of 38 vs 14 [37%] of 38, P = .02). In 32 (84%) of 38 patients, heterogeneous enhancement of the superior mesenteric vein was observed on images obtained in the pancreatic phase subsequent to observation of flow artifacts. However, in all 38 patients, homogeneous enhancement of the superior mesenteric vein was observed on images obtained in the hepatic phase (Fig 3). On images obtained in the hepatic phase, the mean attenuation in the portal and in the hepatic veins was 157 HU ± 33 and 157 HU ± 39, respectively, and these values were similar to the attenuation in the superior mesenteric vein. Both of these values were greater than the mean hepatic attenuation of 106 HU ± 15 (P < .001) on images obtained in the hepatic phase. Four (11%) of 38 patients had differences of less than 10 HU between the attenuation of the hepatic parenchyma and that of the hepatic veins.
Table 4 reports the sensitivity, specificity, and accuracy of detection of vascular invasion for the blinded interpretation of images obtained in each phase of enhancement, as well as the sensitivity of detection of arterial and venous invasion. Twelve patients had vascular invasion; included among them were three with arterial invasion, seven with venous invasion, and two with both. The sensitivity of images obtained in the hepatic phase for prediction of vascular invasion was significantly better than that of those obtained in the arterial phase (P = .016), and there was a trend toward increased sensitivity compared with that of images obtained in the pancreatic phase, although this trend did not indicate a significant difference (Figs 4, 5).
View larger version (161K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 4a. Transverse CT images in 68-year-old man with replaced right hepatic artery in juxtaposition to pancreatic adenocarcinoma in the posterior aspect of the pancreatic head, which indicates arterial invasion. (a) Image obtained in arterial phase on which the tumor was not identified with certainty, which likely led to misdiagnosis of lack of vascular invasion. Replaced right hepatic artery (arrow) and tumor (arrowheads) are seen. (b) Image obtained in pancreatic phase on which vascular invasion was correctly identified. Note increased conspicuity of the tumor (arrowheads) compared with the normal gland (black arrow). The low-attenuation tumor is clearly depicted adjacent to the artery (white arrow). (c) Image obtained in hepatic phase on which arterial invasion was correctly diagnosed shows tumor (arrowheads) and artery (arrow).
|
|
View larger version (161K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 4b. Transverse CT images in 68-year-old man with replaced right hepatic artery in juxtaposition to pancreatic adenocarcinoma in the posterior aspect of the pancreatic head, which indicates arterial invasion. (a) Image obtained in arterial phase on which the tumor was not identified with certainty, which likely led to misdiagnosis of lack of vascular invasion. Replaced right hepatic artery (arrow) and tumor (arrowheads) are seen. (b) Image obtained in pancreatic phase on which vascular invasion was correctly identified. Note increased conspicuity of the tumor (arrowheads) compared with the normal gland (black arrow). The low-attenuation tumor is clearly depicted adjacent to the artery (white arrow). (c) Image obtained in hepatic phase on which arterial invasion was correctly diagnosed shows tumor (arrowheads) and artery (arrow).
|
|
View larger version (158K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 4c. Transverse CT images in 68-year-old man with replaced right hepatic artery in juxtaposition to pancreatic adenocarcinoma in the posterior aspect of the pancreatic head, which indicates arterial invasion. (a) Image obtained in arterial phase on which the tumor was not identified with certainty, which likely led to misdiagnosis of lack of vascular invasion. Replaced right hepatic artery (arrow) and tumor (arrowheads) are seen. (b) Image obtained in pancreatic phase on which vascular invasion was correctly identified. Note increased conspicuity of the tumor (arrowheads) compared with the normal gland (black arrow). The low-attenuation tumor is clearly depicted adjacent to the artery (white arrow). (c) Image obtained in hepatic phase on which arterial invasion was correctly diagnosed shows tumor (arrowheads) and artery (arrow).
|
|
View larger version (114K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 5a. Transverse CT images in 76-year-old man with vascular invasion of the superior mesenteric vein from pancreatic adenocarcinoma. (a) Image obtained in hepatic phase above region of narrowing shows normal-caliber portal vein (arrow). (b) Image obtained in hepatic phase shows focal narrowing and flattening of the superior mesenteric vein (arrows). (c) Image obtained in hepatic phase below narrowing demonstrates normal-caliber superior mesenteric vein (large arrow) and tumor (arrowhead) adjacent to common bile duct stent. Note inferior mesenteric vein (small arrows). (d, e) Images obtained in pancreatic phase at a level corresponding to b and c, respectively, fail to demonstrate convincing narrowing of the superior mesenteric vein (arrow), and contrast enhancement of the superior mesenteric vein is suboptimal. Note inferior mesenteric vein (small arrows) on e.
|
|
View larger version (116K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 5b. Transverse CT images in 76-year-old man with vascular invasion of the superior mesenteric vein from pancreatic adenocarcinoma. (a) Image obtained in hepatic phase above region of narrowing shows normal-caliber portal vein (arrow). (b) Image obtained in hepatic phase shows focal narrowing and flattening of the superior mesenteric vein (arrows). (c) Image obtained in hepatic phase below narrowing demonstrates normal-caliber superior mesenteric vein (large arrow) and tumor (arrowhead) adjacent to common bile duct stent. Note inferior mesenteric vein (small arrows). (d, e) Images obtained in pancreatic phase at a level corresponding to b and c, respectively, fail to demonstrate convincing narrowing of the superior mesenteric vein (arrow), and contrast enhancement of the superior mesenteric vein is suboptimal. Note inferior mesenteric vein (small arrows) on e.
|
|
View larger version (112K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 5c. Transverse CT images in 76-year-old man with vascular invasion of the superior mesenteric vein from pancreatic adenocarcinoma. (a) Image obtained in hepatic phase above region of narrowing shows normal-caliber portal vein (arrow). (b) Image obtained in hepatic phase shows focal narrowing and flattening of the superior mesenteric vein (arrows). (c) Image obtained in hepatic phase below narrowing demonstrates normal-caliber superior mesenteric vein (large arrow) and tumor (arrowhead) adjacent to common bile duct stent. Note inferior mesenteric vein (small arrows). (d, e) Images obtained in pancreatic phase at a level corresponding to b and c, respectively, fail to demonstrate convincing narrowing of the superior mesenteric vein (arrow), and contrast enhancement of the superior mesenteric vein is suboptimal. Note inferior mesenteric vein (small arrows) on e.
|
|
View larger version (132K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 5d. Transverse CT images in 76-year-old man with vascular invasion of the superior mesenteric vein from pancreatic adenocarcinoma. (a) Image obtained in hepatic phase above region of narrowing shows normal-caliber portal vein (arrow). (b) Image obtained in hepatic phase shows focal narrowing and flattening of the superior mesenteric vein (arrows). (c) Image obtained in hepatic phase below narrowing demonstrates normal-caliber superior mesenteric vein (large arrow) and tumor (arrowhead) adjacent to common bile duct stent. Note inferior mesenteric vein (small arrows). (d, e) Images obtained in pancreatic phase at a level corresponding to b and c, respectively, fail to demonstrate convincing narrowing of the superior mesenteric vein (arrow), and contrast enhancement of the superior mesenteric vein is suboptimal. Note inferior mesenteric vein (small arrows) on e.
|
|
View larger version (125K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 5e. Transverse CT images in 76-year-old man with vascular invasion of the superior mesenteric vein from pancreatic adenocarcinoma. (a) Image obtained in hepatic phase above region of narrowing shows normal-caliber portal vein (arrow). (b) Image obtained in hepatic phase shows focal narrowing and flattening of the superior mesenteric vein (arrows). (c) Image obtained in hepatic phase below narrowing demonstrates normal-caliber superior mesenteric vein (large arrow) and tumor (arrowhead) adjacent to common bile duct stent. Note inferior mesenteric vein (small arrows). (d, e) Images obtained in pancreatic phase at a level corresponding to b and c, respectively, fail to demonstrate convincing narrowing of the superior mesenteric vein (arrow), and contrast enhancement of the superior mesenteric vein is suboptimal. Note inferior mesenteric vein (small arrows) on e.
|
|
Four patients had false-positive examinations for prediction of vascular invasion on images obtained in the hepatic phase. In all four of these patients, the reader of the images obtained in the arterial or pancreatic phase also identified vascular invasion of the same structure. Two of these patients did not undergo surgical examination, but they underwent MR imaging and endoscopic US as the reference standard. Findings of both of these examinations did not indicate invasion of the superior mesenteric vein or of the portal vein, which was seen at CT as flattening or narrowing and concavity toward the vessel lumen, respectively. These patients died at 2 and 14 months after the CT examination. Another patient had false-positive findings for superior mesenteric vein and portal vein invasion at CT (ie, 90°–180° apposition and flattening or narrowing) and underwent an attempted Whipple procedure. The mesenteric veins appeared normal at surgical palpation, but the Whipple procedure was aborted after the discovery of metastatic lymphadenopathy, which was located outside the resection specimen. The last patient had false-positive findings for superior mesenteric artery invasion, which was seen as invasion of the periarterial fat plane, but findings for such were negative at surgery and at histologic examination. This patient developed a perisuperior mesenteric artery recurrence, however, which was seen at surveillance CT 4 months after resection, and died 6 months after surgery.
Table 5 includes the sensitivity of the individual readers for images obtained during each phase of enhancement, with respect to detection of tumor and vascular invasion. There was no statistically significant difference in sensitivity for tumor detection on images obtained in any phase of enhancement among the four radiologists. Readers 1 and 2 did not perform as well as did readers 3 and 4 in the interpretation of images obtained in the arterial phase with respect to tumor detection, but they performed virtually identically in the detection of tumors on images obtained in the pancreatic and hepatic phases. There was likewise no significant difference in reader performance with respect to sensitivity for vascular invasion staging. The P values from the 2 test of equality of specificity estimates for detection of tumor and of vascular invasion for the four radiologists were also not significant for any phase of enhancement.
|
DISCUSSION
|
|---|
We evaluated a triple-phase multi–detector row CT protocol with arterial, pancreatic, and hepatic phase imaging in patients who were suspected of having resectable pancreatic adenocarcinoma. With our multi–detector row CT protocol, a narrow section thickness of 2.5 mm was used for each phase of enhancement. We evaluated the relative differences in attenuation between the pancreas, the pancreatic tumor, and the peripancreatic vessels on images obtained in each phase of enhancement, as others have determined for single– and dual–detector row CT (1–6). Such evaluations have implications about tumor and vessel conspicuity in each vascular phase. We also estimated the effect of each phase of enhancement on clinical end points, such as tumor detection and vascular invasion, by having different radiologists interpret images obtained in each phase. Readers who evaluated images obtained in the pancreatic and hepatic phases performed well in the detection of pancreatic neoplasms, with overall sensitivities of 0.94 and 0.91, respectively, and performed significantly better than did readers who interpreted images obtained in the arterial phase (P < .001, pancreatic phase; P = .002, hepatic phase). This trend also was true when they evaluated images of patients with pancreatic ductal adenocarcinoma, with sensitivities for pancreatic adenocarcinoma of 0.63, 0.97, and 0.93, for images obtained in the arterial, pancreatic, and hepatic phases, respectively (P = .002, pancreatic phase; P = .008, hepatic phase). We found that images obtained in the pancreatic and hepatic phases may be complementary in a small number of patients in whom findings are equivocal on images obtained in one of these two phases. For example, in 30 patients, this phenomenon occurred in one patient for images obtained in the pancreatic phase, and it occurred in two patients for images obtained in the hepatic phase. However, these small differences could also be caused by interobserver variability.
Tumor detection depends in part on the attenuation characteristics of the pancreas and the adjacent pancreatic adenocarcinoma. Both the mean pancreatic attenuation and the tumor-to-gland attenuation differences were greatest on images obtained in the pancreatic phase and smallest on those obtained in the arterial phase (Fig 2), and these findings are in agreement with those of others (1,2,7). In approximately 70% of patients, maximal pancreatic enhancement was observed on images obtained in the pancreatic phase, and in a similar percentage, maximal tumor-to-gland attenuation differences were demonstrated on images obtained in the pancreatic phase. More important, in no patients in our study were tumor-to-gland attenuation differences less than 10 HU on images obtained in the pancreatic phase, because attenuation differences of less than 10 HU cannot be identified readily (18). Similar to findings in the study of Lu et al (1), in several patients in this study (five [17%] of 29), tumor-to-gland attenuation differences were observed that were less than 10 HU compared with the attenuation of the surrounding gland in the hepatic phase. Despite this limitation, tumors in all five of these patients were identified by the readers of the images obtained in the hepatic phase, and this finding emphasizes that other signs such as ductal cutoff and mass effect assist in tumor identification. These signs may be more apparent when thin-section techniques are used. In this regard, multi–detector row CT allows the acquisition of images in the hepatic phase with narrow, 2.5-mm or smaller, section thickness in multiple phases of enhancement without concerns for tube heating. We have shown that tumor-to-gland attenuation differences are smallest on images obtained in the arterial phase and that the findings on images obtained in the arterial phase are unlikely to complement the tumor depiction capabilities of images obtained in the pancreatic and hepatic phases. We conclude that it is unnecessary to acquire images in the arterial phase when one is searching for pancreatic ductal adenocarcinoma with a multi–detector row CT protocol.
In regard to the clinical end point of correctly predicting vascular invasion, radiologists who interpreted images obtained in the hepatic phase performed significantly better than those who interpreted images obtained in the arterial phase (P = .016), with a trend toward improved performance versus that in interpretation of images obtained in the pancreatic phase. These results are likely to result from flow artifacts in the superior mesenteric vein, which we observed in the pancreatic phase, as described (Fig 3). In regard to arterial invasion, those who interpreted images obtained in the hepatic phase demonstrated slightly improved performance compared with those who interpreted images obtained in the other phases (Fig 4). As a result of these findings, we fail to find evidence to support the acquisition of images in the arterial and pancreatic phases for the staging of pancreatic ductal adenocarcinoma once a tumor has been identified on thin-section images obtained in the hepatic phase. More important, with our protocol, thin 2.5-mm sections were used to obtain images in the hepatic phase, and this may have contributed to our higher accuracy of interpretation of findings on images obtained in the hepatic phase. This finding may be particularly important when a tumor is identified at routine screening CT of the abdomen with multi–detector row CT performed in the hepatic phase: Eight– or 16–detector row scanners can retrospectively construct 2.5-mm-thick sections from a routine scan in which 5.0-mm section thickness is used, thus allowing the creation of a thin-section data set that could be used to predict vascular invasion with a high degree of accuracy. This potential will allow such patients to forego a second CT examination for staging of the vascular invasion. In this regard, we found that the sensitivity for vascular invasion was identical between readers in the current study who exclusively evaluated images obtained in the hepatic phase (sensitivity, 0.83) and readers in our earlier study who simultaneously examined images obtained in all three phases of vascular enhancement (8).
To accurately assess vascular invasion, the peripancreatic arteries and veins must have a homogeneous attenuation that is noticeably different from that of the pancreatic tumor and parenchyma. Inhomogeneous attenuation that results from flow artifacts or from suboptimal opacification of the superior mesenteric vein makes it difficult to distinguish flattening or narrowing of the vessel, particularly when the vessel is in juxtaposition to a low-attenuation tumor (Fig 5). Just as McNulty et al (7) observed, we found that venous enhancement was greatest on images obtained in the hepatic phase, whereas arterial enhancement was greatest on images obtained in the arterial and pancreatic phases. We also found that the difference between the attenuation of the superior mesenteric vein and that of the pancreas was also greater on images obtained in the hepatic phase. However, in most of our patients (32 [84%] of 38), enhancement of the superior mesenteric vein in the pancreatic phase was inhomogeneous because of flow artifacts; thus, inhomogeneity may compromise the staging of vascular invasion in this phase. This finding was different from the observation of Lu et al (1), who defined the pancreatic phase as the period during which "tumor-pancreas contrast is maximal with adequate portal venous and arterial opacification." Both Lu et al (1) and Boland et al (2) found that the superior mesenteric vein had higher attenuation on images obtained in the pancreatic phase than it did on those obtained in the hepatic phase. We suggest that the speed of pancreatic phase acquisition with multi–detector row CT is sufficiently faster than that with single– or dual–detector row CT so that flow artifacts are now present in much greater frequency (Figs 1, 3). The faster speed of multi–detector row CT scanners means that the period of maximal pancreatic enhancement is temporally separated from that of maximal homogeneous venous enhancement. This fact raises the possibility that there is an intermediate phase between the pancreatic and hepatic phases that can be seen with multi–detector row CT in which mesenteric venous opacification is maximal but pancreatic and hepatic attenuation is submaximal (Lu D, oral communication, 2001).
Although sensitivity was improved at imaging performed in the hepatic phase for detection of vascular invasion compared with sensitivity at imaging performed in the arterial and pancreatic phases, there were false-positive examinations for vascular invasion in four patients, compared with findings with our reference standard (specificity, 0.82). Readers of the images obtained in the other phases of enhancement detected the same findings in all four of these patients. This finding led us to conclude that either our imaging criteria for vascular invasion or our reference standard did not reflect truth in these cases.
All three cases of purported venous invasion demonstrated only 90°–180° juxtaposition of the tumor and the vein, with concavity toward the vessel or flattening of the vein in each case. These findings are known to be sensitive for venous invasion but occasionally can yield false-positive results (14,16). In two of these cases, a consensus reading of MR images and endoscopic US scans was used as the reference standard, and the discrepancy between the findings on CT scans and on the MR images and US scans may reflect the potentially superior sensitivity of CT for vascular invasion (8). Finally, one area of suspected arterial invasion at CT was resectable surgically, and it had histologic margins negative for tumor. However, a perisuperior mesenteric artery recurrence was seen at CT 4 months later, and this finding led us to believe that the surgical reference standard was flawed in this case.
McNulty et al (7) described a triple-phase multi–detector row CT protocol for the pancreas, but they used a section thickness of 5.0 mm to obtain images in the portal phase (7). With their protocol, they used different section thicknesses for obtaining images in each phase of enhancement, and they did not report results with respect to clinical end points, such as tumor detection and vascular invasion staging. They also concluded that images obtained in the arterial phase were noncontributive to the diagnosis of pancreatic adenocarcinoma. They reported a tumor-to-gland attenuation difference similar to that observed in our study, despite a slightly earlier timing of their pancreatic phase of enhancement and their use of a thicker section of 5.0 mm for images obtained in the hepatic phase, compared with the 2.5-mm section thickness used in our study. Figure 1 gives the timing of different phases of pancreatic enhancement for several studies.
There are several limitations of our study. First, our study population included patients who were most likely to benefit from additional pancreatic imaging, that is, patients who were suspected of having resectable pancreatic adenocarcinoma. Our results reflect the performance of triple-phase CT of the pancreas in this select population with a high proportion of smaller tumors (8). Although the estimated sensitivities and specificities do not measure the performance of each phase of enhancement in a larger screening population, they do provide insight into the relative contribution of each phase to clinical end points, such as tumor detection and vascular invasion staging. Because we excluded patients with gross vascular invasion determined at prior imaging, the relative percentage of patients with less specific signs of vessel invasion was likely to be increased, and this increase affected our reported specificity for vascular invasion. Second and more important, we did not have one radiologist interpret images obtained in all three phases of vascular enhancement for each patient and we did not randomize which scan each radiologist would interpret. We believed that to do so would introduce recall bias, because many of the tumors were small and had an idiosyncratic appearance (ie, they would be recalled by the interpreting radiologist when he or she evaluated tumor presence on images obtained in different phases of enhancement). Consequently, we chose to have different radiologists with similar experience interpret images obtained in each phase of enhancement in each patient. This method, however, introduced interobserver variability (ie, a "radiologist effect"), which could confound the effect of the phase of enhancement on the clinical end points of detection of tumor and vascular invasion if the radiologists had performed differently. We believed that potential recall bias from using one radiologist to interpret images obtained in each phase of enhancement was more likely to be a confounding factor in examination of the contribution of each phase to clinical end points, such as detection of tumor and vascular invasion, than was interobserver variability between radiologists. To assess the effect of these weaknesses in our study design, we looked for differences in the estimates of sensitivity and specificity between our four radiologists, and we found that these differences were not considered statistically significant. Therefore, we believe it is reasonable to examine the overall estimates for each phase of enhancement. We do not believe differences in interpretation that resulted from interobserver variability accounted for the significantly increased sensitivity of the pancreatic and hepatic phases compared with the sensitivity of the arterial phase.
Finally, we allowed different rates of contrast material injection for each patient and did not subcategorize our patient population according to these rates of injection. However, we sought to use a protocol that could be used with all patients. Given the short periods of acquisition of approximately 5 seconds for the arterial and pancreatic phases with our multi–detector row CT protocol, the temporal window during which images were acquired in each phase still overlapped with the scanning of images in similar phases with the single–detector row protocols (Figs 1, 3), regardless of the injection rate used. Furthermore, McNulty et al (7) demonstrated similar findings with their multi–detector row CT protocol for the pancreas, and they used a constant injection rate.
We can draw several conclusions from our evaluation of a triple-phase multi–detector row CT protocol for the pancreas. First, routine acquisition of images in the arterial phase is unnecessary in the detection and staging of pancreatic ductal adenocarcinoma. The acquisition of images in the pancreatic phase is helpful when detection of pancreatic malignancies is desired because tumor-to-gland attenuation differences are greatest on images obtained in this phase. The review of images obtained in the pancreatic phase may be helpful in the small number of patients with equivocal findings on images obtained in the hepatic phase. Second, images of narrow section thickness obtained during the hepatic phase allow tumor detection nearly equivalent to that of those obtained during the pancreatic phase, and they most accurately depict vascular invasion. When these images are available for review, the acquisition of images during the arterial and pancreatic phases is likely to be of little benefit in the improvement of the prediction of tumor resectability. Finally, the speed of acquisition of multi–detector row CT scans results in a pancreatic phase that appears to be different from that which results with earlier generation helical CT scanners: Maximal pancreatic enhancement precedes maximal venous enhancement, and flow artifacts are seen within the veins adjacent to the pancreas during the pancreatic phase.
|
FOOTNOTES
|
|---|
Author contributions: Guarantor of integrity of entire study, J.G.F.; study concepts, J.G.F., D.H.S.; study design, J.G.F., M.J.W., D.H.S., C.D.J., E.M.W.; literature research, J.G.F.; clinical studies, J.G.F., M.J.W., M.A.F., J.L.F., L.J.B., T.K., C.D.J., D.H.S., E.M.W.; data acquisition, J.G.F., M.J.W., M.A.F., J.L.F., L.J.B., T.K., C.D.J., D.H.S., E.M.W.; data analysis/interpretation, J.G.F., M.J.W., W.S.H.; statistical analysis, J.G.F., M.J.W., W.S.H.; manuscript preparation, editing, and final version approval, J.G.F.; manuscript definition of intellectual content, all authors; manuscript revision/review, J.G.F., W.S.H.
|
REFERENCES
|
|---|
- Lu DSK, Vedantham S, Krasny RM, Kadell B, Berger WL, Reber HA. Two-phase helical CT for pancreatic tumors: pancreatic versus hepatic phase enhancement of tumor, pancreas and vascular structures. Radiology 1996; 199:697-701.[Abstract/Free Full Text]
- Boland GW, O’Malley ME, Saez M, Fernandez-del-Castillo C, Warshaw AL, Mueller PR. Pancreatic-phase versus portal vein-phase helical CT of the pancreas: optimal temporal window for evaluation of pancreatic adenocarcinoma. AJR Am J Roentgenol 1999; 172:605-608.[Abstract/Free Full Text]
- Hollett MD, Jorgensen MJ, Jeffrey RB. Quantitative evaluation of pancreatic enhancement during dual-phase helical CT. Radiology 1995; 195:359-361.[Abstract/Free Full Text]
- Tabuchi T, Itoh K, Ohshio G, et al. Tumor staging of pancreatic adenocarcinoma using early- and late-phase helical CT. AJR Am J Roentgenol 1999; 173:375-380.[Abstract/Free Full Text]
- Keogan MT, McDermott VG, Paulson EK, et al. Pancreatic malignancy: effect of dual-phase helical CT in tumor detection and vascular opacification. Radiology 1997; 205:513-518.[Abstract/Free Full Text]
- Graf O, Boland GW, Warshaw AL, Fernandez-del Castillo C, Hahn PF, Mueller PR. Arterial versus portal venous helical CT for revealing pancreatic adenocarcinoma: conspicuity of tumor and critical vascular anatomy. AJR Am J Roentgenol 1997; 169:119-123.[Abstract/Free Full Text]
- McNulty NJ, Francis IR, Platt JF, Cohan RH, Korobkin M, Gebremariam A. Multi-detector row helical CT of the pancreas: effect of contrast-enhanced multiphasic imaging on enhancement of the pancreas, peripancreatic vasculature, and pancreatic adenocarcinoma. Radiology 2001; 220:97-102.[Abstract/Free Full Text]
- Wiersema MJ, Fletcher JG, Burgart LJ, et al. Prospective evaluation of potentially resectable pancreatic ductal adenocarcinoma: performance of triple-phase helical CT, gadolinium-enhanced MR and endoscopic ultrasound (abstr). Radiology 2001; 221 (P):250.
- Kalbhen CL, Yetter EM, Olson MC, Posniak HV, Aranha GV. Assessing the resectability of pancreatic carcinoma: the value of reinterpreting abdominal CT performed at other institutions. AJR Am J Roentgenol 1998; 171:1571-1576.[Abstract/Free Full Text]
- Tublin ME, Tessler FM, Cheng SL, Peters TL, McGovern PC. Effect of injection rate of contrast medium on pancreatic and hepatic helical CT. Radiology 1999; 210:97-101.[Abstract/Free Full Text]
- Kim T, Murakami T, Takahashi S, et al. Effects of injection rates of contrast material on arterial phase hepatic CT. AJR Am J Roentgenol 1998; 171:429-432.[Abstract/Free Full Text]
- Kim T, Murakami T, Takahashi S, et al. Pancreatic CT imaging: effects of different injection rates and doses of contrast material. Radiology 1999; 212:219-225.[Abstract/Free Full Text]
- Charnsangavej C. Staging of pancreatic tumors: state-of-the-art and helical CT. Presented at the Society for Gastrointestinal Radiology Postgraduate Course 2000, Kauai, Hawaii, March 15–20 2000.
- Lu DSK, Reber HA, Krasny RM, Kadell BM, Sayre J. Local staging of pancreatic cancer: criteria for unresectability of major vessels as revealed by pancreatic-phase, thin-section helical CT. AJR Am J Roentgenol 1997; 168:1439-1443.[Abstract/Free Full Text]
- O’Malley ME, Boland GW, Wood BJ, Fernandez-del Castillo C, Warshaw AL, Mueller PR. Adenocarcinoma of the head of the pancreas: determination of surgical unresectability with thin-section pancreatic-phase helical CT. AJR Am J Roentgenol 1999; 173:1513-1518.[Abstract]
- Hough TJ, Raptopoulos V, Siewert B, Matthews JG. Teardrop superior mesenteric vein: CT sign for unresectable carcinoma of the pancreas. AJR Am J Roentgenol 1999; 173:1509-1512.[Abstract]
- Phoa SSKS, Reeders JWAJ, Stoker J, Rauws EAJ, Gouma DJ, Lameris JS. CT criteria for venous invasion in patients with pancreatic head carcinoma. Br J Radiol 2000; 73:1159-1164.[Abstract]
- Baron R. Understanding and optimizing use of contrast material for CT of the liver. AJR Am J Roentgenol 1994; 163:323-331.[Abstract/Free Full Text]
This article has been cited by other articles:
|
|
|
|
 
H. Furukawa, K. Uesaka, and N. Boku
Treatment Decision Making in Pancreatic Adenocarcinoma: Multidisciplinary Team Discussion With Multidetector-Row Computed Tomography
Arch Surg,
March 1, 2008;
143(3):
275 - 280.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
Y. Yanaga, K. Awai, Y. Nakayama, T. Nakaura, Y. Tamura, M. Hatemura, and Y. Yamashita
Pancreas: Patient Body Weight tailored Contrast Material Injection Protocol versus Fixed Dose Protocol at Dynamic CT
Radiology,
November 1, 2007;
245(2):
475 - 482.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
<
TOP
ABSTRACT
INTRODUCTION
= pancreas, 
= tumor, 
= mean tumor-to-gland attenuation difference.