| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Gastrointestinal Imaging |
1 From the Department of Radiology, Yamanashi Medical University School of Medicine, 1110 Shimokato, Tamaho, Nakakoma, Yamanashi 409-3815, Japan (T.I., T.A.); the Department of Radiology, University of Pittsburgh Medical Center, Pa (T.I., M.S.P., M.P.F., R.L.B., Y.K.); the Department of Radiology, National Cancer Center East Hospital, Chiba, Japan (H.H., S.N.). Received April 14, 1999; revision requested May 12; final revision received October 8; accepted October 26. Address correspondence to T.I. (e-mail: ichikawa@res.yamanashi-med.ac.jp).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: Retrospective quantitative, qualitative, and receiver operating characteristic analyses of biphasic CT and MR imaging were performed in 19 patients with 26 histopathologically proved islet cell tumors. Delayed arterial dominant–phase (AP) and portal venous–phase (PVP) biphasic CT was performed after the administration of contrast material. MR imaging included T1-weighted spin-echo (SE) and T2-weighted SE or fast SE imaging, fat-saturated T1-weighted SE imaging, dynamic contrast material–enhanced T1-weighted gradient-echo imaging, and delayed enhanced T1-weighted SE imaging with or without fat saturation.
RESULTS: PVP CT and delayed enhanced T1-weighted MR imaging had the highest Az values (0.98 and 0.97, respectively; P < .05). Delayed enhanced T1-weighted MR imaging had the highest relative sensitivity (14–15 [74%–79%] of 19 lesions), followed by PVP CT (18–19 [69%–73%] of 26 lesions), AP CT (17–19 [65%–73%] of 26 lesions), fat-saturated T1-weighted MR imaging (eight to 10 [57%–71%] of 14 lesions), T2-weighted (16–17 [62%–65%] of 26 lesions), T1-weighted (15–18 [58%–69%] of 26 lesions) MR imaging, and dynamic MR imaging (nine [56%] of 16 lesions).
CONCLUSION: Biphasic (especially PVP) CT and MR imaging have similar effectiveness in the detection of islet cell tumors if fat-saturated T1-weighted and delayed enhanced T1-weighted MR imaging are included.
Index terms: Computed tomography (CT), comparative studies, 770.12111, 770.12112, 770.12114, 770.12115 • Magnetic resonance (MR), comparative studies, 770.121411, 770.121412, 770.121415, 770.12143 • Pancreas, CT, 770.12111, 770.12112, 770.12114, 770.12115 • Pancreas, MR, 770.121411, 770.121412, 770.121415, 770.12143 • Pancreas, neoplasms, 770.3191 • Receiver operating characteristic (ROC) curve
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Dynamic contrast material–enhanced computed tomography (CT) and magnetic resonance (MR) imaging are common noninvasive preoperative radiologic techniques used in the evaluation of islet cell tumors (1–3,7–17). Recently, helical CT has been successfully applied in the imaging of pancreatic disease, as it allows multiphase imaging during a single bolus administration of contrast material to optimize organ enhancement and lesion conspicuity (9–11,18–21). In the evaluation of islet cell tumors, several previous articles (9–11) have emphasized that the combination of arterial dominant–phase (AP) and portal venous–phase (PVP) CT improves the detection of hepatic metastases and primary tumors. The recent consensus based on efficacy results is that, among the various enhanced CT methods, biphasic CT is the most sensitive method in the detection of islet cell tumors (9–11).
Several investigators (12–17) have also emphasized the advantages of MR imaging over CT, including higher contrast resolution and higher sensitivity for contrast enhancement in the investigation of islet cell tumors. On the basis of these advantages, some investigators have concluded that MR imaging is a promising tool in the detection of islet cell tumors that is possibly superior to dynamic enhanced CT (15–17).
The purpose of this study was to compare the effectiveness of biphasic CT and MR imaging with various pulse sequences in the detection of islet cell tumors in a patient group that is larger than those previously reported.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Thirteen patients underwent complete tumor resection at either partial pancreatectomy (n = 12; seven with distal pancreatectomy and five with pancreatoduodenectomy) or tumor enucleation (n = 1). The remaining six patients were first suspected of having nonfunctioning islet cell tumors at percutaneous needle biopsy of multiple hepatic tumors. Subsequently, a histopathologic diagnosis of nonfunctioning malignant islet cell tumor of the pancreas was established at exploratory laparotomy (n = 4) or percutaneous pancreatic biopsy (n = 2).
The final histopathologic diagnosis in eight patients was functioning islet cell tumors (four gastrinomas, one insulinoma, three carcinoids). The remaining 11 patients had nonfunctioning islet cell tumors (one benign lesion, 10 malignant lesions). Three patients with gastrinomas had clinical manifestations of the Zollinger-Ellison syndrome. Three patients with gastrinomas and one patient with a carcinoid had the multiple endocrine neoplasia type 1 syndrome. Four patients had multiple pancreatic tumors at surgery; one patient had two carcinoids, two patients with nonfunctioning islet cell tumors had two lesions each, and one patient with a gastrinoma had five tumors.
In summary, the 19 patients had a total of 26 tumors (13 functioning, 13 nonfunctioning). Each of the 26 tumors had a histopathologic diagnosis. At histopathologic examination, the size of these islet cell tumors ranged from 4 to 60 mm (mean, 22 mm). Nonfunctioning islet cell tumors (9–60 mm; mean, 28 mm) tended to be larger than the functioning ones (4–25 mm; mean, 12 mm).
All CT examinations were performed with commercially available helical CT scanners (HiSpeed Advantage, GE Medical System, Milwaukee, Wis; X-Vigor, X-Vision, Toshiba Medical, Tokyo, Japan). All patients underwent nonenhanced and enhanced biphasic CT that included AP and PVP CT. Patients intravenously received 150 mL of either iothalamate meglumine (Conray-60; Mallinckrodt Medical, St Louis, Mo) or ioversol (Optiray 320; Mallinckrodt Medical) by means of a power injector (model OP 100; Medrad, Pittsburgh, Pa). Contrast material was administrated at a rate of 3–5 mL/sec, with a monophasic rate of injection in all patients. After the infusion of contrast material was initiated, AP CT scanning was started after a delay of 20–28 seconds, and PVP CT scanning was started after a delay of 60–70 seconds. All scans were acquired in a cephalocaudal direction. The section thickness was 5–7 mm, and the table incremental speed was 5–7 mm/sec.
All MR imaging examinations were performed with commercially available 1.5-T MR imagers (Signa Advantage or Horizon, GE Medical Systems; Magnetom SP, Siemens, Erlangen, Germany). As this study was a retrospective study, there was moderate variability in the MR imaging units and sequence parameters, which usually involved the pulse sequence, repetition and echo times, flip angle, image matrix, or scanning method (eg, breath hold or non–breath hold).
All patients underwent T1-weighted spin-echo (SE) and T2-weighted SE (15 patients) or fast SE (four patients) imaging. Additional MR imaging included nonenhanced fat-saturated T1-weighted SE imaging (14 lesions in 10 patients). After T1- and T2-weighted images were obtained, dynamic contrast-enhanced T1-weighted gradient-echo MR imaging (16 lesions in 13 patients) was performed during an intravenous bolus administration of 0.1 mmol/kg gadolinium-based contrast material (Magnevist; Schering, Berlin). Delayed enhanced T1-weighted SE images (19 lesions in 17 patients) with (15 patients) or without (two patients) fat saturation were obtained 5–10 minutes after the administration of the contrast material.
Detailed parameters for the MR imaging sequences were the following: (a) T1-weighted SE sequence, 366–660/12–20 (repetition time msec/echo time msec), 128–512 x 256–512 matrix, 5–10-mm section thickness, 1–2-mm intersection gap, two to four signals acquired, and 8–12-minute acquisition time; (b) T2-weighted SE sequence, 2,000–4,100/70–140, 128–192 x 256 matrix; 5–7-mm section thickness, 1–3-mm intersection gap, one or two signals acquired, and 4–10-minute acquisition time; (c) T2-weighted fast SE sequence, 2,500–6,666/84–105, echo train length of eight to 20, 192–256 x 256 matrix, 5–8-mm section thickness, 1–3-mm intersection gap, one or two signals acquired, and 25-second 7-minute acquisition time; (d) fat-saturated and delayed enhanced (with or without fat saturation) T1-weighted SE sequences, 450–733/12–20, 128–192 x 256 matrix, 5–10-mm section thickness, 1–2-mm intersection gap, one to four signals acquired, and 16-minute 24-second acquisition time; (e) dynamic non–breath hold sequential enhanced gradient-echo T1-weighted sequence, 8.8–10.6/2.0–3.3, 30° flip angle, 128 x 256 matrix, 10-mm section thickness, 2-mm intersection gap, one signal acquired, and 2–3-minute acquisition time for 90–162 sections; and (f) multisection dynamic breath-hold enhanced gradient-echo T1-weighted sequence, 80–180/1.8–6.0, 60°–90° flip angle, 128 x 256 matrix, 7–10-mm section thickness, 2–3-mm intersection gap, one signal acquired, and 18–27-second acquisition time.
Quantitative measurements of CT enhancement (in Hounsfield units) were made in the 15 lesions that were larger than 10 mm. For this purpose, one radiologist (T.I.) placed regions of interest over the lesions and normal pancreas on the nonenhanced and enhanced biphasic CT images. The region of interest for each lesion was carefully placed within the confines of the entire lesion. As a rule, for heterogeneous lesions, the regions of interest were placed to include the entire lesion without excluding the various components with different CT attenuation values. However, regions in a lesion or in the pancreas with calcifications or cystic foci were avoided in the region-of-interest analysis.
CT attenuation values were measured three times and were averaged for each lesion. Mean enhancement of the normal pancreas or tumor was calculated by subtracting attenuation values before contrast enhancement from the respective attenuation values after contrast enhancement. The difference between the mean attenuation of the tumor and pancreas, or the tumor-to-pancreas contrast, was calculated on both AP and PVP CT images by using the following equation: tumor-to-pancreas contrast = mean tumor enhancement ÷ mean normal pancreatic enhancement.
To investigate the possibility that image quality affects lesion detectability, two reviewers (M.S.P. and Y.K.) separately evaluated the quality of all CT and MR images. A five-point grading system (5 was excellent, 4 was good, 3 was marginally adequate, 2 was poor, and 1 was nondiagnostic) was used. Images were subjectively graded on the basis of the contrast enhancement or conspicuity of the pancreas. Images were considered nondiagnostic if contrast enhancement of the pancreas was not depicted on enhanced AP CT or MR images or if the pancreas could not be confidently identified because of any kind of image degradation (eg, presence of motion-related artifacts on MR images). If the CT and MR images had good contrast enhancement, if the pancreas could be clearly depicted, and if no image degradation was noted, the image quality was considered to be excellent. Levels of image quality between nondiagnostic and excellent were subjectively scored. If the subjective grading of the two readers disagreed, that of a third reviewer (H.H.) was used to achieve a majority rule.
Image interpretation in tumor detection was performed in only primary pancreatic lesions. Although assessment of distant metastases (especially hepatic metastases) is important in the evaluation of islet cell tumors, it was not possible because the range of some of the MR images in this study was not large enough to cover the whole liver.
In the receiver operating characteristic (ROC) analysis, one radiologist (T.I.) who had knowledge of the clinical and histopathologic findings served as the study coordinator and initially reviewed all CT and MR images. On the basis of findings in the surgical pathology reports, he attempted to determine the true number and location of all tumors in all patients. He also attempted to correlate the anatomic and imaging findings of the histopathologically proved tumors as accurately as possible to allow detection of false-positive readings. Therefore, the study coordinator considered lesions other than the histopathologically proved tumors to be false-positive.
If different kinds of CT or MR images were initially printed on the same piece of film, the study coordinator reprinted the different types or series of images on separate pieces of film. Also, if the initial hard-copy images had suboptimal quality, the study coordinator reprinted the images. No normal CT or MR images of patients not included in the study population were used in the ROC analysis because of the difficulty in finding suitable images, especially MR images, that matched those included in the study with regard to image quality, imaging parameters, or degree of contrast enhancement in the pancreas. Therefore, image interpretation was conducted on a portion-by-portion basis by dividing the pancreas into the head, body, and tail according to pancreatic anatomy.
All CT and MR images were then independently interpreted in random order by three experienced abdominal radiologists (M.S.P, Y.K., H.H.). The study coordinator separately presented the images obtained with all CT and MR sequences. The images were randomly presented to each reader in terms of patient and sequence order. The purpose of the study was to compare biphasic CT and MR images in the detection of islet cell tumors and also to determine the kinds of CT or MR imaging that are necessary to include in clinical protocols; therefore, biphasic CT images and the various kinds of MR images were not interpreted together in any combination.
Readers knew that the CT and MR examinations were performed in each patient to evaluate possible islet cell tumors in the upper abdominal organs. However, readers were blinded to patient identity, clinical history, and results of other imaging and histopathologic evaluations. Although all biphasic CT and MR images were presented to the blinded readers in a randomized fashion, it was difficult to blind the readers to the phase of contrast enhancement because of obvious differences in vascular and renal enhancement at biphasic CT and MR imaging. In addition, readers were aware that images were T1-weighted when they interpreted dynamic MR images that also included images obtained before contrast enhancement.
On all CT and MR images, each reader graded the presence of a tumor on a five-point confidence scale (1 was definitely absent, 2 was probably absent, 3 was equivocal, 4 was probably present, and 5 was definitely present). If tumors were considered to be present in the pancreas on any CT or MR image, the number and location of tumors were recorded. In addition, the predominant attenuation or signal intensity of the tumors was described as hypo-, iso-, or hyperintense relative to that of the adjacent normal pancreatic parenchyma.
Statistical Analysis
The Wilcoxon signed rank test was used to determine whether the calculated mean enhancement in tumors and normal pancreas and whether the tumor-to-pancreas contrast on the biphasic CT images significantly differed between AP and PVP CT images. For the analysis of the scoring of image quality with each technique, the quantitative image-grading data for each technique were analyzed with the Kruskal-Wallis test, followed by the Wilcoxon signed rank test.
Before ROC analysis, interobserver agreement in image interpretation with each kind of CT and MR image was assessed to establish the reliability of image interpretation in this study. The degree of interobserver agreement between each combination of two readers was also calculated with chance-corrected 
statistics. In general, a statistic of greater than 0.75 indicates excellent agreement; 0.40–0.75, fair to good agreement; and less than 0.40, poor agreement (22).
For ROC analysis, ROC curves were created for each kind of CT and MR image. The findings were analyzed by means of the ROCFIT (Metz CE, University of Chicago, Ill) maximum likelihood estimation of a binomial ROC curve grading data (23). The diagnostic accuracy of each CT and MR imaging technique for each reader was evaluated by calculating the area under the ROC curve, Az. Factors with Az values of greater than 0.80 were considered to have good diagnostic accuracy on the basis of results of a previous study in which ROC analysis was used (23). Az values for each CT and MR imaging technique were calculated from an ROC curve for each reader and were compared statistically by using the paired t test. Differences between the ROC curves of individual readers were not tested because we used multiple portions from each patient in image interpretation.
When a portion of the pancreas was assigned a grade of 4 or 5 (tumor probably or definitely present), a tumor was considered to be present, and a relative sensitivity value was calculated. On biphasic CT and MR images that depicted tumors, the tumor attenuation or signal intensity relative to that of the normal pancreatic parenchyma was also assessed for each reader. The significance of the differences in the relative sensitivity of each type of CT and MR image and the results of tumor attenuation on biphasic CT images were subsequently estimated by using the McNemar test (24).
For all tests, a P value of .05 indicated a statistically significant difference.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
|
values that indicated confidence levels in image interpretation between the three readers are summarized in Table 3. The values were excellent ( > 0.75 for AP CT and fat-saturated T1-weighted images between readers 1 and 2 and for delayed enhanced T1-weighted images between readers 2 and 3) or fair to good ( = 0.40–0.75) in the interpretation of all images. The values for the T2-weighted images between all readers and for the dynamic MR images for reader 1 versus reader 2 tended to be poor ( < 0.40) compared with the values for the other kinds of images.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Regarding tumor detection with MR images, tumors that could not be detected on fat-saturated T1-weighted or delayed enhanced T1-weighted images were also not detected on other MR images, including T1-weighted, T2-weighted, and dynamic enhanced images. Three tumors were detected on only fat-saturated T1-weighted images by all readers, and one tumor was detected on only delayed enhanced T1-weighted images by readers 1 and 2 (Fig 7). In nine patients with nine tumors who underwent both fat-saturated T1-weighted and delayed enhanced T1-weighted imaging, two tumors were detected on only fat-saturated T1-weighted images, one tumor by readers 2 and 3 and two tumors by reader 1. (The same tumor was detected by readers 2 and 3. The tumor detected by readers 2 and 3 was one of the tumors detected by reader 1.) Therefore, the relative sensitivity obtained with the combination of fat-saturated T1-weighted and delayed enhanced T1-weighted images was 78% (seven of nine lesions) for all readers.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The results of previous studies (11,18–20) have been controversial with regard to the phase of enhanced biphasic CT imaging that is most reliable for the depiction of pancreatic neoplasms in terms of tumor-to-pancreas contrast. For pancreatic islet cell tumors, Stafford-Johnson et al (11) reported that, in six nonfunctioning tumors, both normal pancreatic enhancement and tumor-to-pancreas contrast were greater on AP CT images than on PVP CT images. Although our findings tended to be similar to theirs, we found no significant difference between AP and PVP CT images in terms of normal pancreatic and tumor enhancement and tumor-to-pancreas contrast. Moreover, on the basis of ROC analysis of the qualitative data for pancreatic islet cell tumor detection, we found that PVP CT images were significantly superior to AP CT images. Therefore, despite the earlier reports of greater tissue enhancement with AP CT imaging, our data suggest that AP CT images are less helpful than PVP CT images.
In the angiography literature (2), it was reported that the majority of both primary pancreatic islet cell tumors and hepatic metastases are expected to be hyperattenuating on AP CT images. However, several tumors in our series were hypo- or isoattenuating relative to the enhanced pancreas on the AP CT images and were not be depicted on these images. This unexpected pattern of enhancement may be one of the major reasons for our less favorable results with AP CT images. Van Hoe et al (9) and Keogan et al (19) also found that 45% (five of 11) and 40% (two of five) of islet cell tumors did not show hyperattenuation. Previous investigators (9) also reported that isoattenuating islet cell tumors on AP CT images are sometimes detected on only PVP CT images. Radiologists should also recognize that the effective depiction of hyperattenuating islet cell tumors on AP CT images depends on the timing of scanning relative to the injection of contrast material.
Regarding the optimal CT scanning phase when imaging the pancreas, Lu et al (21) have advocated the "pancreatic phase." They emphasized that images obtained in this phase demonstrated substantially greater normal pancreatic parenchymal enhancement and tumor-to-pancreas contrast in pancreatic ductal adenocarcinoma than did PVP CT images. As this pancreatic phase was defined at 40–70 seconds after the bolus administration of a contrast agent at a rate of 3 mL/sec, the pancreatic phase is considered to occur in the interval between the AP and the PVP, slightly overlapping the PVP. Whether images obtained during this pancreatic phase better depict islet cell tumors, hepatic metastases from islet cell tumors, and hypovascular pancreatic ductal adenocarcinomas than do biphasic CT images is an interesting concept that merits further study. Furthermore, the elimination of one phase from biphasic CT scanning would also have the benefit of reducing the patient's radiation exposure.
The sensitivity of MR imaging in the detection of pancreatic islet cell tumors has been reported in prior studies (12–17) and ranges widely, from 24% to 100%, probably because of the small patient series and variations in the imaging sequences used. Moreover, tumor detection may be affected by selection bias, since the majority of these studies predominantly included functioning islet cell tumors, which often have readily apparent clinical symptoms.
In our study, fat-saturated T1-weighted and delayed enhanced T1-weighted images obtained 5–10 minutes after the administration of gadolinium-based contrast material proved to be useful in the detection of islet cell tumors. Our successful results with tumor detection with delayed enhanced T1-weighted images may be due to the fact that some islet cell tumors (nine [35%] of 26 lesions) in our series were proved at histopathologic examination to be scirrhous tumors with abundant fibrous tissue. It has been reported that some hepatic tumors, especially cholangiocarcinoma, with fibrous tissue components show prolonged enhancement on delayed-phase CT or MR images (25–27). However, this hypothesis is highly speculative because, to our knowledge, such features have not been previously reported in pancreatic lesions, and, in this study, we did not strictly discuss with a pathologist the correlation of imaging features with the scirrhous findings in pancreatic islet cell tumors.
Our study has several noteworthy limitations. First, the study population was small and did not include healthy control subjects. If control subjects were included in this study, there would have been more of a real-life situation. Second, the lack of assessment of hepatic metastases was a major limitation in this study. It was easier to identify a malignant pancreatic islet cell tumor when the readers knew that the patient had hepatic metastases. Third, the study population had different types of islet cell tumors, and nonfunctioning islet cell tumors tended to be larger than functioning ones. Obviously, small tumors are harder to detect, and the more nonfunctioning (particularly if malignant) islet cell tumors there were, the more likely they would be depicted at imaging. Therefore, if the study population were homogeneous, our results may have been different.
This retrospective study has an obvious bias, since all readers knew that each patient had a pancreatic tumor. In addition, the study may have had a selection bias, as it was unknown why patients in the population underwent both biphasic CT and MR imaging and why some patients underwent MR imaging with certain pulse sequences (particularly those that were proved to be best, namely, fat-saturated T1-weighted and delayed enhanced T1-weighted sequences) while others did not. For example, if the initial CT or MR imaging examinations (eg, T1-weighted and T2-weighted imaging) demonstrated large lesions or hepatic metastases, the patients did not undergo further examination or MR imaging with other pulse sequences (eg, fat-saturated T1-weighted and delayed enhanced T1-weighted sequences). Therefore, our results cannot indicate the absolute sensitivity, specificity, or accuracy of CT or MR imaging in the prospective diagnosis of pancreatic islet cell tumors.
It has been reported that AP CT images are of particular value in the detection of hepatic metastases from islet cell tumors. Therefore, AP CT images cannot be justifiably eliminated from CT protocols for patients with islet cell tumors, especially malignant tumors, although the images may be of lesser value in the detection of primary pancreatic lesions. For this reason, the timing of scanning at pancreatic AP CT imaging was identical to that of hepatic imaging; this practice allowed us to assess hepatic metastases and to evaluate primary pancreatic tumors, but the scanning timing used at AP CT may not have been optimized for pancreatic imaging. However, this problem may be solved in the future by using a multidetector-array CT scanner to obtain images during an early (hepatic arterial) AP, late (pancreatic) AP, and a PVP.
Some of the MR imaging techniques used in this study may be somewhat out of date, especially when we consider a recent study of state-of-the-art fast MR imaging techniques for pancreatic tumors (28). Our discouraging results in islet cell tumor detection with T2-weighted and dynamic MR imaging may be associated with poorer image quality and large interobserver variation in image interpretation. Moreover, the 10-mm-thick sections used at MR imaging may not be fairly compared with the 5-mm-thick sections used at biphasic CT images used in this study.
In conclusion, biphasic CT and MR imaging have similar effectiveness in the detection of primary pancreatic islet cell tumors. With the reliability of biphasic CT in the detection of pancreatic islet cell tumors, the improved results with PVP CT images compared with AP CT images indicate that the use of AP CT could be limited to those patients who require an evaluation for hepatic metastases. However, a much larger series will be needed to finally resolve the issue of the utility of the AP CT images.
Regarding MR imaging protocols, both fat-saturated T1-weighted and delayed enhanced T1-weighted imaging should be included. Delayed enhanced T1-weighted images are especially valuable in the detection of the scirrhous type of malignant islet cell tumors. Because of the rarity of islet cell tumors and the rapid evolution of CT (eg, multidetector-array CT scanners) and MR imaging techniques (eg, fast MR imaging), the best imaging technique will remain difficult to define. We believe that our results should prove to be useful in the evaluation of current imaging strategies; they may provide a basis for future prospective and comparative evaluations.
|
FOOTNOTES |
|---|
Author contributions: Guarantor of integrity of entire study, T.I.; study concepts, R.L.B., M.P.F.; study design, T.I.; definition of intellectual content, T.I., M.P.F.; literature research, T.I.; clinical studies, M.S.P., H.H., S.N.; data acquisition, T.I., M.S.P., H.H., Y.K.; data analysis, T.I.; statistical analysis, T.I.; manuscript preparation, T.I.; manuscript editing, T.I., M.P.F.; manuscript review, M.S.P., M.P.F., R.L.B., T.A.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  S. Herwick, F. H. Miller, and A. L. Keppke MRI of islet cell tumors of the pancreas. Am. J. Roentgenol., November 1, 2006; 187(5): W472 - W480. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. F. Scarsbrook, R. V. Thakker, J. A. H. Wass, F. V. Gleeson, and R. R. Phillips Multiple Endocrine Neoplasia: Spectrum of Radiologic Appearances and Discussion of a Multitechnique Imaging Approach. RadioGraphics, March 1, 2006; 26(2): 433 - 451. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. M. Horton, R. H. Hruban, C. Yeo, and E. K. Fishman Multi-Detector Row CT of Pancreatic Islet Cell Tumors. RadioGraphics, March 1, 2006; 26(2): 453 - 464. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P Manoharan and M B Sheridan Neoplasms of the pancreas Imaging, September 1, 2004; 16(4): 323 - 337. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. A. Kaltsas, G. M. Besser, and A. B. Grossman The Diagnosis and Medical Management of Advanced Neuroendocrine Tumors Endocr. Rev., June 1, 2004; 25(3): 458 - 511. [Abstract] [Full Text] [PDF]
|
|
|
|
