DOI: 10.1148/radiol.2442060909
(Radiology 2007;244:583-590.)
© RSNA, 2007
Semiautomated MIP Images Created Directly on 16-Section Multidetector CT Console for Evaluation of Living Renal Donors1
Anand Kumar Singh, DMRD,
Dushyant V. Sahani, MD,
Christopher R. Kagay, MD,
Sanjeeva P. Kalva, MD,
Mukta C. Joshi,
Nahel Elias, MD, and
Tatsuo Kawai, MD
1 From the Departments of Radiology (A.K.S., D.V.S., C.R.K., S.P.K.) and Surgery (N.E., T.K.), Massachusetts General Hospital, Harvard Medical School, 55 Fruit St, White 270, Boston, MA 02114; and Department of FCT Engineering, GE Healthcare, Belmont, Mass (M.C.J.). D.V.S has received Research Grant support from GE Healthcare and has a consultant agreement with Bracco Diagnostics. Received May 24, 2006; revision requested July 24; revision received September 6; accepted October 12; final version accepted January 3, 2007.
Address correspondence to D.V.S. (e-mail: dsahani{at}partners.org).
 |
ABSTRACT
|
|---|
Institutional Review Board approval was obtained and informed consent was waived for this HIPAA-compliant study. The aim of this study was to retrospectively compare the accuracy of semiautomated maximum intensity projection (MIP) images created at a 16-section multidetector CT console with three-dimensional (3D)–workstation-generated images for the definition of renal donor anatomy, with intraoperative findings as a reference standard. In examining 40 renal donors (21 men and 19 women; age range, 24–56 years; mean age, 40.4 years), the sensitivity and accuracy for mapping donor anatomy by two readers were greater than 95%, interobserver agreement was excellent (
= 0.89–1.00). The 95% confidence interval for sensitivity was also calculated. Simple MIPs compared well with 3D–workstation images. MIPs from a predesigned protocol on the scanner console were generated more quickly than similar images from 3D workstations; postprocessing demands (eg, for renal donors) can be quickly fulfilled at the scanner console itself. The average time to generate simple MIPs at the console was 3.4 minutes (range, 1.7–4.4 minutes), and 22.3 minutes (range, 15–30 minutes) to create images at the 3D workstation.
© RSNA, 2007
|
INTRODUCTION
|
|---|
The limited surgical visibility and exposure associated with laparoscopic removal of kidneys is a substantial technical challenge for surgeons, making preoperative computed tomography (CT) evaluation of donor anatomy critical (1,2). Multidetector CT plays a key role in preoperative evaluations, with maximum intensity projection (MIP) and volume-rendered (VR) images being used preoperatively to define renal vasculature in many centers (3–7). The isotropic voxel resolution and high contrast material–enhancement of vessels in multidetector CT thin transverse data sets aid in better postprocessing of three-dimensional (3D) images (8–10).
Generation of reconstructed 3D images generally requires an expensive setup, including a dedicated 3D workstation and a specially trained technologist or radiologist. With the advent of faster multidetector CT scanners, there has been exponential growth in the number of imaging studies that need postprocessing, adding to the ever-increasing workload placed on workstations. Further, the process of image transfer from the scanner and the image generation by the technologist at the 3D workstation delays the availability of these images to the radiologist for interpretation (11).
Given these limitations affecting traditional 3D reconstructions, we created a semiautomated system to quickly deliver easily constructed, limited 3D images to the radiologist without the delays inherent in traditional 3D reconstruction systems (8).
Thus, the purpose of our study was to retrospectively compare the accuracy of semiautomated MIP images created at a 16-section multidetector CT console with 3D-workstation–generated images for the definition of renal donor anatomy, using intraoperative findings as a reference standard.
|
MATERIALS AND METHODS
|
|---|
Our institutional review board approved our study with waiver of informed consent. The study was in compliance with the Health Insurance Portability and Accountability Act. The equipment for the study was provided by GE Healthcare (Xtream; GE Healthcare, Chalfont, St Giles, England). The authors who are not employees of GE Healthcare (A.K.S., D.V.S., S.P.K., N.E., and T.K.) had control over inclusion of any data and information submitted for publication.
Renal Donors
This was a retrospective analysis of the preoperative multidetector CT findings of 40 consecutive renal donors who underwent donor nephrectomy from October 2003 to March 2005 (21 men, 19 women; aged 24–56 years; mean age, 40.4 years).
Multidetector CT Technique
The CT examinations were performed on a 16-section multidetector CT scanner (Light Speed 16; GE Healthcare). No oral contrast material was administered and unenhanced CT was performed from vertebrae T12 through L5. This was followed by an injection of nonionic contrast material (300 mg iodine per milliliter; iohexol, Isovue300; Bracco Diagnostics, Princeton, NJ) administered at a dose of 2 mL/kg and at a rate of 4–5 mL/sec. The maximum permissible limit with respect to injectable volume of contrast material was 150 mL. A late arterial phase image was obtained with a bolus tracking technique (Smart Prep; GE Healthcare) to visualize both arteries and veins; the scan was initiated after a 150-HU (Hounsfield unit) enhancement threshold was achieved in the aorta at the level of the superior mesenteric artery. The arterial phase scanning was performed by using a section thickness of 1.25 mm, intersection spacing of 0.6 mm, and table speed of 9.37 mm per rotation. Excretory phase scanning was performed by using a section thickness of 2.5 mm, intersection spacing of 1.25 mm, and table speed of 18.75 mm per rotation. Constant parameters were 0.5-second gantry rotation, 120-140 kV, 75–380 mA (automatic tube current modulation), and pitch of 0.938.
Image Postprocessing
MIPs from console.—The 16-section multidetector CT scanner console enables creation of predesigned protocols consisting of image reformations in desired planes using algorithms like MIPs or minimum intensity projections. These protocols can be assigned a name and saved according to sections of different body parts to allow easy retrieval from a database when needed. Image processing with the designed protocol can then be facilitated with the use of few button clicks.
The desired protocol is then applied to the selected sections of thin transverse data sets. A "batch film" (a small window showing options for saving one series and running the next series) and a "layout" (four-screen format of two transverse, one coronal, and one sagittal image of the chosen data set), where different acquisition slabs appear automatically after clicking the "next" button on a batch film, are generated. The batch film also displays field of view, spacing used in planning, and section thickness used for the predesigned protocol.
The arterial and excretory phase transverse data sets of the 40 donors were retrieved from the picture archiving and communication system (PACS version 4.0; Agfa, Richmond, Va) and transferred to the Xtream scanner console for postprocessing. MIP image series were then generated by a radiology fellow (A.K.S., 1 year experience in MIP imaging) using a predesigned "renal donor" protocol (Table 1), which consisted of slabs from which MIP images of renal vasculature (straight coronal, transverse, and both oblique coronal planes) and the collecting system (one oblique coronal plane) were acquired. Oblique coronal MIPs were planned to produce images parallel to the obliquity of the renal vessels, thereby ensuring visualization of the entire length and course of renal vessels. Oblique coronal MIPs on delayed phase were planned parallel to the renal axis on a delayed phase sagittal image to check for any anomaly or duplication of the collecting system.
Image processing: 3D workstation.—Image reconstructions were also performed on a workstation (Advantage 3.1; GE Healthcare) where two trained technologists (with 5 years and 11 months of image processing experience) used volumetric multidetector CT data sets for reconstruction. Two-dimensional and 3D VR and MIP maps (screen shots of single images) of renal arteries, renal veins, and the collecting system were generated in five to seven angles. For delineation of the collecting system, three coronal and oblique coronal MIP and VR screen captures (single images) were obtained. Field of view range of 18–26 cm was used to obtain screen captures of MIP (2.5-mm thick) and VR (1-mm thick, with 0.6-mm spacing) images. By using the VR tools from the 3D viewport, opacity and color of the VR images were adjusted and parameters were optimized for bone removal. Screen captures at different angles were available by moving the bounding box (3D control box) on the viewport.
Image Analysis
Image analysis was performed on an Xtream 16-section multidetector CT scanner console by two independent, experienced genitourinary radiologists (D.V.S., 13 years experience; S.P.K., 10 years experience) and two transplant surgeons with genitourinary experience (T.K., 20 years experience; N.E., 6 years experience). The readers were blinded to observations made at each donor surgery and the prospective interpretations of the donor CT examinations. A predefined template was used to systematically review the imaging data sets. The 16-section multidetector CT console MIPs were presented first to each radiologist, immediately followed by the 3D-workstation–generated images from the same donor. The two radiologists used a subjective five-point scale to grade both the image quality (1 = unacceptable, 2 = suboptimal, 3 = acceptable, 4 = good, 5 = excellent) and diagnostic confidence (1 = poor, 2 = low, 3 = intermediate, 4 = high, 5 = excellent) of the console MIPs. Subsequently, they recorded details of renal vascular anatomy, documenting the number of renal arteries, veins, and their variants, including early bifurcation of the vessels. The delayed urographic phase images were then graded on a subjective three-point scale (1 = poor, 2 = good, 3 = excellent) for the collecting system delineation and quality of ureteric opacification. Any difference in diagnostic confidence for the findings on donor anatomy from the 3D workstation images and simple console MIPs was also noted.
To test the acceptability of the console MIPs before surgery, two transplant surgeons were presented with the image data sets in the same manner that the radiologists were. Using the same scale, they graded the console MIPs for image quality, diagnostic confidence, and their adequacy for surgery. The need for reviewing transverse image sets after review of 3D images by the surgeons was also noted.
Standard of Reference
The intraoperative findings from each donor nephrectomy with respect to the renal vasculature and the collecting system constituted the standard of reference for the imaging findings. One of the authors (A.K.S.) retrieved surgical reports of all 40 donors from the institution's clinical database and compared the findings of the operated kidney with the interpretations made by readers. The findings of the unoperated kidney could not be confirmed with the surgical report. Since there is no standard of reference, comparison of both image sets was limited to the evaluation of their agreement.
Time for Image Generation
A record of the time taken for 3D image creation by a trained technologist from the workstation was obtained from a 3D workstation log sheet, while a research fellow maintained record of time for generation of console MIPs.
Statistical Analysis
By using the surgical findings as the standard of reference, the sensitivity for detection of the renal arteries and veins, and the sensitivity, specificity, and accuracy for the categorization of the collecting system and variants were calculated for MIPs (from both console and 3D workstation image sets) based on the findings of the two radiologists. It should be noted that specificity and accuracy are not defined for the detection of renal arteries and veins since the number of true-negatives is not definable. SAS statistical software (version 9.1, 2002–2003; SAS Institute, Cary, NC) was used to calculate the weighted values and the confidence intervals for sensitivity for their findings. The McNemar test for paired data was used to test for differences between the methods in the number of arteries and veins and for categorization of the collecting system. A P value of .05 or less was considered to indicate a statistically significant difference.
|
RESULTS
|
|---|
Renal Vasculature
Surgical records revealed a total of 55 arteries and 46 veins in the 40 removed kidneys (Fig 1). After review of console MIPs, radiologist 1 detected 54 of 55 renal arteries on the operated side, and radiologist 2 detected 53 of 55 (Fig 2). Both failed to detect an accessory renal artery 1.5 mm in diameter on the left side, which was found during surgery. Radiologist 2 also missed another left accessory renal artery 2 mm in diameter. Both radiologists detected all 46 veins at review of console MIPs. All six accessories and variants such as one prominent lumbar perforator vein and one retroaortic renal vein were accurately detected by both radiologists on console MIPs (Table 2; Fig 3). The number of arteries and veins detected in each of the unoperated kidneys using the console MIP image sets was identical to the numbers detected during review of 3D workstation image sets for each radiologist.
View larger version (163K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2a: CT angiographic images show right kidney in 56-year-old female donor. (a) Reconstruction design used at 16-section multidetector CT console for generation of 6-mm straight coronal MIPs. (b) Console-generated coronal MIP shows three arteries (straight arrows) supplying right kidney, branching of right main renal artery (arrowhead), and two renal veins (curved arrows). (c) Coronal volume MIP image from 3D workstation shows same findings.
|
|
View larger version (135K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2b: CT angiographic images show right kidney in 56-year-old female donor. (a) Reconstruction design used at 16-section multidetector CT console for generation of 6-mm straight coronal MIPs. (b) Console-generated coronal MIP shows three arteries (straight arrows) supplying right kidney, branching of right main renal artery (arrowhead), and two renal veins (curved arrows). (c) Coronal volume MIP image from 3D workstation shows same findings.
|
|
View larger version (89K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2c: CT angiographic images show right kidney in 56-year-old female donor. (a) Reconstruction design used at 16-section multidetector CT console for generation of 6-mm straight coronal MIPs. (b) Console-generated coronal MIP shows three arteries (straight arrows) supplying right kidney, branching of right main renal artery (arrowhead), and two renal veins (curved arrows). (c) Coronal volume MIP image from 3D workstation shows same findings.
|
|
View larger version (166K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3a: CT angiographic images show left kidney in 32-year-old female donor. (a) Reconstruction design used at 16-section multidetector CT console for generation of 6-mm oblique coronal MIPs, parallel to obliquity of renal vasculature. (b) Console-generated oblique coronal MIP shows accessory artery (black arrow) and vein (white arrow) at lower pole. Accessory venous origin is left common iliac vein. (c) Coronal volume MIP image from 3D workstation shows same findings.
|
|
View larger version (163K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3b: CT angiographic images show left kidney in 32-year-old female donor. (a) Reconstruction design used at 16-section multidetector CT console for generation of 6-mm oblique coronal MIPs, parallel to obliquity of renal vasculature. (b) Console-generated oblique coronal MIP shows accessory artery (black arrow) and vein (white arrow) at lower pole. Accessory venous origin is left common iliac vein. (c) Coronal volume MIP image from 3D workstation shows same findings.
|
|
View larger version (158K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3c: CT angiographic images show left kidney in 32-year-old female donor. (a) Reconstruction design used at 16-section multidetector CT console for generation of 6-mm oblique coronal MIPs, parallel to obliquity of renal vasculature. (b) Console-generated oblique coronal MIP shows accessory artery (black arrow) and vein (white arrow) at lower pole. Accessory venous origin is left common iliac vein. (c) Coronal volume MIP image from 3D workstation shows same findings.
|
|
Pelvicaliceal System and Ureter
After viewing the console-generated coronal MIPs for the delayed phase scan, both radiologists detected normal pelvicaliceal and ureteric anatomy in all but one donor, where duplication of the left collecting system was seen (Fig 4). This was also well seen by both radiologists on the 3D workstation images, which did not provide any additional information about the pelvicaliceal system.
View larger version (136K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 4a: CT urographic images of 56-year-old female donor. (a) Reconstruction design used at 16-section multidetector CT console for generation of 7-mm oblique coronal MIPs. (b) Console-generated MIP shows incomplete duplication of left ureter (arrow). (c) Same finding is shown in coronal 3D workstation MIP.
|
|
View larger version (137K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 4b: CT urographic images of 56-year-old female donor. (a) Reconstruction design used at 16-section multidetector CT console for generation of 7-mm oblique coronal MIPs. (b) Console-generated MIP shows incomplete duplication of left ureter (arrow). (c) Same finding is shown in coronal 3D workstation MIP.
|
|
View larger version (99K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 4c: CT urographic images of 56-year-old female donor. (a) Reconstruction design used at 16-section multidetector CT console for generation of 7-mm oblique coronal MIPs. (b) Console-generated MIP shows incomplete duplication of left ureter (arrow). (c) Same finding is shown in coronal 3D workstation MIP.
|
|
Statistical Analysis
The overall sensitivity, specificity, accuracy, and values for detection of renal vessels and their variants and the collecting system for both radiologists at review of the console MIPs and 3D workstation image sets were the same (Table 3).
The test for the difference between the detection of the number of arteries and veins, and the anatomy of the collecting system on 16-section multidetector CT console MIPs and 3D workstation images was not statistically significant (P > .99).
Qualitative Assessment
In all cases, the image quality of console MIPs was consistently rated as either good (grade 4) or excellent (grade 5) by both radiologists and surgeons on a five-point scale (mean grade = 4.4). The diagnostic confidence of these assessors was either high (grade 4) or excellent (grade 5) for these cases (mean grade = 4.6). Image quality and diagnostic confidence for ureteric delineation on console MIPs were rated as excellent (grade 5) by both radiologists and surgeons. Console MIPs were graded better than 3D workstation images for diagnostic confidence on a few occasions (radiologist 1, n = 3; radiologist 2, n = 2; surgeon 2, n = 2); 3D workstation images were rated better than console MIPs in other cases (radiologist 1, n = 1; surgeon 1, n = 1; surgeon 2, n = 2). Adequacy of console MIPs for surgery was rated as good (grade 4) to excellent (grade 5) by both surgeons on both image sets on all occasions except for three donor console MIPs and their corresponding 3D workstation images, which were rated as acceptable (grade 3) for surgery by surgeon 1 (average grading: surgeon 1, 3.9; surgeon 2, 4.8). These three donors had larger body size and the CT data sets had greater than expected image noise for the selected noise index. The surgeons in some cases (surgeon 1, n = 5; surgeon 2, n = 3) reviewed transverse images carefully to confirm early arterial branching.
Time to Create Images
The average time to generate simple MIPs at the console was 3.4 minutes (range, 1.7–4.4 minutes), and 22.3 minutes (range, 15–30 minutes) to create images at the 3D workstation.
|
DISCUSSION
|
|---|
In our study, we compared console-generated MIPs with 3D-workstation–generated images.
The console MIP findings, with respect to detection of renal arteries by radiologists, were found to be similar to those in multidetector CT studies by Kawamoto et al (12) (accuracy, 93%; accuracy range for three readers, 89%–97%) and Sahani et al (1) (accuracy, 94%).
The MIP technique is excellent for emphasizing high-attenuation structures such as arteries and can provide 3D information if multiple MIPs are generated with different viewing angles and then flipped over or displayed as a cine loop (8,13). Hence, we designed the protocol to create a set of MIP images in various planes: the straight coronal, oblique coronal parallel to the obliquity of the renal vessels, and thin and better-focused transverse MIPs with decreased field of view. Although VR images are aesthetically more pleasing and enable better 3D perception than do MIPs, we refrained from incorporating VR in our protocol, owing to the large computational burden that makes the procedure cumbersome and time consuming (14,15).
With respect to renal veins, console MIPs and 3D image set findings were concordant. The sensitivity was 100% (46 of 46 veins) for both radiologists, with an excellent value of 1.0. This result was better than the previous multidetector CT study by Sahani et al (1) (sensitivity, 75%; nine of 12), and similar to that of Kim et al (10), who revealed a depiction rate of 98% (83 of 85) veins. From the surgeon's perspective, appropriate venous variants such as lumbar vein prominence must be identified and localized (16). Owing to the use of an automatic bolus triggering technique for obtaining late arterial phase images with our CT scanners, the effects on interpretation of both sets of images were accurate and specific for detection of venous variants (5,17).
In a few cases, the surgeons needed transverse images for confirmation of early arterial branching. When images are presented in multiple planes, some distortion of the true anatomic relationship of the structures does occur and should therefore be interpreted together with the transverse images (18).
Our study revealed that MIPs from a predesigned protocol on the scanner console were quicker to generate than similar images from 3D workstations. We believe that 3D workstations are appropriate for complex postprocessing demands, like organ volume estimations, while postprocessing demands such as those for renal donors can be quickly fulfilled at the scanner console itself. The postprocessing can either take place on the same console after scanning the patient or on a separate 16-section multidetector CT console.
There were limitations to our study. The study was retrospective and the donor cohort was small. The donor kidney with normal or less complicated vascular anatomy was chosen by the surgeon for surgery. Hence, we could not always compare the more complicated variants detected on console MIPs and 3D workstation images with those seen at surgery. Sequential readouts were performed by readers, where console MIPs for donor examination were presented first, followed by a 3D workstation image set for the same examination. Reviewing both image sets in a blinded and separate fashion would have ensured optimal control over the reader's bias. Confidence intervals for sensitivity for detection of arteries and veins are only approximate, as they were calculated on the basis of an assumption of independence, whereas in truth there may be data clustering due to multiple arteries in the same patient. The result may be manufacturer and model specific since we used data only from a Lightspeed 16 scanner (GE Healthcare) and processed it on the Xtream console (GE Healthcare).
In conclusion, our study revealed that the MIPs from a predesigned protocol on the scanner console were much quicker to generate than similar images from 3D workstations, and postprocessing demands (eg, the needs of renal donors) can be quickly fulfilled at the scanner console itself.
|
ADVANCES IN KNOWLEDGE
|
|---|
- Three-dimensional (3D) CT angiographic images for renal donors can be directly created at the 16-section multidetector CT scanner console.
- In detection of renal vascular and collecting system variants, console-generated 3D images are as accurate as the 3D workstation images that are created by using advanced tools.
|
IMPLICATION FOR PATIENT CARE
|
|---|
- In this study, one person generated the images at the console, taking an average of 3.4 minutes, thus reducing the image-generation time compared with that for 3D workstations; this can improve CT workflow for patient care.
|
ACKNOWLEDGMENTS
|
|---|
The authors acknowledge the input of Elkan Halpern, PhD, for statistical analysis.
|
FOOTNOTES
|
|---|
Abbreviations: MIP = maximum intensity projection • 3D = three-dimensional • VR = volume rendered
See Materials and Methods for pertinent disclosures.
Author contributions: Guarantor of integrity of entire study, D.V.S.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; approval of final version of submitted manuscript, all authors; literature research, A.K.S.; clinical studies, D.V.S., S.P.K., N.E., T.K.; statistical analysis, A.K.S.; and manuscript editing, all authors.
|
References
|
|---|
- Sahani DV, Rastogi N, Greenfield AC, et al. Multi-detector row CT in evaluation of 94 living renal donors by readers with varied experience. Radiology 2005;235(3):905–910.[Abstract/Free Full Text]
- Ratner LE, Kavoussi LR, Sroka M, et al. Laparoscopic assisted live donor nephrectomy: a comparison with the open approach. Transplantation 1997;63(2):229–233.[CrossRef][Medline]
- Del Pizzo JJ, Sklar GN, You-Cheong JW, Levin B, Krebs T, Jacobs SC. Helical computerized tomography arteriography for evaluation of live renal donors undergoing laparoscopic nephrectomy. J Urol 1999;162(1):31–34.[CrossRef][Medline]
- Patil UD, Ragavan A, Nadaraj, et al. Helical CT angiography in evaluation of live kidney donors. Nephrol Dial Transplant 2001;16(9):1900–1904.[Abstract/Free Full Text]
- El Fettouh HA, Herts BR, Nimeh T, et al. Prospective comparison of 3-dimensional volume rendered computerized tomography and conventional renal arteriography for surgical planning in patients undergoing laparoscopic donor nephrectomy. J Urol 2003;170(1):57–60.[CrossRef][Medline]
- Behar JV, Nelson RC, Zidar JP, DeLong DM, Smith TP. Thin-section multidetector CT angiography of renal artery stents. AJR Am J Roentgenol 2002;178(5):1155–1159.[Abstract/Free Full Text]
- Prokop M. General principles of MDCT. Eur J Radiol 2003;45(suppl 1):S4–S10.[CrossRef][Medline]
- Kirchgeorg MA, Prokop M. Increasing spiral CT benefits with postprocessing applications. Eur J Radiol 1998;28(1):39–54.[CrossRef][Medline]
- Costello P, Gaa J. Spiral CT angiography of abdominal aortic aneurysms. RadioGraphics 1995;15(2):397–406.[Abstract]
- Kim JK, Kim JH, Bae SJ, Cho KS. CT angiography for evaluation of living renal donors: comparison of four reconstruction methods. AJR Am J Roentgenol 2004;183(2):471–477.[Abstract/Free Full Text]
- van Elzakker T. Current perspectives on CT technology. Radiol Manage 2004;26(5):54–58.[Medline]
- Kawamoto S, Montgomery RA, Lawler LP, Horton KM, Fishman EK. Multidetector CT angiography for preoperative evaluation of living laparoscopic kidney donors. AJR Am J Roentgenol 2003;180(6):1633–1638.[Abstract/Free Full Text]
- Marks MP, Napel S, Jordan JE, Enzmann DR. Diagnosis of carotid artery disease: preliminary experience with maximum-intensity-projection spiral CT angiography. AJR Am J Roentgenol 1993;160(6):1267–1271.[Abstract/Free Full Text]
- Soyer P, Heath D, Bluemke DA, et al. Three-dimensional helical CT of intrahepatic venous structures: comparison of three rendering techniques. J Comput Assist Tomogr 1996;20(1):122–127.[CrossRef][Medline]
- Johnson PT, Heath DG, Kuszyk BS, Fishman EK. CT angiography with volume rendering: advantages and applications in splanchnic vascular imaging. Radiology 1996;200(2):564–568.[Abstract/Free Full Text]
- Raman SS, Pojchamarnwiputh S, Muangsomboom K, Schulam PG, Gritisch HA, Lu DS. Surgically relevant normal and variant renal parenchymal and vascular anatomy in preoperative 16-MDCT evaluation of potential laparoscopic renal donors. AJR Am J Roentgenol 2007;188(1):105–114.[Abstract/Free Full Text]
- Fishman EK, Magid D, Ney DR, et al. Three-dimensional imaging.Radiology 1991;181(2):321–337.[Free Full Text]
- Galanski M, Prokop M, Chavan A, Schaefer CM, Jandeleit K, Nischelsky JE. Renal arterial stenoses: spiral CT angiography. Radiology 1993;189(1):185–192.[Abstract/Free Full Text]
TOP
ABSTRACT
INTRODUCTION
