| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Gastrointestinal Imaging |
1 From the Department of Radiology, New York University Medical Center, MRI-Basement, Schwartz Bldg, 530 First Ave, New York, NY 10016 (N.M.R., V.S.L., M.A.P., G.A.K., M.M.A., J.C.W.), and Siemens Medical Systems (G.L., D.T.). Received October 1, 1998; revision requested November 11; revision received December 28; accepted April 1, 1999. Address reprint requests to N.M.R.
 |
Abstract |
|---|
 TOP Abstract IntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
MATERIALS AND METHODS: Twenty consecutive patients underwent 1.5-T MR imaging. The examinations included pre- and postcontrast (20 mL gadopentetate dimeglumine) fat-saturated 2D GRE breath-hold imaging and fat-saturated volumetric interpolated breath-hold imaging before, during (arterial phase), and after injection, with thin (2-mm source images) and thick (8-mm reconstruction images) sections. The three images were compared qualitatively and quantitatively (signal-to-noise ratio [SNR] and contrast-to-noise ratio [CNR]).
RESULTS: Qualitatively, the 2-mm source images had poorer pancreatic edge definition on precontrast images compared with the other two data sets (P < .05). On gadolinium-enhanced images, scores for clarity of pancreatic edge, number of vessels visualized, and arterial ghosting were significantly lower for the postcontrast 2D GRE images. Quantitatively, SNR measurements in the liver, aorta, and renal cortex on pre- and postcontrast images were significantly higher for the 8-mm reconstruction images than for the 2D GRE or 2-mm source images (P < .05). Aorta-to-fat CNR was significantly higher on the 8-mm reconstruction images.
CONCLUSION: Fat-saturated volumetric interpolated breath-hold images have quality comparable to that of conventional fat-saturated 2D GRE images.
Index terms: Abdomen, MR, 70.121412, 70.121415, 70.121419, 70.12143 • Magnetic resonance (MR), technology, 70.121412, 70.121415, 70.121419
|
Introduction |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
For 2D breath-hold techniques in the abdomen, the quality and efficacy of imaging is limited by the need to acquire enough sections to cover a relatively large region, typically at least 160–200 mm. This must be accomplished with the finite number of sections that can be obtained in less than 25 seconds and thus requires use of relatively thick sections (8–10 mm). With 2D GRE pulse sequences in the abdomen, subcentimeter lesions could be missed or insufficiently characterized owing to partial volume averaging and low contrast-to-noise ratio (CNR) (8).
Short acquisition times place serious constraints on imaging; increased anatomic coverage comes at the expense of decreased spatial resolution and may require use of higher bandwidths that reduce the signal-to-noise ratio (SNR). In addition, fat-saturation methods, which can improve CNR on contrast-enhanced images, require additional imaging time. The application of fat saturation to GRE imaging on a routine basis has not been popular since most strategies do not provide adequate anatomic coverage in the time frame of breath-hold imaging.
Three-dimensional (3D) Fourier transform GRE imaging has potential advantages over 2D imaging. In comparison with traditional 2D GRE sequences, properly structured 3D GRE sequences have the capacity to provide thinner sections, no gaps, fat saturation, higher SNR, and comparable image contrast in the same breath-hold time frame. Furthermore, with appropriately thin sections and accurate timing, the same data set could be used to generate high-quality MR angiograms and thus provide added value. Therefore, 3D GRE imaging has the potential to yield a comprehensive evaluation of the upper abdomen. However, previous efforts with 3D GRE imaging have been hindered by relatively limited resolution and anatomic coverage (6,7,9).
In this study, we evaluated an alternative approach to 3D GRE imaging that addresses those limitations. We call the technique "volumetric interpolated breath-hold examination," or VIBE. With volumetric interpolated breath-hold imaging, data sets that have nearly isotropic resolution in three dimensions (on the order of 2-mm voxel size) can be obtained while preserving adequate anatomic coverage and uniform fat saturation within a breath hold.
We propose the volumetric interpolated breath-hold examination is a versatile and efficient approach to body MR imaging. The goals of this study were to assess the feasibility of implementing the approach; to compare the SNR, CNR, and fat saturation of the sequence to those with a more conventional 2D GRE sequence; and to illustrate reconstruction and angiographic possibilities with this 3D volumetric imaging method.
|
MATERIALS AND METHODS |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Breath-hold 2D axial T1-weighted fat-saturated images and breath-hold 3D T1-weighted fat-saturated volumetric interpolated breath-hold images were obtained before contrast material administration. For all sequences, patients were instructed to suspend respiration at end expiration. According to the method detailed by Earls et al (10), a timing image was then obtained with a test dose of 1 mL gadopentetate dimeglumine (Magnevist; Berlex Laboratories, Wayne, NJ) followed by 20 mL saline solution, both injected at a rate of 2 mL/sec. This allowed estimation of patient circulation time (bolus transit time from arm vein to abdominal aorta) and was used to ensure acquisition of optimal arterial phase images.
Dynamic imaging was performed with 19 mL gadopentetate dimeglumine. During dynamic imaging, three identical acquisitions with the fat-saturated volumetric interpolated breath-hold sequence were performed in quick succession (20–30 seconds between each to allow patients to catch their breath). The first acquisition was timed for arterial enhancement. After the third acquisition, postcontrast 2D fat-saturated imaging (with parameters identical to those in the precontrast 2D fat-saturated sequence) was performed.
MR Sequences and Parameters
The field of view, craniocaudal coverage, and the number of in-plane phase-encoding steps were tailored for each patient, and they were kept constant in the 2D fat-saturated GRE and fat-saturated volumetric interpolated breath-hold sequences for each individual. A bandwidth of 488 Hz per pixel was used in both sequences.
Two-dimensional GRE.—This is a fast low-angle shot, or FLASH, GRE sequence in which periodic chemically selective fat-saturation pulses have been incorporated. The field of view was 300–370 mm with a rectangular configuration in the phase-encoding (anteroposterior) dimension. The rectangular field of view was modified to accommodate the individual's body habitus. An initial matrix of 160 x 256 was used to yield in-plane spatial resolution of 2.3 x 1.4 mm or less and was modified to 125 phase-encoding steps with the rectangular field of view.
A 2D fat-saturation strategy was used in which two fat-saturation pulses were applied every 20 sections. The first pulse was applied at the beginning of the line acquisition (phase encoding) for the first section, and the second was applied after the line data from 10 sections were obtained. This strategy was repeated until the matrix was complete. This yielded an effective combination of fat saturation and anatomic coverage (20 sections) in the time frame of a breath hold. Although this fat-saturation strategy was experimental, it has since become commercially available. The section thickness was 5 mm with no gap between sections (n = 3) or 8 mm (n = 17) with no gap between sections (n = 15) or a 20% gap (n = 2). This allowed craniocaudal coverage of 100, 160, and 200 mm, respectively.
The 2D sequence was performed with repetition time msec/echo time msec of 180/2.8 with 80° flip angle, and 20 sections were generated during a breath hold of less than 23 seconds. The precise acquisition time depended on the rectangular field of view used.
Fat-saturated volumetric interpolated breath-hold examination.—In this application, a 3D radio-frequency–spoiled GRE sequence with isotropic or nearly isotropic spatial resolution was implemented. For this purpose, a symmetric echo was used in the read direction with a bandwidth of 488 Hz per pixel and 160 readout points. The k space was filled with 160 points along ky and 40 points along kz. Interpolation was performed in the section-select direction, as shown in Figure 1, which resulted in 80 partitions. Each partition was interpolated to a 256 square matrix. This strategy was selected to reduce the voxel size and maintain a short imaging time. As is typical of 3D sequences, there was no intersection gap. The field of view was 300–370 mm with a rectangular configuration in the phase-encoding (anteroposterior) dimension matched to that in the 2D fat-saturated sequence of a particular patient. An initial matrix of 160 x 160 data points was used to yield in-plane spatial resolution of less than 2.3 x 2.3 mm depending on the choice of field of view. The imaging matrix was 256 x 256 with a pixel size of less than 1.5 mm. The slab thickness was 160–200 mm divided into 80 partitions with a partition thickness of 2.0–2.5 mm.
|
This sequence incorporated a sequential phase-encode order in the ky direction and a centric phase-encode reordering in the kz direction (partitions loop). A chemically selective fat-saturation pulse was applied just before each partitions loop. The partitions loop was centric reordered to optimize fat-saturation uniformity. This fat-saturation strategy has been available in a previous commercially available 3D sequence (Siemens Medical Systems).
The 3D sequence was performed with the following parameters: 4.2/1.8 with 12° flip angle, 40 partitions were generated in a breath hold of less than 24 seconds, depending on the rectangular field of view. The 12° flip angle was selected after comparing liver-to-spleen CNR with multiple flip angles in eight patients before this study (unpublished data). The 3D volumetric interpolated breath-hold examination was performed before, during, and after administration of gadopentetate dimeglumine, with 20 seconds between each breath-hold acquisition.
Image Processing
Postprocessing of the fat-saturated volumetric interpolated breath-hold images was performed on a commercially available MR workstation, a satellite console of the MR unit. Reformations of the data sets before and after contrast material administration were performed by one investigator (M.A.P) who was not involved with the qualitative comparison. These reformations included axial 2-mm-thick sections obtained at intervals of 5–8 mm to match the locations of the 20 sections obtained with the 2D fat-saturated sequence. The fat-saturated volumetric interpolated breath-hold data sets were also reformatted as 5–8-mm-thick sections (no gap) and matched in position to the 2D fat-saturated images.
In addition, sagittal and coronal multiplanar reconstruction images were generated and reviewed interactively on the console, and selected images were acquired as deemed necessary by the interpreting radiologist. Similarly, maximum intensity projection (MIP) images were generated interactively from postcontrast fat-saturated volumetric interpolated breath-hold images with use of restricted volumes of interest, and hard copies of selected MIP images were obtained.
Data Analysis: Quantitative Assessment of Image Quality
Blinded to imaging results, a single investigator (M.A.P.) performed region-of-interest analysis for each of the six sets of images for each of the 20 patients: pre- and postcontrast 2D fat-saturated images, pre- and postcontrast 2-mm fat-saturated volumetric interpolated breath-hold source images, and pre- and postcontrast 8-mm fat-saturated volumetric interpolated breath-hold reconstruction images. The postcontrast 2-mm data that were quantitatively analyzed were obtained in the third of three postcontrast acquisitions performed just before the postcontrast 2D fat-saturated sequence. This was done to minimize differences in contrast media dynamics.
For each patient, matching regions of interest were placed over the liver, pancreas, spleen, retroperitoneal fat, subcutaneous fat, renal cortex, aorta, and air outside the patient's body. Care was taken to place regions within normal-appearing portions of each organ and away from intraparenchymal blood vessels. Means and SDs of signal intensity were recorded. On the basis of these measurements, the following parameters were computed: liver, renal cortex, and aorta SNRs, and liver-to-spleen, liver-to–retroperitoneal fat, pancreas-to-fat, renal cortex-to-fat, and aorta-to-fat CNRs. Signal noise was estimated on the basis of the SD of signal intensity measured in the region of interest outside the patient's body. For each sequence of reformatted data sets, regions of interest were placed at comparable section positions.
CNRs were calculated as |SItissueA - SItissueB|/SDnoise, where SI is signal intensity; tissue A is liver, pancreas, renal cortex, or aorta; and tissue B is spleen or retroperitoneal fat.
Data Analysis: Qualitative Analysis of Image Quality
Three independent reviewers (N.M.R., V.S.L., G.A.K.), who were blinded to patient and sequence information, viewed sets of 20 images in a random fashion. Each set consisted of the 20 images obtained with a given sequence displayed on a single sheet of film. There were three precontrast and three postcontrast sets (2D fat-saturated images, 2-mm fat-saturated volumetric interpolated breath-hold source images, or 8-mm reconstruction images) for a total of six sets per patient. Thus, 120 images per patient were evaluated, 20 images at a time.
Image quality was assessed by scoring the following parameters on a scale of 1–5, with 5 representing the highest quality image or the least degradation by artifact. The parameters included clarity of liver edge, clarity of pancreas edge, ghosting, clarity of vessels, uniformity of fat saturation centrally (within the peritoneum and retroperitoneum) and peripherally (subcutaneous fat), number of vessels identified, and arterial ghosting.
The following scale was used to evaluate clarity of the liver and pancreas edge: 1, unreadable; 2, extreme blur; 3, moderate blur; 4, mild blur; and 5, sharp. For ghosting, the following scale was used: 1, unreadable study; 2, more than three ghosts; 3, two ghosts; 4, one ghost; 5, no ghosting. For fat saturation, the approximate percentage of area that demonstrated uniform fat saturation was graded as follows: 1, no fat saturation; 2, 0%–24% fat saturation; 3, 25%–49%; 4, 50%–74%; and 5, 75%–100%. For the number of vessels seen, readers were instructed to attempt to identify as many of the following vessels as possible: right, left, and middle hepatic veins; main portal vein; right and left portal veins; and right and left segmental veins (total of 10 vessels). The number identified was then converted to a scale of 1–5 as follows: 1, no vessels; 2, one to three vessels; 3, four to seven vessels; 4, seven to nine vessels; 5, 10 vessels.
Statitical Analysis
Comparisons of SNR and CNR with the three imaging sequences both before and after contrast material administration were made by means of one-way analysis of variance (EXCEL; Microsoft, Redmond, Wash). For those parameters with which a statistically significant difference (P < .05) was observed, comparisons between pairs of methods were then made by means of a two-tailed Student t test.
With subjective ratings of image quality, the Kendall coefficient of concordance was used to evaluate the degree of agreement among the three readers, and then the Friedman nonparametric test for related samples (SPSS 6.1; SPSS, Chicago, Ill) was used to assess for statistically significant differences among the three methods. With parameters for which a statistically significant difference in ratings was observed, a comparison between pairs of techniques was made by means of the Wilcoxon rank sum test.
|
RESULTS |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
|
|
|
|
|
|
|
|
Although the difference did not achieve a threshold of a P value less than .05 for statistical significance, a consistent trend in CNR was observed in which the 8-mm reconstruction images had higher liver-to-fat, pancreas-to-fat, and renal cortex–to-fat CNR on both on pre- and postcontrast studies. Aorta-to-fat CNR was significantly higher on the 8-mm reconstruction images compared with images obtained with the other two methods on only postcontrast images.
Qualitative Results
The Kendall coefficient of concordance, used to evaluate the degree of agreement among the three readers, showed that there was no significant difference at a level of significance of a P value less than .05. Average values for subjective measures of image quality are shown in the Table. Overall, the qualitative values support the quantitative results. Average ratings exceeded a value of 4 (on a scale of 1–5 with 5 representing highest quality) for all techniques both before and after contrast material administration. For precontrast images, differences in clarity of the pancreas edge achieved statistical significance (P < .05) with values for the 2-mm fat-saturated volumetric interpolated breath-hold source images, which were significantly lower than those for images obtained with the other techniques (Fig 5). The remaining image quality parameters were not significantly different among the three methods before contrast material administration (Fig 5).
|
|
|
|
|
|
|
DISCUSSION |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Our quantitative and qualitative results show that images obtained with the fat-saturated volumetric interpolated breath-hold sequence are comparable to or better than those obtained with the 2D fat-saturated GRE sequence. The prohibitive reduction in SNR that accompanies use of very thin sections with 2D techniques seems mitigated with fat-saturated volumetric interpolated breath-hold imaging, perhaps owing to the inherent improvement in SNR with 3D techniques and possibly by means of the judicious selection of flip angle (11). In our study, the flip angle was selected to optimize liver-to-spleen CNR. This may account for the slightly greater precontrast pancreas-to-fat CNR observed on 2D fat-saturated images compared with that on the 2-mm fat-saturated volumetric interpolated breath-hold source images or the 8-mm reconstruction images (Table). Given the shorter T1 of the pancreas, it is likely that a larger flip angle could have been selected for the 3D sequence to optimize pancreas-to-background CNR.
We recognize that on a strictly scientific basis, it would have been useful to obtain arterial phase postcontrast studies for both the 2D and 3D approaches and compare these. Such a comparison would have required a separate imaging session to acquire distinct arterial phase images for both approaches, and we believed that the additional time and contrast material exposure was not warranted. The data we acquired for both the precontrast and delayed postcontrast images show the efficacy of the 3D approach. We do not believe that the arterial phase evaluations would have any other differences. Moreover, meaningful multiplanar reformation and MIP images possible with the 3D approach cannot be obtained with the 8-mm-thick sections needed for adequate anatomic coverage with the 2D approach.
Our strategy of using an intermittent fat-saturation pulse for both 2D and 3D techniques enabled fat saturation with minimum added imaging time. Fat saturation improves CNR (12–15) and reduces the potential degradation of image quality resulting from motion-related artifacts. Saturation of fat also improves the depiction of enhanced vascular structures on gadolinium-enhanced MR angiograms. We note that volumetric interpolated breath-hold imaging can also be performed without fat saturation, such as when insufficient magnetic field homogeneity precludes effective frequency-selected fat saturation.
At the time this article was written, there were only a handful of reports about 3D fast GRE acquisitions for abdominal imaging (6,7,16,17). Results in these studies were limited by relatively poor spatial resolution, insufficient anatomic coverage, or both. For example, Soyer et al (6) reported results with a 3D GRE technique that allowed 12 partitions in a 21-second breath hold or 20 partitions in a 30-second breath hold with a 128 x 256 matrix. In that study, partition thickness was 10 mm and therefore the 21-second breath hold could provide only 120 mm of craniocaudal coverage. Such applications of 3D imaging do not provide clear advantages over conventional 2D GRE imaging. Recently, the use of thin-section 3D imaging was described for evaluation of the kidneys in which a limited slab thickness preserved breath-hold capacity (18).
In our study, improved imaging efficiency has been pursued by taking advantage of pulse sequence and reconstruction tools that are unique to 3D sequence structures. By means of zero filling in the section-select direction, we achieved interpolated data that reduced the voxel dimension by a factor of two. By combining these strategies with use of extremely short repetition times (the latter made possible with commercially available high-performance gradients), the fat-saturated volumetric interpolated breath-hold sequence provided volumetric images with 2-mm effective partition thickness within the time frame of a breath hold. With the smaller voxel size, our preliminary observations predict a corresponding improvement in the detection and characterization of small lesions. A scientific study that evaluates the impact of voxel size on lesion detection and characterization is needed.
Results in prior work have shown that zero-filling interpolation (19) and section doubling (20) can reduce partial volume artifacts on 3D MR angiograms. Previously, these techniques have not, to our knowledge, been pursued for body imaging. The precise effects of zero filling on section-select resolution can be complex (21). We expect that the section-select resolution we achieved with interpolation is slightly less than that achieved by doubling the number of partitions without interpolation; however, the latter requires a time penalty and in many cases will eliminate the possibility of breath-hold imaging.
The temporal dynamics of sequential, contrast-enhanced studies bring out important differences between 2D and volumetric interpolated breath-hold imaging techniques. With use of a 2D imaging approach, separate acquisitions are required to obtain alternative imaging planes; thus, image contrast varies with the temporal dynamics of contrast media distribution and therefore with the plane of each separate acquisition. In contradistinction, multiplanar reconstruction images from volumetric interpolated breath-hold data contain image contrast identical to that obtained in the initial acquisition plane. Thus, volumetric interpolated breath-hold examinations can offer images in any obliquity from the data sets obtained during critical phases of enhancement (arterial and portal venous phases of liver enhancement, for example). We predict that the use of multiplanar reconstruction techniques to reformat dynamically acquired volumetric data sets will translate into improved lesion detection and characterization. Verification of this prediction requires further study.
There are additional practical implications for breath-hold volumetric imaging in the body. Multiplanar reconstruction images of volumetric data sets can be used to improve localization of lesions and pretherapeutic planning. For example, for patients who may undergo radio-frequency or cryoablation of liver masses, assessment of proximity of lesions to nearby biliary and vascular structures can be vital. In addition, volumetric imaging can improve definition of segmental anatomy and help the planning of segmental resection of liver masses (22–24). Similarly, the feasibility of partial nephrectomy can be aided by means of precise localization of the masses (25–27).
With use of timing imaging to ensure acquisition of a volumetric data set during peak arterial enhancement, the arterial phase volumetric interpolated breath-hold images that are used to evaluate parenchymal disease can also be manipulated to obtain angiographic images. Postcontrast volumetric interpolated breath-hold images can be reconstructed by using MIP algorithms (Figs 2–4), thereby incorporating the advantages of gadolinium-enhanced 3D MR angiography (10,28–30). The effective demonstration of arterial and venous anatomy is important in the staging of tumors such as renal and pancreatic neoplasms. With volumetric interpolated breath-hold imaging, this can be accomplished by using the same data sets from which cross-sectional axial images are prospectively obtained. Finally, there may be fortuitous angiographic findings demonstrated on angiograms obtained with the volumetric interpolated breath-hold sequence (eg, renal artery stenoses or splenic artery aneurysms) that can lead to early diagnoses and possible elimination of the need for additional tests.
We recognize that the large number of images generated with such a volumetric approach can present challenges for interpretation and also affect data storage considerations. An option for presentation of data is to make hard copies of selected acquisitions as limited reformatted images (eg, 20 images per acquisition consisting of 8-mm reformatted sections). Aside from minimizing film, shipping, and storage costs, images reformatted at larger section thicknesses also provide improved SNR compared with that on the original thin-section images, as we have shown. Similar results have been reported previously (31). While improved SNR could translate into improved image aesthetics, the reformation images reintroduce greater partial volume averaging effects. Therefore, to capture the true benefits of thin-section imaging, a review of source images is important for interpretation. The volumetric approach to imaging may thus lend itself to direct interpretation at the workstation and allow the user to "sweep" through the data sets in standard and oblique planes.
The volumetric interpolated breath-hold imaging sequence was modified from 3D sequences originally structured for gadolinium-enhanced MR angiography. It differs from the MR angiographic sequences by replacing an asymmetric echo in the read direction with an echo that is symmetric; this change improves the in-plane spatial resolution. The sequence also differs by including a decreased flip angle to better address abdominal imaging contrast. It is possible for the reader to attempt 3D body imaging with section-interpolated MR angiographic sequences; however, our initial observations with them prompted the aforementioned changes. The pursuit of alternate zero-fill strategies offers opportunities for further research and potential improvements in volumetric interpolated breath-hold techniques.
In summary, the volumetric interpolated breath-hold examination offers a more comprehensive and efficient approach to body MR imaging. We demonstrated the feasibility of the fat-saturated examination in the clinical setting, showed its comparability in image quality to that with a more conventional fat-saturated 2D GRE imaging strategy, and illustrated examples of its reconstruction and angiographic capabilities. If further studies document that volumetric interpolated breath-hold examination improves the accuracy and utility of body MR imaging, the technique might replace other breath-hold, T1-weighted imaging strategies.
|
Footnotes |
|---|
Author contributions: Guarantors of integrity of entire study, N.M.R., V.S.L.; study concepts, N.M.R., D.T., J.C.W.; study design, N.M.R.; definition of intellectual content, N.M.R., G.L., J.C.W.; literature research, N.M.R., D.T.; clinical studies, N.M.R., V.S.L., G.A.K.; experimental studies, N.M.R., M.A.P., G.L., D.T.; data acquisition, M.A.P.; data analysis, V.S.L., N.M.R.; statistical analysis, V.S.L., M.M.A.; manuscript preparation, N.M.R., V.S.L.; manuscript editing, V.S.L., G.A.K., J.C.W.; manuscript review, V.S.L., G.L., G.A.K., J.C.W.
|
References |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
3.0 cm) renal masses. Magn Reson Imaging Clin N Am 1997; 5:67-81.[Medline]
This article has been cited by other articles:
|
|
 |
|
  S. H. Choi, J. M. Lee, N. C. Yu, K.-S. Suh, J.-J. Jang, S. H. Kim, and B. I. Choi Hepatocellular Carcinoma in Liver Transplantation Candidates: Detection with Gadobenate Dimeglumine-Enhanced MRI Am. J. Roentgenol., August 1, 2008; 191(2): 529 - 536. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 I. Pedrosa, M. R. Sun, M. Spencer, E. M. Genega, A. F. Olumi, W. C. Dewolf, and N. M. Rofsky MR Imaging of Renal Masses: Correlation with Findings at Surgery and Pathologic Analysis RadioGraphics, July 1, 2008; 28(4): 985 - 1003. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. M. Willatt, H. K. Hussain, S. Adusumilli, and J. A. Marrero MR Imaging of Hepatocellular Carcinoma in the Cirrhotic Liver: Challenges and Controversies Radiology, May 1, 2008; 247(2): 311 - 330. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N. Takahashi, A. Kawashima, J. F. Glockner, R. P. Hartman, B. C. Leibovich, A. C. S. Brau, P. J. Beatty, and B. F. King Small (<2-cm) Upper-Tract Urothelial Carcinoma: Evaluation with Gadolinium-enhanced Three-dimensional Spoiled Gradient-Recalled Echo MR Urography Radiology, May 1, 2008; 247(2): 451 - 457. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
F. Y. Donmez, E. Yekeler, V. Saeidi, A. Tunaci, M. Tunaci, and G. Acunas Dynamic Contrast Enhancement Patterns of Solitary Pulmonary Nodules on 3D Gradient-Recalled Echo MRI Am. J. Roentgenol., December 1, 2007; 189(6): 1380 - 1386. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. Inaoka, H. Sugimori, Y. Sasaki, K. Takahashi, K. Sengoku, N. Takada, and T. Aburano VIBE MRI for Evaluating the Normal and Abnormal Gastrointestinal Tract in Fetuses Am. J. Roentgenol., December 1, 2007; 189(6): W303 - W308. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Singh, R. Danrad, P. F. Hahn, M. A. Blake, P. R. Mueller, and R. A. Novelline MR Imaging of the Acute Abdomen and Pelvis: Acute Appendicitis and Beyond RadioGraphics, September 1, 2007; 27(5): 1419 - 1431. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
I. Pedrosa, E. A. Zeikus, D. Levine, and N. M. Rofsky MR Imaging of Acute Right Lower Quadrant Pain in Pregnant and Nonpregnant Patients RadioGraphics, May 1, 2007; 27(3): 721 - 743. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. N. Gandhi, M. A. Brown, J. G. Wong, D. A. Aguirre, and C. B. Sirlin MR Contrast Agents for Liver Imaging: What, When, How RadioGraphics, November 1, 2006; 26(6): 1621 - 1636. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Tozaki, T. Igarashi, and K. Fukuda Breast MRI using the VIBE sequence: clustered ring enhancement in the differential diagnosis of lesions showing non-masslike enhancement. Am. J. Roentgenol., August 1, 2006; 187(2): 313 - 321. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Tozaki and K. Fukuda High-spatial-resolution MRI of non-masslike breast lesions: interpretation model based on BI-RADS MRI descriptors. Am. J. Roentgenol., August 1, 2006; 187(2): 330 - 337. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. M. Hecht, A. E. Holland, G. M. Israel, W. Y. Hahn, D. C. Kim, A. B. West, J. S. Babb, B. Taouli, V. S. Lee, and G. A. Krinsky Hepatocellular Carcinoma in the Cirrhotic Liver: Gadolinium-enhanced 3D T1-weighted MR Imaging as a Stand-alone Sequence for Diagnosis. Radiology, May 1, 2006; 239(2): 438 - 447. [Abstract] [Full Text] [PDF]
|
|
