| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Department of Nuclear Medicine and Diagnostic Imaging, Graduate School of
Medicine, Kyoto University, 54 Shogoinkawaharacho, Sakyo, Kyoto 606-8507,
Japan.
2 Department of Radiology, Kyoto City Hospital, Kyoto, Japan.
3 Department of Radiology, Graduate School of Medicine, Kyoto University, Kyoto,
Japan.
4 Department of Otolaryngology and Head and Neck Surgery, Graduate School of
Medicine, Kyoto University, Kyoto, Japan.
Received December 3, 2003;
accepted after revision May 27, 2004.
Address correspondence to M. Kataoka
(makok{at}kyotou.ac.jp).
Abstract
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
SUBJECTS AND METHODS. The subjects were 33 consecutive patients
referred for MRI for preoperative assessment of head and neck tumors. After
administration of gadodiamide hydrate, images were obtained using postcontrast
fat-saturated VIBE sequence for a 35-sec acquisition time and then a
postcontrast fat-saturated spin-echo T1-weighted sequence for a 269-sec
acquisition time (
4.5 min). Quantitative comparisons of the two methods
were made by calculating signal-to-noise and contrast-to-noise ratios for both
methods, and qualitative comparisons were made on the basis of the scoring of
three independent reviewers concerning image quality and tumor
conspicuity.
RESULTS. No significant difference was detected quantitatively between the two sequences. However, in qualitative assessments, the degree of image degradation by artifacts was significantly smaller for VIBE images than for spin-echo T1-weighted images (p = 0.029).
CONCLUSION. In preoperative evaluations of head and neck tumors, the postcontrast VIBE sequence is capable of decreasing acquisition time without degrading image quality or tumor conspicuity; thus, it is an acceptable alternative to postcontrast spin-echo T1-weighted imaging.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Volumetric interpolated breath-hold examination (VIBE) is a relatively new gradient-echo MR sequence that affords short acquisition times without degrading image quality. This sequence allows the generation of T1-weighted images with fat suppression. Signal characteristics of VIBE sequences may be similar to those of conventional gradient-echo or fast low-angle shot gradient-recalled echo sequences. Some studies already have shown that the VIBE sequence provided images comparable or superior to those obtained using conventional gradient-recalled echo sequences and that therefore the VIBE sequence is useful in abdominal, brain, and chest imaging [4–7]. We hypothesized that this rapid scanning technique could produce images in the head and neck region with acceptable image quality. In this study, we aimed to evaluate the feasibility of using VIBE images as a substitute for conventional postcontrast spin-echo T1-weighted images. Both quantitative and qualitative methods were used to assess image quality.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
|
|
|
|
|
|
|
|
|
|
Referring diagnoses were evaluations for the following: hypopharyngeal carcinoma (n = 18), oropharyngeal carcinoma (n = 5), nasopharyngeal carcinoma (n = 3), lymphoma (n = 3), laryngeal carcinoma (n = 2), tongue carcinoma (n = 1), and mucoepidermoid carcinoma (n = 1). Histopathologic diagnoses were confirmed by surgery or biopsy. Two cases initially suspected of being hypopharyngeal cancers were found to be epidermoid cysts after MR study, but these cases were included.
MRI Protocol
All patients underwent imaging on a 1.5-T MR system (Symphony, Siemens)
with high-performance gradients (25 mT/m), maximum gradient strength, and a
120-µsec rise time using a head and neck coil (combined use of circular
polarized head array and circular polarized neck array coil) as a receiver.
Section thickness was 4 mm with a 1-mm intersection gap for all sequences
including contrast-enhanced VIBE and spin-echo T1-weighted sequences. The
field of view was 18 x 18 cm or 20 x 20 cm in all of the
sequences. Before an MRI examination began, a 22-gauge IV catheter was placed
in an anterior cubital vein of the patient.
All patients underwent conventional spin-echo T1-weighted and fast spin-echo T2-weighted imaging according to our routine protocols in the head and neck area with a TR/TE of 570/9.8 and 5,140/114, respectively. Gadodiamide hydrate (Omniscan, Daiichi Pharmaceutical) then was administered at a dosage of 20 mL at a rate of 2 mL/sec. The delay time between contrast material administration and the start of contrast-enhanced fat-saturated VIBE imaging was 50 sec. Eighty-five seconds after contrast material administration, fat-saturated spin-echo T1-weighted images were obtained. VIBE images always were obtained first, followed by spin-echo T1-weighed images; this order was not randomized.
Section thickness, slice gap, and field of view were the same as those used
for the unenhanced images. Thirty slices were obtained. The fat-saturated VIBE
images were obtained with a TR/TE of 4.3/2.0, flip angle of 15°, bandwidth
of 490 Hz/pixel, and 2 signal averages. The fat-saturated spinecho T1-weighted
images were obtained with a TR range/TE of 497
502/17.0, bandwidth of 130
Hz/pixel, 256 x 256 matrix, and 1 signal average. Flow compensation was
applied to reduce flow-related artifacts. The acquisition time for a
fat-saturated VIBE image was 35 sec, whereas that for a fat-saturated
spin-echo T1-weighted image was 269 sec. In contrast to the protocol used for
imaging the abdomen [4,
5], patients were not required
to hold their breath during scanning, but they were required to avoid
swallowing and coughing for this protocol.
The VIBE sequence is a spoiled gradient-echo acquisition. Asymmetric echo was used in the read direction with a bandwidth of 490 Hz/pixel. The parameters can be summarized as follows: phase resolution of 80% and slice resolution of 63% with partial Fourier sampling of 75%. An initial matrix of 205 x 256 was interpolated by zero filling to a 410 x 512 matrix.
To determine the optimal scanning protocol for this study, we conducted preliminary trials to explore the number of signal averages for the VIBE sequence with five subjects with head and neck cancer, none of whom was included in our study population. Theoretically, image quality improves with an increasing number of signal averages. The drawbacks of increasing the number of signal averages in head and neck imaging are the increase in scanning time and possible image degradation by motion artifacts. Thus, we aimed at an image quality for VIBE imaging that was at least comparable to spin-echo T1-weighted imaging with a minimum scanning time and a minimum number of averages. We started with 1 signal average, a default for abdominal imaging, which produced images of somewhat lower quality than a conventional spin-echo T1-weighted image. The number of signal averages was increased to obtain an image quality comparable to that of spin-echo T1-weighted images. We then set up the MRI protocol for VIBE imaging with 2 signal averages to upgrade image quality and to maintain the advantages of rapid scanning.
Quantitative Assessment of Images
Region-of-interest (ROI) analyses were performed by one investigator on a
satellite console of the MR unit. Means and SDs of signal intensities (SIs) of
the primary tumor, muscle, and air were recorded for postcontrast
fat-saturated VIBE images and spin-echo T1-weighted images. Only primary
tumors were evaluated, and metastatic lymph nodes were not included in the
evaluation. The ROIs of tumor, muscle, and air were selected within the same
slice and were identical on both VIBE and spin-echo T1-weighted images to
eliminate differences among slices. The ROI of the tumor was selected
carefully so that the area was more than 0.5 cm2. In case the tumor
was associated with a large area of central necrosis, the ROI of the tumor was
placed in a nonnecrotic area because it is the contrast, or difference in
intensity between the peripheral part of the tumor and the surrounding
structures, that contributes to delineate tumor. Including a large area of
necrosis might decrease the signal intensity of the tumor, losing the
information about the peripheral area of the tumor.
On MRI, a necrosis was defined as an area of low signal on T1-weighted
images, high signal intensity on T2-weighted images, and lack of enhancement
on contrast-enhanced images
[6]. The regions for primary
tumor and muscle were selected away from intratumoral or intraparenchymal
blood vessels. According to the location of the primary tumor, ipsilateral
sternocleidomastoid, strap (sternothyroid and sternohyoid), or lateral
pterygoid muscles were selected for the ROI analysis of muscle. The ROI of air
was selected away from the patient's body to avoid the effects of artifacts.
Based on the measured SI, the signal-to-noise ratio (SNR) of the tumor, the
SNR of the muscle, and the contrast-to-noise ratio (CNR) of the tumor using
muscle as a reference were calculated on both postcontrast fat-saturated VIBE
and spin-echo T1-weighted images according to the following equations:
 |
 |
 |
Qualitative Assessment of Images
Three independent reviewers, all of whom were experienced in interpreting
head and neck imaging, assessed the two sequences; all were blinded to the
location, diagnosis, and choice of sequence but were told that the images were
postcontrast T1-weighted images. First, each reviewer had a preliminary review
session with 12 MR image series (six spin-echo T1-weighted and six VIBE image
series) to understand the criteria for evaluation. These 12 MR images in the
series used for the preliminary review were selected from the images in
preliminary trials to decide the optimal scanning protocol and were not
included in the analysis of this study. This was followed by a review session
consisting of 66 MR image series (33 spin-echo T1-weighted and 33 VIBE image
series) presented in a random fashion. Each reviewer independently evaluated
66 image series with respect to overall image quality, artifact, and tumor
conspicuity.
Image quality.—Thirty-three cases were included in the evaluations of image quality. Image quality was further classified into two major categories: overall image quality and artifacts, which are a major cause of decreased image quality. Overall image quality was defined as that assessed after a comprehensive evaluation based on the overall clarity of anatomic structures and landmarks in the images. Overall image quality was assessed by three reviewers scoring the parameters on the following scale: 1, unacceptable; 2, poor; 3, fair; 4, good; and 5, excellent.
Artifacts.—The degree of artifacts was assessed by assigning a score on a scale of 1–5. In contrast to the evaluation of overall image quality, the allocated score was related inversely to the degree of artifacts so that a higher score indicated images with fewer artifacts and thus a better score in terms of quality: 1, unacceptable; 2, severe; 3, moderate; 4, mild; and 5, absent. A score of 5 indicated that the image was free of artifacts. In this evaluation, we focused on decreases in image quality caused by artifacts; hence, the effects of the severity of motion-related artifacts, flow-related artifacts, or other kinds of artifacts were evaluated together.
Tumor conspicuity.—For the evaluation of tumor conspicuity, cases in which the tumor was detectable on a contrast-enhanced spin-echo T1-weighted image by all three reviewers were included. If a tumor was too difficult to identify on MRI and was missed by one of the reviewers, it was excluded. Tumor conspicuity was evaluated according to the following scoring system: 1, unacceptable; 2, poor; 3, fair; 4, good; and 5, excellent. A score of 5 corresponds to an image that shows a clear contrast between the tumor and the surrounding tissue and a delineated margin, whereas a score of 1 indicates an image in which a tumor is difficult to identify.
Evaluation of Lymph Nodes
Because metastatic lymph nodes were suspected in only 10 of the 33 cases,
comparisons of VIBE and spin-echo T1-weighted images were performed for these
cases only. Primary tumors were hypopharyngeal cancer (n = 7),
nasopharyngeal cancer (n = 1), and lymphoma (n = 2). We used
the most prominent lymph nodes for comparison because pathologic confirmation
often was obtained only by biopsy. The conspicuity of the suspected metastatic
lymph nodes was scored using the same criteria as for tumor conspicuity: 1,
unacceptable; 2, poor; 3, fair; 4, good; and 5, excellent.
Statistical Analysis
Quantitative assessment.—Cases of cystic lesions or patients
with an undetectable tumor were excluded from quantitative assessment because
it was impossible to select an ROI of the tumor. The paired Student's
t test was used to compare image parameters (SNR and CNR) measured
for both post-contrast fat-saturated VIBE and spin-echo T1-weighted
images.
Qualitative assessment.—The kappa value was calculated to evaluate the degree of agreement among the three reviewers. When the three reviewers disagreed on a score, the score that two of the reviewers agreed on was used for the final evaluation. Comparisons of scores between the two sequences were made using the Wilcoxon's rank sum test. Because of the small number of cases, statistical testing was not performed for evaluations of lymph nodes.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The SNR and CNR of the tumor and muscle in each of the sequences are shown in Table 1. Values are displayed as means ± SD. The mean values of SNR of tumor, SNR of muscle, and CNR of tumor were 41.4, 18.8, and 22.7 on contrast-enhanced VIBE images and 40.5, 18.8, and 21.6 on contrast-enhanced spin-echo T1-weighted images, respectively. As for SNR and CNR, no significant difference was detected between the two sequences (at a level of significance of p < 0.05).
|
Qualitative Assessment
The calculated kappa values ranged from 0.625 to 0.685 (average, 0.655) for
overall image quality, from 0.725 to 0.807 (0.779) for artifacts, and from
0.750 to 0.883 (0.789) for tumor conspicuity, showing relatively good
agreement of scores among the three reviewers. The reviewers differed by no
more than one score on this scaling system; thus, final evaluations were based
on the score agreed to by at least two of the reviewers. Scores of subjective
image parameters for each of the sequences are shown in
Table 2. The values displayed
are means ± SD. Average scores exceeded a value of 4 (on a scale of
1–5 where 5 is the best score) for both techniques and all parameters,
suggesting excellence in image quality and tumor detectability.
|
Image quality.—The mean value of overall image quality was 4.55 on contrast-enhanced VIBE images and 4.36 on contrast-enhanced spin-echo T1-weighted images. No significant difference was detected between the two sequences (p = 0.096), although average scores tended to be higher for VIBE images than for spin-echo T1-weighted images. The mean value of artifacts was 4.39 on contrast-enhanced VIBE images and 4.12 on contrast-enhanced spin-echo T1-weighted images. For artifacts, the scores for VIBE images were significantly higher (corresponding to fewer artifacts and better quality) than those for spin-echo T1-weighted images (p = 0.029).
Tumor conspicuity.—The tumor was too small or too difficult to identify on spin-echo T1-weighted images in three cases. These three tumors also were missed on VIBE images by at least one of the reviewers; thus, 30 cases were included in the evaluation of tumor conspicuity. The mean value of tumor conspicuity was 4.37 on contrast-enhanced VIBE images and 4.20 on contrast-enhanced spin-echo T1-weighted images. No significant difference was detected between the two sequences (p = 0.109).
Evaluation of Lymph Nodes
The results of imaging, including scoring conspicuity of metastatic lymph
nodes (the most prominent lymph nodes) on spin-echo T1-weighted images, are
summarized in Table 3. The
conspicuity of suspected lymph node metastasis was scored equal in six cases,
higher on the spin-echo T1-weighted image in one, and higher on VIBE images in
three. Scores tended to be higher in cases with lymph nodes of more than 10 mm
in maximum diameter, including two cases of lymphoma. Overall, conspicuity of
these lymph nodes on VIBE images was equivalent to or possibly higher than on
spin-echo T1-weighted images.
|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
For preoperative evaluation and staging of head and neck tumors, MRI with the use of contrast enhancement, with its ability to delineate various soft tissues, is the technique of choice to evaluate deep tumoral extensions and metastases to regional lymph nodes. Conventional enhanced spin-echo T1-weighted images with fat saturation, however, take as long as 4–5 min. As a result, image quality can be degraded because of patient movements such as breathing, coughing, snoring, swallowing, and other involuntary and voluntary body movements. In fact, approximately 10–15% of all MRI studies on laryngopharyngeal cancers are nondiagnostic mainly because of motion artifacts [1, 2].
The VIBE sequence is an interpolated T1-weighted gradient-echo MR sequence first described for abdominal imaging by Rofsky et al. in 1999 [4]. They showed that the VIBE sequence affords short acquisition times and image quality comparable or superior to conventional sequences. Thereafter, the feasibility of this technique for imaging the abdominal organs, such as liver, and imaging the whole body, chest, central nervous system, and intracranial venous system has been reported [4–9]. The strength of this spoiled gradient-echo sequence lies in its superior resolution irrespective of the considerable decrease in acquisition time. With the introduction of this sequence, it is possible to shorten the acquisition time to approximately one fifth to one tenth that of conventional spin-echo T1-weighted imaging.
Based on our results, that there are fewer artifacts in the head and neck area might be one of the advantages for using VIBE imaging instead of spin-echo T1-weighted imaging. Comparisons of the severity of artifacts between the VIBE sequence and other sequences have been reported. For example, Wetzel et al. [6] compared VIBE images and spin-echo T1-weighted images and found that VIBE images showed fewer flow-related artifacts but more truncation artifacts than spinecho T1-weighted images [6]. Rofsky et al. [4] compared fat-saturated VIBE and 2D gradient-echo sequences in postcontrast images of the abdomen and found that VIBE images had less degradation by arterial ghosting. Karabulut et al. [7] compared fat-saturated VIBE and 2D gradient-echo sequences on postcontrast images of the chest and found no significant difference in motion artifacts, but they did find a decrease of pulsation artifacts on the VIBE images. These studies differed from ours in imaging parameters, the targeted area of the body, the sequence used for comparison, and the use of contrast medium or fat suppression. Nevertheless, the VIBE sequence seems to be less prone to artifacts.
In this study, we planned initially for a simple analysis: evaluating the severity of several kinds of artifacts together in terms of the degree of image degradation. Thus, detailed analysis and comparison with previous studies could not be made with regard to individual artifacts such as motion-related and flow-related artifacts and with arterial ghosting. Our observations were that artifacts, if seen, were either motion- or flow-related. Because the decrease in the number of artifacts appears to be one of the advantages of VIBE images, further analysis of each kind of artifact might be informative to improve image quality.
High-speed scanning has several advantages; for example, most patients with head and neck cancer are of advanced age and sometimes have breathing difficulties (e.g., due to airway narrowing caused by the tumor itself or tobacco-associated lung disease) [2]. Therefore, such patients, even if instructed not to move or swallow, have difficulty in enduring half an hour without moving. Indeed, rescanning is sometimes required, or the patients may give up on undergoing MR examinations. VIBE images, which can be obtained with decreased acquisition time, might be applicable for examination of such patients. In patients who can endure relatively longer scanning, adding coronal and sagittal images might be helpful for surgeons in understanding the spatial extension of the tumor. In such cases, the effect of saving scanning time is remarkable. Another advantage of rapid MR scanning includes the possibility of introducing contrast-enhanced dynamic study, which is thought to be useful in evaluating tumor vascularity, the vascular anatomy of the region, or both [5, 9, 10].
VIBE images also can be used as 3D MR images. The effectiveness of VIBE imaging already has been reported in studies of the liver and brain [4–6, 9]; in those studies, authors supported the usefulness of 3D reconstructed images for preoperative evaluation of vessels or identification of tumors. Advantages of isotropic 3D imaging include avoidance of partial volume artifacts associated with 2D imaging and the potential to reformat images in any plane without loss of image resolution. In our study, we used a 4-mm scan, the same parameter as for conventional spinecho T1-weighted imaging, for comparison because our primary purpose was to evaluate the feasibility of using VIBE images as a substitute for conventional postcontrast spinecho T1-weighted images. Although we did not examine the usefulness of 3D images in this study, adding 3D images with isovoxel data acquisition for a focused area such as a vocal cord or hypopharynx would be an effective way of providing detailed anatomic and pathologic information in selected cases.
A possible limitation of our study is the effect of scanning timing between the spin-echo T1-weighted images and VIBE images in our protocol. The 35-sec delay in obtaining spinecho T1-weighted images after the administration of gadodiamide hydrate compared with VIBE images, which were obtained soon after the administration of the agent, might have affected the enhancement pattern, image quality, or both. However, the total acquisition time of a fat-saturated spin-echo T1-weighted image was as long as 4.5 min; thus, a delay of only 35 sec should not have affected image quality or tumor conspicuity significantly on a spin-echo T1-weighted image. VIBE imaging was set to start 50 sec after the bolus injection of contrast medium to allow time for the enhancement of tumors. However, VIBE images may have been acquired too early, leading to a decrease in the CNR of the tumor and tumor conspicuity.
The methodology of analysis in respect to the successful blinding of the sequence may have some difficulty; an experienced reviewer may be able to distinguish these two sequences even if blinding is attempted. For reviewers to make unbiased judgments, it is important that evaluations are done in a manner that is as blind as possible. We instituted the blinding here as best we could, although it might not have been completely successful. As for tumor conspicuity, the exclusion of three cases because the tumor was missed by at least one reviewer on a spin-echo T1-weighted image theoretically creates some bias that could lead to an underestimate of the evaluations of VIBE images. However, because we used a spin-echo T1-weighted image as the gold standard (not a pathologic sample or surgical specimen) and looked at the feasibility that VIBE images can be a substitute for spin-echo T1-weighted images, we believe that missing a lesion detectable on spin-echo T1-weighted images has more effect than detecting a lesion missed on spin-echo T1-weighted images. Indeed, the same lesions were missed on both spin-echo T1-weighted images and VIBE images; thus, we think that no bias was introduced into the study.
The lack of detailed pathologic correlation such as the presence of peritumoral extension and perineural invasion between MRI findings and operative findings might be another possible limitation. Although such comparisons would be informative, frequent use of preoperative chemotherapy and radiation can modify tumoral extension, making accurate comparisons difficult within this study group. Also, evaluations with regard to metastatic lymph nodes were limited because of the small number of cases with metastatic lymph nodes, although the results suggested equivocal conspicuity in metastatic lymph nodes. The role of MRI in the evaluation of these lesions is important, so we are investigating the role of VIBE imaging in evaluating these lesions as a next stage in our studies.
In conclusion, for the preoperative evaluations of head and neck tumors, postcontrast VIBE imaging is capable of decreasing acquisition time considerably—less than one fifth of the acquisition time needed for routine spinecho T1-weighted imaging—without degrading image quality and tumor conspicuity. Therefore, the VIBE sequence is an effective alternative approach and can be substituted for postcontrast spin-echo T1-weighted images in the evaluation of head and neck tumors.
Acknowledgments
We are grateful to Akira Hiraga and all MRI technologists for their help
with the examinations and for their valuable suggestions.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
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]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
