| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1
Department of Radiology, Stanford University School of Medicine, 300 Pasteur
Dr., Rm. H1307, Stanford, CA 94305
2
Present address: Department of Radiology, St. Alphonsus Regional Medical
Center, 1055 N. Curtis Rd., Boise, ID 837060.
3
Present address: Diagnostic Radiology, National Institutes of Health, Bldg.
10, Room 1C660, 10 Center Dr., MSC 1182, Bethesda, MD 20892-1182.
Received May 23, 2001;
accepted after revision August 27, 2001.
Supported in part by grant T32-CA90695 from the National Institutes of
Health.
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. A retrospective review of abdominal MR imaging examinations in 61 patients was performed. All MR examinations included unenhanced spin-echo T1-weighted, unenhanced fat-suppressed fast spin-echo T2-weighted, and multiphasic gadolinium-enhanced 3D fast spoiled gradient-recalled echo sequences obtained during successive breath-holds. The liver was evaluated for focal lesions first with the 3D spoiled gradient-recalled echo sequences and then, during a separate sitting, with the T1- and T2-weighted sequences. The usefulness of each sequence in the detection and characterization of lesions was recorded. The gold standard for lesion detection and characterization was all three imaging sequences reviewed together.
RESULTS. A total of 114 focal liver lesions were identified, 54 of which were simple cysts. The 3D spoiled gradient-recalled echo sequence alone detected 92 (81%) of the 114 lesions, and the T1- and T2-weighted sequences detected 95 (83%) of the 114 lesions. Of the 60 lesions that were not simple cysts, the 3D spoiled gradient-recalled echo sequence alone detected 58 (97%), and T1- and T2-weighted sequences detected 51 (85%). In 24% of the patients with lesions, the T1- and T2-weighted sequences were found to be helpful for the characterization of lesions.
CONCLUSION. A multiphasic contrast-enhanced 3D fast spoiled gradient-recalled echo sequence alone detects most of the clinically relevant focal liver lesions. Additional liver examination using both unenhanced T1- and T2-weighted sequences is helpful for lesion characterization but increases the detection rate only minimally.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The radiologists reviewed the studies together, and consensus agreement was achieved in findings for all patients. Both radiologists were unaware of whether findings for the patients had been normal or abnormal. The 3D sequences were reviewed first, independently of the unenhanced T1- and T2-weighted sequences. During a separate sitting, the T1- and T2-weighted images were analyzed without the 3D sequences. Finally, all three sequences were reviewed simultaneously. The gold standard for detection and characterization was consensus of two radiologists trained in abdominal imaging reviewing all three sequences simultaneously (3D, T1-weighted, and T2-weighted).
All patients were imaged on one of two 1.5-T scanners (Signa EchoSpeed;
General Electric Medical Systems). The body coil was used for radiofrequency
transmission, and either the body coil or 4-element torso phased array surface
coil was used for signal reception. Axial T1- and T2-weighted images had been
acquired before the administration of IV contrast material. Axial T1-weighted
images were obtained as a standard spin-echo sequence using respiratory
compensation, superior and inferior spatial saturation bands, no phase wrap
option, and the following parameters: TR range/TE, 300-400/minimum full; flip
angle, 90°; receiver bandwidth, ±16 kHz; slice thickness, 7 mm;
skip, 3 mm; field of view, 32-48 cm; frequency encodes, 512; phase encodes,
192; and signal averages, 2. Axial T2-weighted images were obtained as a fast
spin-echo sequence using a water-selective spectral-spatial excitation pulse
(to eliminate a lipid signal)
[4], superior and inferior
spatial saturation bands, no phase wrap option, and the following parameters:
TR/TE, 
4,000/102; flip angle, 90°; echo train length, 8-12; receiver
bandwidth, ±32 kHz; slice thickness, 9 mm; skip, 1 mm; field of view,
32-48 cm; frequency encodes, 512; phase encodes, 192; and signal averages, 4.
Imaging times were 115-154 sec for the T1-weighted sequence and approximately
300 sec for the T2-weighted sequence.
After the T1-weighted and T2-weighted sequences were performed, the patients were scanned with a 3D pulse sequence developed at our institution that uses faster slew rates than standard manufacturer product sequences [5]. With high-speed imaging gradients (maximum gradient strength, 22 mT/m; maximum gradient slew rate, 120 mT/m per millisecond), this pulse sequence allows high-resolution (512 x 192 x 64) volume acquisitions during sequential 30-sec breath-holds. It is possible to perform pure arterial phase imaging while maintaining full scan coverage and thin (< 6 mm) section thickness.
Our liver imaging protocol calls for 3D acquisitions to be obtained before the contrast material is injected and then also during multiple phases of gadopentetate dimeglumine enhancement. The unenhanced 3D images provide T1-weighted information. A power injector and a 20-gauge IV catheter inserted into a superficial vein of an arm are used to administer 0.1-0.2 mmol/kg of gadopentetate dimeglumine (concentration, 0.5 mmol/mL) at a rate of 2 mL/sec. The arterial phase acquisition begins 15-20 sec after the start of the injection of contrast material. With sequential k-space filling, the low frequency phase encoding is centered approximately 30-35 sec after the start of contrast material injection. After the first 30-sec acquisition, the patient is instructed to breathe for 10 sec before suspending respiration again for the portal venous phase acquisition. This pattern is repeated for the early delayed phase. Several minutes later, a late delayed phase is acquired, which completes the multiphasic 3D sequences. The parameters of the 3D sequence are: TR/TE, 4.6/1; flip angle, 20°; receiver bandwidth, ±64-125 kHz; fractional echo; average number of signals, 1; inplane field of view, 32-48 cm; inplane matrix, 512 x 192; z-phase encodes, as many as 32; section thickness, 3-6 mm, with no order filling in z, resulting in section spacing of 1.5-3.0 mm; and scan time, 30 sec. The scan volume, prescribed graphically, covers the entire liver.
The patients in our study were scanned in either the coronal or axial plane for the 3D sequence. Most were scanned in the coronal plane because the liver can be imaged with a thinner section thickness during the 30-sec breath-hold than is possible in the axial plane. The late delayed 3D fast spoiled gradient-recalled echo sequence was typically performed using partial fat-saturation [6,7,8]. In the partial fat-saturation procedure, a spectrally selective lipid saturation pulse is applied at every eighth repetition. Of the eight phase encodes that follow each saturation pulse, the low-frequency phase encodes are obtained first so that the center of k-space is devoid of fat signal [7,9].
Focal liver lesions revealed by each sequence type were counted. Patients
whose liver contained innumerable (
15) lesions were excluded from
statistical analysis to prevent the results from being heavily influenced by
only a few individuals. The detection rate was determined for all lesions, as
well as for the subset of lesions that were not simple cysts. The sensitivity
and specificity for lesion detection rate were determined for 3D alone and for
all sequences combined. The radiologists interpreting the images were asked to
record whether and how the T1-weighted and T2-weighted sequences helped with
lesion characterization.
The overall image quality of the 3D sequence was rated by consensus using a 5-point scale, 1 representing nondiagnostic and 5 representing excellent image quality. Factors that degraded image quality were noted. The interpreters recorded their opinion on whether the late delayed 3D sequence was beneficial. The quality of the arterial phase 3D sequence was rated in regard to the timing of the gadolinium bolus. An arterial phase considered perfect was one that revealed intense enhancement of the hepatic artery and no more than a trace of portal venous enhancement. The hepatic arterial phase was considered mildly late if the portal vein signal intensity was more than a trace but less than that of the hepatic artery, moderately late if the portal vein signal intensity was greater than that of the hepatic artery, and severely late if any hepatic venous enhancement was observed. The first phase of imaging was deemed acceptable for an arterial phase if the timing was either optimal or mildly late. Moderately or severely late arterial phases were deemed unacceptable for true arterial phase imaging.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
|
Because most of the 114 focal liver lesions were simple cysts and had no clinical relevance, the subgroup of noncystic lesions was further evaluated. Twenty-four patients had livers with lesions that were not simple cysts, including three patients with livers that contained too many solid lesions to count. In the livers of the remaining patients, 60 lesions that were not simple cysts were identified (range, 1-11 per liver; mean, 2.7; median, 1). Of the 60 noncystic lesions, 35 (58%) were suspected to be malignant, 17 (28%) were indeterminate for malignancy, six (10%) were typical hemangiomas, and two (3%) were focal nodular hyperplasia. Of the 17 indeterminate lesions, eight were hyperenhancing lesions in the liver of a 12-month-old female infant with congenital abnormalities (Fig. 3A,3B,3C,3D); four were in an adult patient's liver and were thought most likely to be hemangiomas on the basis of the findings of MR imaging, but the imaging features were not compelling enough to be conclusive; one lesion contained fluid, with a differential diagnosis of abscess or cystic neoplasm; and four lesions were nonspecific solitary liver lesions measuring 1 cm or less in diameter. Of 60 noncystic lesions, 58 (97%) were revealed by the 3D sequence as were 50 (96%) of 52 suspicious or indeterminate lesions. In comparison, T1- and T2-weighted sequences revealed 51 (85%) of 60 noncystic lesions and 44 (83%) of 53 suspicious or indeterminate lesions (Fig. 1).
|
|
|
|
Of the two false-negatives in the 3D sequence, one in retrospect could be seen as a subtly enhancing lesion and was considered suspicious for malignancy. The other was small and could not be clearly discerned even in retrospect; it was considered indeterminate for malignancy. These two lesions were missed because of lack of sufficient signal contrast between lesion and liver parenchyma. The reasons that nine lesions were missed on the T1- and T2-weighted sequences were lack of sufficient lesion contrast (Figs. 4A,4B,4C and 5A) and blurring and phase ghosting artifacts stemming from respiratory motion (Fig. 5A,5B,5C,5D).
|
|
|
|
|
|
|
Of the 38 patients with liver lesions, the T1- and T2-weighted sequences helped with characterization of focal liver lesions in nine (24%). The T1- and T2-weighted sequences were found to be helpful in characterizing small cysts, hemangiomas, and focal nodular hyperplasia. In the remaining 29 patients (76%), the T1- and T2- weighted sequences provided information about lesion characterization that would have neither changed the diagnosis nor improved confidence in the interpretation of the 3D sequence alone. In two patients, the T1- and T2-weighted sequences revealed noncystic lesions that had not been detected prospectively using the 3D sequence. Thus, T1- and T2-weighted sequences either helped with characterization or detection of clinically relevant lesions in 11 (29%) of 38 patients.
Image quality averaged 4.7 on a 5-point scale. In those images judged to be suboptimal, the quality was degraded by respiratory motion, phase artifacts (e.g., cardiac phase), and low signal-to-noise ratio. Using a scan delay of 15-20 sec for the arterial phase acquisition was deemed acceptable in 53 (87%) of 61 examinations. The timing of the arterial phase was optimal in 37 (61%) of 61 and mildly late in 16 (26%) of 61. Of the eight examinations that did not have acceptable arterial phase images, four were obtained before the arrival of the contrast material bolus to the hepatic artery, and four were obtained too late, resulting in moderately late arterial phase imaging (signal intensity in portal veins greater than in hepatic arteries, no hepatic vein enhancement) in three patients and severely late arterial phase imaging (hepatic vein enhancement observed) in one patient.
Overall, the late delayed 3D sequence was believed to be helpful in 25 (41%) of 61 studies. In only one (4%) of 23 studies with no focal liver lesion was the late delayed phase judged to be helpful, and in this one case, the sequence helped to disprove the presence of a lesion suspected because of the findings of another imaging sequence. Of the 38 livers containing at least one focal lesion, the late delayed phase was thought to be helpful in 24 (63%) of 38. In 22 (92%) of these 24 cases, the late delayed phase helped in lesion characterization; in one case (4%), it helped to confirm the presence of a subtle lesion, and in one other case (4%), the late delayed phase helped to better delineate the extent of a central liver mass. Of the sub-group of 24 patients with liver lesions that were not simple cysts, the late delayed phase was deemed helpful in 17 (71%) of 24-15 (88%) of 17 because of helping with lesion characterization, one (6%) of 17 for helping with confirmation of the presence of a lesion, and one of (6%) 17 for helping with delineation of the extent of a lesion.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
|
|
|
The false-negative rate for detection of solid lesions with contrast-enhanced 3D alone was two of (3%) 60. Of these two lesions, one was suspected to be malignant, and another was considered to be indeterminate for malignancy. Both images were rated either good or excellent in quality, and both had arterial phases that were optimal. In retrospect, only one of the two lesions could be clearly discerned, and neither lesion was identified prospectively because of low lesion-to-liver contrast.
Of the 38 patients with liver lesions, the interpreting radiologists were able to confidently characterize lesions using contrast-enhanced 3D sequences alone in 29 (76%). The T1- and T2-weighted sequences did help in lesion characterization in nine (24%) of 38 patients, and the lesions in these patients were usually cysts, hemangiomas, or focal nodular hyperplasias. Thus, a multiphasic 3D sequence alone proved to be quite sensitive in revealing clinically relevant lesions, but additional imaging with T1- and T2-weighted sequences added confidence for lesion characterization in 24% of patients with lesions.
On the basis of this evidence, one could construct a limited protocol for focal liver disease: Begin with a multiphasic contrast-enhanced 3D sequence, and then review the images at the MR imaging console while the patient is still in the scanner. If no abnormalities are identified and the images are of high quality, the examination can be terminated with only a small loss of sensitivity. Additional imaging with a minimum of a T2-weighted sequence is indicated if lesions are detected but cannot be confidently characterized, if the images are of suboptimal quality, or if delineation of lesions is vague or the presence of a lesion is in question. In the last scenario, additional imaging would improve specificity.
In patients for whom additional T2-weighted imaging is desired, the imaging can be performed immediately after the contrast-enhanced sequences. One concern in obtaining T2-weighted images after administration of contrast material is the shortening effect of gadolinium in enhanced lesions; however, it is extremely unlikely that the concentration of gadolinium accumulating in liver lesions, even cavernous hemangiomas, would be sufficient to become problematic. Reviewing images on the MR imaging console may not be practical in many radiology practices. In these cases, patients could be recalled for additional unenhanced imaging if necessary. Even if the patient had a liver lesion, such a recall would only be necessary approximately 25% of the time.
We believe that unenhanced imaging of the liver with a fast two-dimensional gradient-recalled echo sequence for T1-weighting and single-shot fast spin-echo sequence for T2-weighting [11, 12] could each be performed as breath-held sequences and could potentially provide all information needed for lesion characterization without substantially adding to imaging time. Our opinion on this issue is purely speculative; we did not investigate the matter in our study.
The indication for the MR imaging will influence the imaging protocol. In many cases, a focal liver lesion has been detected on another imaging modality such as CT or sonography, and MR imaging is requested for further characterization. In cases such as this, it is probably wise to image the liver with a complete protocol that includes high-quality T1- and T2-weighted imaging in addition to imaging with exogenous contrast material. Potential candidates for an abbreviated liver protocol may include patients without known liver disease, patients who are foregoing CT imaging because of renal insufficiency or allergy to iodine contrast material, and patients with cirrhosis undergoing routine screening for hepatoma.
The image quality of the 3D sequences was rated very good overall (4.7 on a 5-point scale) by consensus of two interpreters. This result is in keeping with the recent investigation by Lee et al. [13]. The reasons for image degradation were breathing motion artifacts, cardiac motion artifacts (phase ghosting from the heart), and poor signal-to-noise ratio (Fig. 6A,6B,6C,6D).
Several options are available if the patient is not able to suspend respiration for 30 sec. The scan time can be reduced by increasing slice thickness or decreasing the y matrix to reduce the number of phase encodes. In certain cases, increasing the receiver bandwidth may also be an option to reduce scan time. One must be aware that these techniques will reduce the signal-to-noise ratio, potentially compromising image quality. When imaging patients who are unable to suspend respiration at all, increasing the imaging time by reducing the receiver band-width or increasing the number of signal averages will likely improve image quality by averaging through more respiratory cycles and improving the signal-to-noise ratio at the expense of a compromised arterial phase scan length. For maximal signal-to-noise ratio, a multielement torso phased array surface coil should be used for breath-held 3D imaging of the liver. Because older scanner platforms were used, many of the 3D sequences in our study were performed with the body coil.
A substantial improvement in image quality was noted in those acquisitions using the torso phased array surface coil. A common artifact in coronal images was a phase ghosting artifact in the liver dome associated with cardiac motion. Coronal imaging allows thinner slice thickness in the same imaging time as axial imaging because the anterior-to-posterior dimension of the liver is usually less than the superior-to-inferior dimension. With a slight loss in spatial resolution but improvement in signal-to-noise ratio, axial imaging could be performed to avoid the artifacts in the dome of the right lobe. Increased cardiac motion artifacts in the left lobe can be expected with axial 3D imaging, however. Another solution would be to perform an additional delayed breath-held 3D sequence in the imaging plane orthogonal to the initial phases.
Using an empirical delay of 15-20 sec (depending on the patient's age and overall state of health), we considered the first 3D phase truly arterial in 53 (87%) of 61 patients and either truly arterial or late arterial in 57 (93%) of 61. These results may be acceptable in most situations. The use of automated bolus tracking would likely result in even greater success in achieving optimal arterial phase scanning; we do not use automated bolus tracking because our 3D sequence is not compatible with such an option at this time. A drawback to automated bolus tracking lies in the lack of the leading edge of the gadolinium bolus during the scan acquisition. Use of a timing bolus with 2-3 mL of gadopentetate dimeglumine would likely improve the rate of optimal arterial phase scans, but would add additional time—and possibly confusion—to the examination. In patients for whom an optimal arterial phase scan is critical or patients with known cardiac disease or central venous obstruction, the use of a timing bolus or automated bolus tracking is suggested [13].
The late delayed 3D sequence, obtained 3-5 min after the administration of gadopentetate dimeglumine, was helpful in 25 (41%) of 61 patients. The main benefit was in lesion characterization, and in one of the 61 cases, the late delayed scan helped in the detection of a liver lesion. The value of delayed venous phase has been previously reported by Hawighorst et al. [10], and our study confirms their findings. Given that the additional acquisition takes relatively little time, we believe in routinely obtaining 3-5 min delayed scans after administering contrast material.
A limitation to our study is the lack of histologic proof of the liver lesions. We have emphasized the accuracy of the 3D sequence in detecting lesions that are not simple cysts; the gold standard is having all imaging sequences interpreted together. However, the point of our study was not to determine the accuracy of MR imaging per se but rather to determine the accuracy of a multiphasic 3D sequence alone compared with our routine liver MR imaging protocol. We do not use liver-specific contrast agents, such as ferroxides or mangafodipir, that may provide greater sensitivity for the detection and characterization of focal liver lesions [14,15,16,17,18].
Other limitations were also present. The patient population in this study was biased. Most patients undergoing liver MR imaging at our institution were suspected of having a liver abnormality because of the findings of an initial evaluation with either sonography or CT. Because we are proposing the use of a rapid MR imaging protocol for focal liver lesion screening, a screening population should be the focus of a study in the future. In addition, this study only evaluated focal liver lesions. Further investigation is required to evaluate the accuracy of the 3D sequence in the detection and characterization of diffuse liver disease and extrahepatic abdominal disease.
If examination time can be reduced by selectively eliminating pulse sequences, the cost of an abdominal MR examination could be reduced, making MR imaging a cost-effective alternative to CT for liver imaging. A multiphasic gadolinium-enhanced 3D sequence alone was 97% sensitive and 95% specific in revealing clinically relevant focal liver lesions compared with a combination of 3D, T1-weighted, and fat-suppressed T2-weighted sequences. Hence, a contrast-enhanced 3D sequence alone may be adequate for either the initial or follow-up MR evaluation of the liver in certain circumstances. Routine acquisition of a late delayed 3D sequence after the portal venous phase acquisition is frequently helpful, requires little additional time, and is therefore recommended.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  M. Kanematsu, S. Goshima, H. Kondo, R. Yokoyama, K. Kajita, H. Hoshi, M. Onozuka, A. Nozaki, M. Hirano, Y. Shiratori, et al. Double hepatic arterial phase MRI of the liver with switching of reversed centric and centric K-space reordering. Am. J. Roentgenol., August 1, 2006; 187(2): 464 - 472. [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]
|
|
|
|
|
|
Y. K. Kim, C. S. Kim, G. H. Chung, S. B. Jeon, and J. M. Lee Feasibility of Application of Sensitivity Encoding to the Breath-Hold T2-Weighted Turbo Spin-Echo Sequence for Evaluation of Focal Hepatic Tumors Am. J. Roentgenol., February 1, 2005; 184(2): 497 - 504. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. K. Hussain, I. Syed, H. V. Nghiem, T. D. Johnson, R. C. Carlos, W. J. Weadock, and I. R. Francis T2-weighted MR Imaging in the Assessment of Cirrhotic Liver Radiology, March 1, 2004; 230(3): 637 - 644. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. C. Mueller, H. K. Hussain, R. C. Carlos, H. V. Nghiem, and I. R. Francis Effectiveness of MR Imaging in Characterizing Small Hepatic Lesions: Routine Versus Expert Interpretation Am. J. Roentgenol., March 1, 2003; 180(3): 673 - 680. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
= noncystic lesions (n = 60); T1 = T1-weighted sequence; T2 =
T2-weighted sequence; 3D = three-dimensional fast spoiled gradient-recalled
echo sequence.
