| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Gastrointestinal Imaging |
1 From the Department of Radiology, CB 7510, University of North Carolina, Chapel Hill, NC 27599-7510 (R.C.S., N.C.B.); Free Academic Hospital, University of Brussels, Belgium (B.O.d.B.); and the Department of Radiology, Montreal General Hospital, Canada (C.R.). Received April 13, 1998; revision requested June 29; revision received July 16; accepted October 14. Address reprint requests to R.C.S.
 |
Abstract |
|---|
 TOP Abstract IntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
MATERIALS AND METHODS: Images obtained with selected breathing-independent and breath-hold MR sequences, with 2 minutes of total acquisition time and an estimated 10 minutes of total study time, in consecutive MR examinations of the upper abdomen in 72 patients (age range, 23–87 years) were retrospectively reviewed in a blinded fashion by two separate interpreters. Determination was made of major and minor findings, and the two separate retrospective interpretations and the prospective clinical interpretation were correlated by using 
statistics. Surgical and clinical findings were also correlated with imaging findings.
RESULTS: In 61 patients, all major and minor findings were identical in the original clinical interpretation and the two retrospective readings. In 66 patients, the major findings were identical in these three readings. Close agreement was present between the two separate retrospective readings and the prospective clinical interpretation ( = 0.49–1.00).
CONCLUSION: The findings suggest that the diagnostic information provided by a shortened MR imaging protocol that includes breath-hold and breathing-independent sequences is in close agreement with lengthier MR protocols. The advantages of a shortened protocol include increased patient throughput and decreased study cost.
Index terms: Abdomen, diseases, 70.28, 70.77, 70.81, 80.81, 981.73, 99.83 • Abdomen, neoplasms, 70.30, 80.30 • Magnetic resonance (MR), comparative studies, 70.121412, 70.12143, 80.121412, 80.12143, 981.129412, 981.12943, 99.129412, 99.12943 • Magnetic resonance (MR), half-Fourier imaging, 70.121412, 70.12143, 80.121412, 80.12143, 981.129412, 981.12943, 99.129412, 99.12943 • Magnetic resonance (MR), pulse sequences, 70.121412, 70.12143, 80.121412, 80.12143, 981.129412, 981.12943, 99.129412, 99.12943
|
Introduction |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
There are clear advantages to replacing longer breathing-averaged sequences with shorter breath-hold or breathing-independent sequences. The term "breathing independent" reflects that sequences are less than 2 seconds in duration, are minimally sensitive to artifacts from patient breathing and motion, and therefore do not require that patients breathe in a regular fashion or suspend respiration. The intention of this study was to determine whether selected breath-hold and breathing-independent sequences provide as much diagnostic information as breathing-averaged sequences in a long MR examination.
|
MATERIALS AND METHODS |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
MR Imaging
MR imaging was performed with a 1.5-T system (Vision; Siemens Medical Systems, Iselin, NJ). A phased-array body coil was used in 66 of 72 studies. The majority of the MR studies of the upper abdomen were performed by using the following sequences: breathing-averaged (without respiratory triggering), fat-suppressed, T2-weighted, turbo spin-echo imaging; axial and coronal in-phase, spoiled gradient-echo (SGE) imaging; out-of-phase SGE imaging; coronal, breathing-independent, single-shot, half-Fourier rapid acquisition with relaxation enhancement (RARE) imaging; SGE imaging immediately after and 45 seconds after the intravenous administration of 0.1 mmol/kg gadopentetate dimeglumine (Magnevist; Berlex, Wayne, NJ); fat-suppressed SGE imaging 1.5 minute after gadopentetate dimeglumine administration; and SGE imaging 5–10 minutes after gadopentetate dimeglumine administration. The above sequences composed the fixed core of the lengthier original MR examinations. Approximately one-third of studies were performed with additional sequences, including additional precontrast fat-suppressed SGE and sagittal gadolinium-enhanced SGE imaging. The total estimated imaging time for the core sequences was 10 minutes; the mean estimated study duration was 45 minutes.
The selected sequences that were used for the retrospective evaluation included the following: coronal, breathing-independent, single-shot, half-Fourier RARE; in-phase SGE; out-of-phase SGE; SGE immediately and 45 seconds after gadopentetate dimeglumine administration; and fat-suppressed SGE 1.5 minute after gadopentetate dimeglumine administration. The total duration of imaging was 2 minutes, with an estimated study duration of 10 minutes. The intravenous catheter is routinely established in all patients outside the MR examination room, so the time to establish this was not included in our study duration calculation.
Table 1 displays the sequence parameters for the core sequences of the original lengthy examination and the shortened examination, with estimates of imaging and study times.
|
The retrospective interpretation was performed by two independent experienced investigators (R.C.S., B.O.d.B.). One investigator (in-house interpreter [R.C.S.]) was from the host institution (University of North Carolina at Chapel Hill, NC) and the second (outside interpreter [B.O.d.B.]) was from another site. The retrospective review was performed a minimum of 3 months after the initial interpretation to minimize the potential for study recall in the event that the in-house interpreter was familiar with or was the original interpreter of the case. The in-house interpreter prospectively reported on 38 patients (approximately one-half). The 3-month delay also provided time for surgical and clinical follow-up of the MR findings. The in-house investigator was blinded to clinical history, all information regarding the patient, and the findings of the other interpreter. The outside investigator was given the brief clinical history present on the original MR requisition sheet but was blinded to all other information. In each MR study, major findings and minor findings were determined by each of the retrospective interpreters. Major findings were considered as findings that had the potential to affect patient treatment. Minor findings were considered as findings that were not likely to affect patient treatment. A third investigator (N.C.B.) reviewed the original MR reports to determine the major and minor findings. Data were then reviewed from all patient charts to determine the surgical and clinical correlations.
The statistic was used to determine the extent of agreement between the prospective clinical interpretation and the two retrospective data sets. The MR imaging findings and the surgical and clinical follow-up findings also were correlated to ensure overall accuracy. Major and minor findings were correlated on a patient-by-patient basis.
|
RESULTS |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
= 0.49–1.00). In only one patient was the discrepancy between the data sets considered sufficiently substantial to have a potential adverse effect on patient care. The overall agreement between the three data sets for all major and minor findings was good, with exact agreement in 61 of the 72 patients. In five patients, there was a discrepancy in one major finding (Table 3, Fig 1). In five patients, there was a discrepancy in one minor finding (Table 4). In one patient, there was a discrepancy in one major and one minor finding. The major finding was an undefined 1-cm liver lesion that was missed by the one reader. Results of biopsy of the liver lesion revealed no malignant cells. The minor finding was one small (<1-cm) gallstone diagnosed by the first reader but missed by the second reader and also not identified on the original report. No management change was made on the basis of the gallstone finding. Comprehensive demonstration of various findings with the shortened protocol was good (Fig 2). There were no systematic interpretive errors of any particular entity when the findings from the shortened protocol were compared with the findings from the lengthier protocol.
|
|
|
|
|
|
|
|
|
|
|
|
DISCUSSION |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Contributing factors to the limited use of abdominal MR imaging include lack of adequately trained radiologists; few state-of-the-art systems that are able to perform new sequences; the perception that MR imaging is inferior to computed tomography (CT); lack of reproducible, good image quality; and long study times. The first two points are beyond the scope of this study but are time-limited and reflect that currently, to our knowledge, few centers train radiologists to interpret body MR images and that there is a time delay for older MR imaging systems to be replaced by new ones. A major holdup for more widespread use of abdominal MR imaging is the general perception that MR imaging is inferior to CT. This again is beyond the scope of this study; however, recent spiral CT–MR imaging comparisons on state-of-the-art systems have shown that MR imaging is superior to CT for the demonstration of many abdominal neoplastic diseases (4,5,7).
Lack of reproducible good image quality and lengthy examination times are both problems that are addressed by the development of breath-hold and breathing-independent sequences. The most problematic artifact in abdominal MR imaging is the breathing artifact (3), and the most successful method of dealing with this problem is to avoid breathing artifacts by using either breath-hold or breathing-independent sequences. The important by-product of these short sequences is that they permit the dramatic shortening of examinations.
The explanation of why the implementation of short sequences has not yet resulted in a substantial shortening of MR examinations may be that in many cases it is not clear in a comparison of short and long sequences to what extent information is not obtained or lost. Therefore, there is a hesitancy in the general community to remove longer sequences in favor of shorter sequences because of the desire not to sacrifice information (9). The cumulative results of prior studies suggest that replacing a breathing-averaged spin-echo sequence with a breath-hold SGE sequence does not result in any substantial loss of information for T1-weighted images, with perhaps a gain in reproducibility and lesion detection (2,3,10). However, the problem with SGE imaging is that the patient still must cooperate sufficiently to suspend respiration for 10–20 seconds.
The advantages of breath-hold or breathing-independent T2-weighted sequences over breathing-averaged conventional or fast spin-echo sequences are not as clear-cut. In fact, it does appear clear that on a sequence-by-sequence basis, the short sequences do not demonstrate solid hepatic lesions as well as longer sequences (10–12). This observation explains why many centers that have the ability to use short T2-weighted sequences have not ceased using the longer breathing-averaged sequences. Evaluating MR protocols on a sequence-by-sequence basis is suboptimal because MR studies are interpreted not as individual sequences but as all the sequences combined. Lesions that are not well seen with one sequence may be well seen with another. In fact, many lesions that are not well seen on T2-weighted images and that are therefore particularly poorly seen with short, echo-train spin-echo sequences are relatively well shown on T1-weighted images, particularly T1-weighted SGE images acquired immediately following gadopentetate dimeglumine administration (13–15).
We believe that the evolution of MR imaging protocols requires the substitution of older, lengthier sequences by newer, shorter sequences to increase reproducible image quality and patient throughput. It may, however, be important to develop a protocol design sufficiently comprehensive without much redundancy. In the current study, the sequences selected for the shortened protocol were specifically selected to provide comprehensive information while minimizing repetitious information. Our rationale for using the sequences we selected are briefly as follows.
1. We used coronal, breathing-independent, T2-weighted single-shot half-Fourier RARE to provide information on the fluid content of liver lesions and other focal lesions in other organs; to depict the biliary and pancreatic ductal system, calculous disease of the biliary system, and the iron content of the liver, spleen, and pancreas; and to determine the appearance of the bowel.
2. We used transverse SGE to provide information on overall tissue morphology, to detect lesions with higher fibrous or fluid content than background organs or tissues, and to detect subacute blood.
3. We used transverse out-of-phase SGE to provide information on the presence of fatty liver and the fat content of adrenal masses and other fatty lesions.
4. We used SGE imaging immediately after gadopentetate dimeglumine administration to detect the majority of lesions in solid upper abdominal organs, to characterize focal hepatic lesions and pancreatic lesions, and to provide information on arterial vessels and the blood supply to organs and lesions.
5. We used SGE imaging 45 seconds after gadopentetate dimeglumine administration to improve the conspicuity of hypovascular liver lesions, to provide information on the patency of the hepatic vessels, and to serve as backup for the SGE imaging performed immediately after gadopentetate dimeglumine administration in the event of inaccurate timing of image acquisition or breathing during image acquisition immediately after contrast agent administration.
6. We used fat-suppressed SGE imaging performed 1.5 minute after contrast agent administration to provide information on peritoneal or capsular disease, cholangiocarcinoma, lymph nodes, renal parenchymal masses, and inflammatory diseases.
As new sequences are developed that are faster and more consistently good in image quality and that provide greater diagnostic information, replacement of various of the sequences described earlier is anticipated.
Single-shot spin-echo-train T2-weighted sequences, such as single-shot half-Fourier RARE, are relatively insensitive to T2* field inhomogeneities (16). As a result, they are relatively insensitive to conditions such as mild iron deposition. We observed this in our study in the missed minor diagnosis of mild iron deposition by both of the retrospective readers. It is, however, our impression that the single-shot half-Fourier RARE sequence misses only cases of mild iron deposition, whereas cases of potentially clinically important iron deposition are detected. In our clinical experience, we have found that substantial iron deposition is well shown even on long-echo-train spin-echo sequences such as single-shot half-Fourier RARE. Iron deposition in the liver and spleen may result in signal intensity loss in these organs on SGE images obtained with an echo time of 4 msec, and the combined use of single-shot half-Fourier RARE and SGE may permit adequate detection of iron deposition. It should be noted, however, that the ability of single-shot half-Fourier RARE imaging to depict changes of early idiopathic hemochromatosis is not known at present, and, in patients suspected to have this entity, acquisition of long–echo time spin-echo or gradient-echo images may still be prudent. On the basis of our findings in this study, we have adopted a shortened protocol for our clinical practice; however, we believe that an additional fat-suppressed single-shot half-Fourier RARE sequence should be used in the transverse plane to facilitate the detection of calculi in the gallbladder and liver lesions.
Limitations of this study are that data were acquired at one institution and therefore provide a limited database. The retrospective readings were performed by experienced interpreters and may not be directly indicative of readings by the general radiologist. One of the retrospective readers was an in-house reader and may therefore have recalled some of the MR cases. We attempted to minimize this by performing the retrospective readings more than 3 months after the clinical reading. The retrospective readings are also by their nature less accurate than readings that would have been acquired if the entire study had been a larger, randomized, controlled study. A limited number of patients in this study had histopathologic proof. This, however, reflects that we included consecutive patients who underwent MR examination and that, in general, not many patients who undergo imaging investigation subsequently have specimens obtained for histopathologic analysis. The authors introduced their own bias in sequence selection; other approaches, such as using only breathing-independent T1- and T2-weighted sequences, would have resulted in even shorter examinations. Shorter breathing-independent protocols have been described (16,17); however, a number of disease processes may not be well shown with nonenhanced breathing-independent T1- and T2-weighted sequences. Peritoneal disease would not be visualized well, and many other disease processes, such as hepatocellular carcinoma, small pancreatic tumors, splenic lesions, varices, and renal masses, might be suboptimally studied. The direction we prefer to explore with a shortened MR protocol is the maintenance of a variety of sequences that provide comprehensive information and that collectively may be superior rather than equivalent to CT. The routine use of gadolinium-based contrast agents in all patients who undergo abdominal MR imaging, which is the practice at our institution (University of North Carolina), is also not universally performed. Last, another limitation of the study is that this shortened protocol needs to be compared with CT to define its true role for the evaluation of abdominal diseases.
In summary, the results of our study have shown that a short MR protocol with breath-hold and breathing-independent sequences has results in close agreement with those of a long MR protocol with breathing-averaged sequences. The 10-minute comprehensive MR examination of the upper abdomen is currently a feasible diagnostic approach.
|
Footnotes |
|---|
Author contributions: Guarantor of integrity of entire study, R.C.S.; study concepts and design, R.C.S., N.C.B.; definition of intellectual content, R.C.S., N.C.B., B.O.d.B.; literature research, N.C.B.; clinical studies, R.C.S., B.O.d.B., N.C.B.; data acquisition, N.C.B., B.O.d.B.; data analysis, N.C.B., C.R.; statistical analysis, C.R.; manuscript preparation, R.C.S., N.C.B.; manuscript editing, R.C.S.; manuscript review, R.C.S., B.O.d.B., N.C.B.
|
References |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  A. S. d. A. Bezerra, G. D'Ippolito, R. P. Caldana, D. D. Leopoldino, G. R. Batista, D. R. Borges, G. de Jesus Lopes Filho, and M. Ahmed Differentiating Cirrhosis and Chronic Hepatosplenic Schistosomiasis Using MRI Am. J. Roentgenol., March 1, 2008; 190(3): W201 - W207. [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]
|
|
|
|
 |
|
 R. M. Kwee and T. C. Kwee Imaging in Local Staging of Gastric Cancer: A Systematic Review J. Clin. Oncol., May 20, 2007; 25(15): 2107 - 2116. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. C. Semelka and T. K. G. Helmberger Contrast Agents for MR Imaging of the Liver Radiology, January 1, 2001; 218(1): 27 - 38. [Abstract] [Full Text]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
/90), (c) precontrast T1-weighted SGE image (150/4), and (d) SGE image (140/4) obtained immediately after contrast agent administration. A minimally hyperintense lesion (arrow) is poorly visualized in a and b. The tumor (arrow) is well shown in c and exhibited a predominant rim pattern of enhancement in d, which is consistent with a malignant tumor. The tumor is clearly seen with the sequences used for the short protocol (c and d) while poorly seen with a sequence used for the long protocol from the original data set (a) and with the single-shot half-Fourier RARE sequence (c). The rim pattern of enhancement led one of the retrospective interpreters to describe the tumor as metastasis rather than as hepatocellular carcinoma.
