| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Department of Diagnostic Radiology, Kerckhoff Heart Center, Benekestrasse 2-8,
Bad Nauheim, Germany 61231.
2 Department of Angiology, University of Essen, Essen, Germany.
Received November 23, 2004;
accepted after revision April 6, 2005.
Address correspondence to A. Kluge
(a.kluge{at}kerckhoff-klinik.de).
Abstract
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
SUBJECTS AND METHODS. Sixty-two patients with symptoms indicating acute pulmonary embolism underwent an MRI protocol that progressed from real-time MRI through MR perfusion imaging to MR angiography. The results were compared with those of 16-MDCT, which was the reference standard. Thoracic incidental diagnoses other than pulmonary embolism also were sought with CT and MRI.
RESULTS. Pulmonary embolism was diagnosed with CT in 19 patients for totals of 90 lobar, 245 segmental, and 434 subsegmental arteries. On a per-patient basis, the sensitivities of real-time MRI, MR angiography, MR perfusion imaging, and the combined protocol were 85%, 77%, 100%, and 100%, respectively. The specificities were 98%, 100%, 91%, and 93%. The kappa values in a comparison of the MR techniques with CT were 0.89, 0.87, 0.86, and 0.9. On a per-embolus basis, the sensitivities of real-time MRI, MR angiography, and MR perfusion imaging for lobar pulmonary embolism were 79%, 62%, and 100%. The sensitivities for segmental pulmonary embolism were 86%, 83%, and 97%, respectively. MR perfusion imaging had a sensitivity of 93% for subsegmental pulmonary embolism. Eight of nine incidental findings revealed on CT were also subsequently diagnosed with real-time MRI. MRI failed to reveal a case of emphysema. Mean MRI examination time was 9 minutes 56 seconds.
CONCLUSION. The combined MR protocol is both reliable and sensitive in comparison with 16-MDCT in the diagnosis of pulmonary embolism. MR perfusion imaging is sensitive for the detection of pulmonary embolism, whereas real-time MR and MR angiography are specific.
Keywords: CT angiography • embolism • lung disease • MDCT • MR angiography
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Although PE can be detected with contrast-enhanced MR angiography [14, 15], and both sensitivity and specificity are within the range of those of CT, the established MR angiographic approach to PE has disadvantages that hamper widespread use: The robustness of the technique is limited because the acquisition time is several seconds [16], sufficient sensitivity for PE has been shown only for segmental or larger emboli [15], and limited patient access and the length of examinations raise concern in the care of critically ill patients.
The purposes of this study were to determine the diagnostic accuracy of MRI compared with the reference standard 16-MDCT in the diagnosis of PE; to address the issues of robustness and examination time with different MR techniques; to assess the potential advantages of a multiple-technique MRI approach over each of the single techniques; and to evaluate the diagnostic usefulness of MRI in the detection of subtle subsegmental PE. A stepwise multiple-technique MRI approach was compared with 16-MDCT. A protocol adapted to the patient's clinical condition combined real-time MRI for robustness [16, 17], MRI pulmonary perfusion imaging for sensitivity to the subsegmental level [12, 18, 19], and MR angiography for morphologic detail.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Exclusion criteria for CT examination were history of adverse reaction to iodinated contrast media and serum creatinine level greater than 2 mg/dL. Contraindications to MRI included cardiogenic shock or prolonged low cardiac output (systolic arterial pressure < 60 mm Hg) and the presence of an implanted cardiac pacemaker or implanted cardioverter aggregate, cochlear implants, neurostimulators, intracranial clips, or intraocular metallic foreign bodies. Valve implants and recent cardiac surgery were not considered contraindications to MRI. In accordance with these criteria, two patients were not included in the study because they had a serum creatinine level greater than 2 mg/dL, and one patient was excluded because of the presence of a cardiac pacemaker. Information on clinical course was collected from all patients during their hospital stays but not after hospital discharge. The local ethics committee approved the study, and informed consent was obtained.
Imaging Protocol
CT and MRI were available on a 24-hour basis. Patients were first examined
with CT and were immediately afterward transported to the adjacent MRI area.
The mean time between CT and MRI examinations was 16 ± 17 minutes
(range, 3-61 minutes).
CT—CTA was performed with a 16-MDCT scanner (Somatom Sensation Cardiac, Stratos tube, Siemens Medical Solutions) with the following scan parameters: 100 mAs with geometry-adapted dose reduction (CareDose, Siemens); 100 kV; collimation, 0.75 mm; rotation time, 0.5 seconds; reconstruction slice thickness, 1 mm; reconstruction interval, 0.7 mm. Eighty milliliters of iodinated contrast medium 400 mg I/mL (iomeprol [Imeron 400, Altana Pharma]) were injected at 4 mL/s followed by a saline chaser of 40 mL at 4 mL/s. Image acquisition started 5 seconds after a region of interest in the main pulmonary artery reached 120 H (bolus tracking).
MRI—All examinations were performed with a 1.5-T MRI scanner (Magnetom Sonata, Siemens). The patient was positioned in a supine position with a six-element phased-array coil placed over the chest. The segments were combined with two dorsal segments of the spine-array coil. A 20-gauge peripheral IV line in an antecubital vein was connected to a power injector (Spectris, Medrad). Monitors for blood oxygen saturation and ECG and a mechanical ventilator were on hand.
The MRI protocol consisted of real-time MRI, MR perfusion imaging, and MR angiography. Diagnostic procedures proceeded from urgent to indepth to ensure short examination times with robust sequences for patients in critical condition and sensitive diagnosis for patients in better condition. Reevaluation weighted the achieved diagnostic certitude against the patient's condition after each MRI sequence (30-50 seconds for transverse real-time MRI, 180 seconds for completed real-time MRI, and 7 minutes for completed MR perfusion imaging). The protocol was continued only if the patient was able to tolerate further diagnostic procedures. We calculated examinations using the DICOM tags for series, acquisition, and image time.
Real-time MRI—Three real-time true fast imaging with steady-state free precession sequences (also called balanced fast-field echo and fast imaging using steady-state acquisition) were the first step of the protocol [16]. Images were acquired in single-shot technique, which minimized motion artifacts by limiting them to the short acquisition time of each single image (acquisition time range per image, 0.4-0.5 seconds; TR/TE, 3.1/1.5; flip angle, 59° bandwidth, 975 Hz/pixel). Within 3 minutes, 320 slices with 50% overlap were acquired in three orthogonal orientations: 120 transverse slices (thickness, 3 mm; field of view, 340 mm; matrix size, 256 x 156) followed by 100 coronal slices (thickness, 4 mm; field of view, 360 mm; matrix size, 256 x 192) and 100 sagittal slices (thickness, 4 mm; field of view, 360 mm; matrix size, 256 x 180 pixels). The region from the pulmonary apex to the portal vein was covered by transverse slices, and the region from the manubrium sterni to the spinous process was imaged with coronal slices. The area from the right to the left of the medial clavicular line was imaged with sagittal slices. The inherent T2 contrast made it possible to discriminate embolus and blood without the aid of contrast agents.
Because of the oversized imaging area, no adaptations of slice geometry, position, or sequence parameters were necessary. Measurements began immediately after the patient was placed on the table. No cardiac or respiratory gating was used.
MR pulmonary perfusion imaging—Pulmonary perfusion imaging was performed with a fast 3D fast low-angle shot (FLASH) gradient-echo sequence (1.6/0.6; flip angle, 25° readout bandwidth, 1,500 Hz; matrix size range, 256 x 128 to 256 x 192; field of view, 400 x 200 mm to 400 x 300 mm; voxel size, 2.9 x 1.6 x 10 mm; block thickness, 200 mm; partitions, 20; transverse orientation). The 3D block covered the lung from the apex to the diaphragm. The mean acquisition time was 1.8 seconds (range, 1.5-2.2 seconds), depending on the number of phase-encoding steps needed to cover the patient's thoracic diameter. This sequence was repeated 25 times over 38-55 seconds for visualization of the bolus first pass. Image acquisition started simultaneously with injection of 0.5 mmol/mL gadopentetate dimeglumine solution at a dose of 0.125 mmol/kg of body weight (Magnevist, Schering Laboratories) at 4 mL/s, followed by a 20-mL saline flush at 4 mL/s. Patients were asked to hold their breath for as long as possible before resuming normal breathing to prevent movement of the diaphragm for at least 10 seconds.
MR angiography—MR angiography of the pulmonary arteries was performed with a 3D FLASH sequence (3.2/1.4; flip angle, 25° parallel acquisition technique; acceleration factor of 2 [generalized autocalibrating partially parallel acquisition]; acquisition time, 14 seconds; field of view, 340 mm; matrix size, 512 x 384; number of partitions, 72; voxel size, 0.7 x 1.2 x 1.5 mm). Fat saturation was applied to reduce infolding artifacts from the patient's arms. A bolus timing sequence determined the delay. Contrast medium (gadopentetate dimeglumine) was injected at a dose of 0.125 mmol/kg of body weight at 4 mL/s and was followed by a 20-mL saline flush. A second measurement was obtained 10 seconds later as a backup.
Additional examinations—Additional sequences for cardiac MRI were used in six patients and for MR venography in 45 patients.
Evaluation
CT—Analysis of CT images was performed on transverse slices
and coronal, sagittal, and double oblique multiplanar reformatted images
(multiplanar reconstruction), 1 mm thick. Both soft-tissue windows and lung
windows were used to identify subsegmental bronchi and arteries. Acute PE was
diagnosed when embolic material was directly visualized or when vessel
truncation implied the presence of occlusion. The level of PE was categorized
as central, lobar, segmental, subsegmental, or isolated subsegmental.
Deviations from the nomenclature
[2,
20] were that segment 7 was
not divided into subsegments and that the lingula was counted as a separate
lobe. Subsegmental PE was defined as embolic material in a segmental artery
and in a single branching subsegmental artery. Isolated subsegmental embolism
was defined as an affected subsegmental artery with a normal feeding segmental
artery. Thoracic incidental diagnoses other than PE were also sought.
MRI—Online analysis of real-time MR images and MR perfusion images was used to make immediate clinical decisions. MR angiography was assessed after examination with original coronal images and transverse multiplanar reconstruction images. Maximum intensity projection was not used.
All examinations of all patients were reevaluated for final assessment of the diagnostic accuracy of the MRI techniques used. To avoid recall bias in CT and MRI examinations, as in cases in which the analysis of one technique did not deliver clear results and might have been influenced by the obvious results of another technique, examinations were evaluated by technique and MR technique rather than by patient. Evaluations were performed by two radiologists blinded to clinical data and represent a consensus interpretation.
Real-time MRI—Acute PE was diagnosed if thrombotic material was directly visualized on more than one image in each of two planes or if vessel truncation implied an occlusion. The level of embolic material was categorized as central, lobar, or segmental. The lingula was categorized as a separate lobe.
Vessel visualization was graded as nondiagnostic if a vessel could not be identified or if blurred vessel representation precluded analysis. An entire sequence was classified as nondiagnostic if more than three lobar arteries or more than 10 segmental arteries were not assessed with the sequence. Thoracic incidental diagnoses other than PE were also sought.
MR angiography—Evaluation and criteria for diagnostic quality of MR angiography were similar to those for real-time MRI with one exception. In MR angiography, subsegmental embolism was differentiated from segmental PE.
MR perfusion—Perfusion images were analyzed both in chronological order, showing contrast medium passing at one level, and in spatial order, showing all levels at one phase. Single or multiple sharply delineated perfusion defects in accordance with subsegmental, segmental, or lobar anatomic features were classified as acute PE. The anatomic correlation of subsegmental vessels and parenchymal lung areas had been established in an earlier pilot study. References to CT images and corresponding MR perfusion sections were based on published models [21]. Cases of numerous patchy, diffuse areas of decreased perfusion with blurred delineation were classified as either chronic obstructive pulmonary disease (COPD) or pulmonary disorders other than PE. Diagnoses of COPD were verified by examination of medical records, spirometry, and CT. Criteria for diagnostic quality of MR perfusion images were signal intensity sufficient to allow differentiation of perfusion defects from normal tissue and absence of infolding or motion artifacts.
Statistical Analysis
The diagnostic accuracy (Cohen's kappa) of real-time MRI, MR perfusion
imaging, and MR angiography for the diagnosis of acute PE was assessed
separately by comparison of each technique with CTA as the reference standard.
The comparisons were made on both a per-embolus and a per-examination basis.
In addition, consensus interpretation of the three combined MRI techniques was
compared with that of CT on a per-examination basis. The diagnostic accuracy
of MRI was calculated only for patients who completed the entire imaging
protocol. Arteries not visualized with CTA, real-time MRI, or MR angiography
were excluded from further analysis. The software OpenOffice 1.1.2
(OpenOffice.org, 2004) was used for analysis.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
All of the CT, real-time MRI, and MR perfusion examinations were of diagnostic quality. Previously undetected subclavian vein thrombosis caused sparse pulmonary parenchymal enhancement in one patient but did not preclude qualitative analysis. The mean time to peak enhancement in unaffected pulmonary parenchyma was 11.3 seconds (range, 8-17 seconds). In 49 (79%) of 62 perfusion examinations, the patient's diaphragm stayed in a constant position until peak enhancement. Respiratory motion in the remaining cases did not hamper the visual qualitative analysis in this study. Images in eight (13%) of 62 MR angiographic examinations were not of diagnostic quality because of respiratory motion artifacts. We visualized 2,300 (97.62%) of 2,356 subsegmental arteries with CT. MR angiography depicted 1,384 (60.17%) of the 2,300 subsegmental arteries identified with CT (Table 1). Subsegmental arteries not visualized with MR angiography were located for the most part in the middle lobe, in the lingular segments, and in the lower lobe. The mean ± SD MRI examination time was 9 minutes 56 seconds ± 2 minutes 19 seconds (range, 1 minute 30 seconds to 14 minutes).
|
Diagnostic Accuracy of MRI
PE was diagnosed in 19 (31%) of the 62 patients who completed the MRI
protocol. Figures 1A,
1B,
1C, and
1D shows typical findings in a
patient with a high embolus load. Images obtained with all of the techniques
contained enough information for a diagnosis of PE. Polycystic kidney disease
and polycystic liver disease were also detected incidentally on real-time MRI
images. Figures 2A,
2B,
2C, and
2D shows subtle findings of
isolated subsegmental embolism. These findings were depicted only with MR
perfusion imaging.
|
|
|
|
|
|
|
|
|
|
The findings of combined MRI examinations agreed more closely with those of
CT than with those of each of the MRI techniques alone (sensitivity, 100%;
specificity, 93%; 
= 0.9). Sensitivity and specificity decreased
together with the location of embolic material, decreasing as emboli became
more peripheral. For subsegmental PE, MR perfusion imaging achieved a
sensitivity of 93.33%; the sensitivity of MR angiography was only 55.25%. If
the statistical analysis had been based on all vessels (intention to diagnose)
rather than vessels visualized with the MRI technique under investigation,
results would have been worse for MR angiography.
CT depicted isolated subsegmental embolism in 12 patients, whereas MR perfusion imaging revealed subsegmental perfusion defects in 11 cases (sensitivity, 92%) (Table 2). Because it was not possible to directly visualize segmental embolic material with MR perfusion imaging, it was impossible to differentiate isolated subsegmental embolism and nonisolated subsegmental embolism. The specificity for subsegmental PE was therefore limited to 75%.
Four MR perfusion imaging examinations had false-positive results. Perfusion defects in one patient with centrilobar emphysema and in another patient with basal pulmonary scarring were incorrectly classified as PE. In two other patients it proved impossible to pinpoint the apparent cause of subsegmental MR perfusion disturbances. The pulmonary scarring, however, was subsequently identified with real-time MRI (Figs. 3A, 3B, 3C, and 3D).
|
|
|
|
Coincidental Thoracic Findings
COPD or parenchymal lung disease other than PE was diagnosed in four
patients on the basis of findings on MR perfusion imaging and real-time MRI.
CT examinations also revealed no PE in these patients, instead showing signs
of emphysema in two of them, fibrosis in another, and chronic bronchitis in
the fourth. MRI incorrectly depicted one case of pulmonary emphysema as PE.
One aortic dissection, two cases of marked pleural effusion, and one case of
breast carcinoma with lymphangitic carcinomatosis were likewise diagnosed with
both CT and real-time MRI examinations. Thus eight of the nine coincidental
findings pinpointed at CT examinations were also revealed with MRI. In the
abdomen, polycystic liver disease was diagnosed with real-time MRI and CT;
real-time MRI also depicted polycystic kidney disease (Figs.
1A,
1B,
1C, and
1D, area not covered by
CT).
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
MRI appears to be an alternative technique in cases of suspected PE. An established indication for MRI is known allergy to iodinated contrast media, but MRI also has other positive facets in comparison with CT. Although both MRI and CT can depict signs of acute thoracic vascular disease such as aortic dissection [22] in an emergency setting [16, 17], extensive cardiac MRI can be added to the examination schedule once PE has been ruled out. Additional comprehensive analysis of the venous system with MR venography [6, 8, 23] can also be included in the protocol. Additional advantages of MRI are the good correlation between MR pulmonary perfusion and pulmonary function [10, 11] and the lack of ionizing radiation, which makes followup MRI studies reasonable.
The combination of MRI techniques had advantages over single techniques. MR
perfusion imaging was more sensitive than both real-time MR angiography and MR
angiography (Tables 2 and
3), particularly in cases of
subsegmental embolism (Figs.
2A,
2B,
2C, and
2D). Real-time MRI, on the
other hand, was more specific than MR perfusion imaging. Like CT, but unlike
techniques such as SPECT, real-time MRI was reliable in revealing thoracic
diseases that mimic PE (Figs.
3A,
3B,
3C, and
3D) and explain symptoms.
Diagnoses such as these are reported in as many as to 42% of patients with
suspected PE [24], underlining
the importance of incidental findings. As a consequence of the improved
detection of incidental findings, agreement with CTA findings was higher for
the combined MRI examination ( = 0.9) than it was for each of the
single techniques.
Although the clinical significance of subsegmental PE is controversial, a 4-22% rate of isolated subsegmental PE has been reported [25-27]. Because subsegmental PE may precede recurrent larger PE and may increase the risk of development of chronic pulmonary hypertension [28, 29], accurate depiction of subsegmental segmental PE is advantageous. Pulmonary angiography still is considered the standard for diagnosis of subsegmental PE, but interobserver agreement is low [26]. CT depicts 91-96% of subsegmental arteries [30-32] and facilitates diagnosis of subsegmental PE [33]. Compared with these results, reliable imaging of subsegmental PE with MRI poses a challenge, and only MR angiography gives sufficient spatial resolution. With MR angiography, however, motion artifacts preclude reliable analysis of subsegmental PE (Table 1). MR perfusion imaging bypasses this difficulty by depicting enhancement of smaller parenchymal regions, indicating the patency of feeding vessels that are not directly visualized.
Parallel acquisition technique has shortened the acquisition time of MR angiography and made it less susceptible to motion artifacts. Compared with the results of a study performed without parallel acquisition technique and an acquisition time of 22 seconds [16], the percentage of diagnostic-quality examinations has improved with an acquisition time of 14 seconds. In a 2004 study [34] of MR angiographic examinations conducted with an acquisition time of 4 seconds and parallel acquisition technique, there were no nondiagnostic results. Likewise, all MR perfusion sequences used in our study (acquisition time, 1.8 seconds) were of diagnostic quality. We used parallel acquisition technique for MR angiography but not for MR perfusion imaging because an acquisition time of 1.8 seconds was short enough to avoid motion blurring, and an increase in spatial resolution was neither desired nor required for pulmonary perfusion visualization. Furthermore, parallel acquisition technique entails a lower signal-to-noise ratio and additional reconstruction artifacts.
This study had limitations. For everyday clinical practice, the results can be generalized only to a limited extent. MRI and CT, for example, were available on a 24-hour basis during the study, and radiologists had ample expertise with emergency MRI. Definite pre-requisites for using the procedure on a routine basis are personal expertise with applied algorithms and experience in the clinical handling of critically ill patients. Thrombus load does not correspond exactly to perfusion impairment and clinical condition. An outcome study for evaluating the prognostic capabilities of thrombus-depicting techniques (CTA, real-time MRI, MR angiography) and perfusion imaging techniques such as SPECT [35] and MR perfusion imaging [10, 12, 18, 19] would be beneficial.
We concluded that the combined MRI protocol proved as robust and sensitive as latest-generation 16-MDCT. The protocol over-came past restrictions of MRI for emergency use and may play a larger role in routine diagnostic evaluation of suspected PE.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  J. K. Pahade, D. Litmanovich, I. Pedrosa, J. Romero, A. A. Bankier, and P. M. Boiselle Quality Initiatives: Imaging Pregnant Patients with Suspected Pulmonary Embolism: What the Radiologist Needs to Know RadioGraphics, May 1, 2009; 29(3): 639 - 654. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Remy-Jardin, M. Pistolesi, L. R. Goodman, W. B. Gefter, A. Gottschalk, J. R. Mayo, and H. D. Sostman Management of Suspected Acute Pulmonary Embolism in the Era of CT Angiography: A Statement from the Fleischner Society Radiology, November 1, 2007; 245(2): 315 - 329. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
