Purpose: Materials and Methods: Results: Conclusion: Original Research n Thoracic Imaging

Transcription

1 Note: This copy is for your personal non-commercial use only. To order presentation-ready copies for distribution to your colleagues or clients, contact us at MR Imaging of Pulmonary Embolism: Diagnostic Accuracy of Contrast-enhanced 3D MR Pulmonary Angiography, Contrast-enhanced Low Flip Angle 3D GRE, and Nonenhanced Free-Induction FISP Sequences 1 Original Research n Thoracic Imaging Bobby Kalb, MD 2 Puneet Sharma, PhD 2 Stefan Tigges, MD, MSCR Gaye L. Ray, NP Hiroumi D. Kitajima, PhD James R. Costello, MD, PhD 2 Zhengjia Chen, PhD Diego R. Martin, MD, PhD, FRCPC 2 1 From the Department of Radiology, Emory University School of Medicine, Atlanta, (B.K., P.S., S.T., G.L.R., H.D.K., J.R.C., D.R.M.); and Department of Biostatistics and Bioinformatics, Emory University School of Public Health, Atlanta, Ga (Z.C.). Received February 1, 2011; revision requested March 24; final revision received August 2; accepted August 11; final version accepted November 16. Address correspondence to D.R.M., Department of Radiology, University of Arizona, University Medical Center, 1501 Campbell Ave, P.O. Box , Tucson, AZ ( dmartin@radiology.arizona.edu). 2 Current address: Department of Radiology, University of Arizona, University Medical Center, Tucson, Ariz. q RSNA, 2012 Purpose: Materials and Methods: Results: Conclusion: To evaluate relative detection of pulmonary embolism (PE) with standard bolus-triggered contrast-enhanced breathhold magnetic resonance (MR) pulmonary angiography, contrast-enhanced recirculation-phase breath-hold low flip angle three-dimensional (3D) gradient-echo (GRE), and nonenhanced free-induction cardiac- and respiratory-triggered true fast imaging with steady-state precession (FISP) MR sequences. The study was HIPAA compliant and institutional review board approved. Twenty-two patients with a computed tomographic (CT) angiography diagnosis of PE underwent MR imaging within 48 hours of CT. MR included three complementary techniques: MR pulmonary angiography, 3D GRE, and triggered true FISP. Each sequence was analyzed separately by two independent reviewers who recorded presence of emboli in categorized pulmonary artery anatomic territories. CT angiography results were analyzed by a third independent reviewer, who retrospectively recorded presence of emboli using the same format; these results served as the reference standard. Sensitivity, specificity, and positive and negative predictive values for PE detection were calculated for each MR technique on a per-embolus basis, and 95% confidence intervals were calculated according to the efficient-score method. A two-sample t test was used to compare values among MR techniques. Sensitivities for PE detection were 55% for MR pulmonary angiography, 67% for triggered true FISP, and 73% for 3D GRE MR imaging. Combining all three MR sequences improved overall sensitivity to 84%. Specificity was 100% for all detection methods except for MR pulmonary angiography (one false-positive). Agreement between readers was high (k = 0.87). Embolus detection rates were lowest in the lingula branch for all MR sequences compared with remainder of the vascular territories (P =.07). There are complementary benefits to combining standard MR pulmonary angiography, 3D GRE, and triggered true FISP MR examinations for evaluation of PE. q RSNA, 2012 Radiology: Volume 263: Number 1 April 2012 n radiology.rsna.org 271

2 The clinical manifestation of pulmonary embolism (PE) is often nonspecific and variable, leading to under- or misdiagnosis (1 4). Because of this, imaging has assumed a central role in the diagnosis of PE. Multidetector computed tomographic (CT) pulmonary angiography is highly sensitive and specific for the diagnosis of PE (5) and has become the imaging method of choice in patients suspected of having PE. Limitations of CT include exposure to ionizing radiation, with an effective dose of approximately 10 msv to radiosensitive parts of the body (6) and a dose of approximately 20 msv to the glandular tissue of the breast (3). Current risk models estimate that a 40-year-old woman who undergoes CT pulmonary angiography has a one in 620 risk of developing a radiation-induced cancer (7), and this risk doubles for a Advances in Knowledge nn Contrast-enhanced low flip angle 3D GRE images acquired during the recirculation phase, without the need for bolus-triggered acquisition timing, and free-induction respiratory-triggered cardiac-gated true FISP MR images, acquired with the patient freely breathing, provided more sensitive detection of pulmonary embolism (PE) (73% and 67%, respectively), as compared with conventional bolus-triggered 3D MR pulmonary angiography (55%) (P =.03 [GRE] and.16 [triggered true FISP]). nn Low flip angle 3D GRE, triggered true FISP, and MR pulmonary angiography all demonstrated high specificity for evaluation of PE (100%, 100%, and 99%, respectively). nn Detection of PE is maximized (84% sensitivity) by combining three imaging sequences (3D MR pulmonary angiography, low flip angle 3D GRE, and triggered true FISP) as part of a comprehensive MR examination, in contrast to any single imaging sequence alone. 20-year-old woman. Patients evaluated for PE in the emergency department are frequently young and often present with other disease processes that may mimic symptoms of PE (8). It has been noted that CT pulmonary angiography performed in emergency room patients for detection of PE shows a low yield for positive examinations, with one study (9) finding only a 5% incidence of positive examinations in patients between the ages of 18 and 45 years. Additionally, CT pulmonary angiography requires the administration of iodinated contrast material, and the incidence of contrast material induced nephropathy following CT angiography has been reported to be as high as 12% in patients with paired measurements (10). Magnetic resonance (MR) imaging may provide an important safer alternative to CT pulmonary angiography for the diagnosis of PE, free of long-term cancer risks from ionizing radiation, and potentially has overall reduced risks of complications from contrast material, particularly if newer nonenhanced MR imaging methods are used. Optimized methods for MR evaluation of PE have been proposed (11 13), but little work is available to show systematic comparison of optimal contrast material enhanced or nonenhanced MR imaging techniques. The aim of our study was to evaluate relative detection of pulmonary emboli with standard bolus-triggered contrast-enhanced breath-hold MR pulmonary angiography, contrast-enhanced recirculation-phase breath-hold low flip angle three-dimensional (3D) gradient echo (GRE), and nonenhanced free-induction respiratory-triggered cardiac-gated true fast imaging with steadystate precession (FISP) sequences. Implication for Patient Care nn This study shows potential gains in PE detection by using nonstandard vascular MR imaging methods designed to bypass the need for accurately timed arterial phase contrast enhancement, breath holding, or obviate the need for contrast material administration altogether. Materials and Methods Patients Our study was approved by the institutional review board and was Health Insurance Portability and Accountability Act compliant. Patients with a confirmed diagnosis of PE on the basis of CT pulmonary angiography performed at the time of admission from our emergency facilities were included. Patients were approached with the opportunity to enroll in our study through the process of written informed consent. Exclusion of patients was based on an inability to perform study protocol MR imaging within 48 hours of the CT pulmonary angiography. Five patients were excluded on this basis (three women [average age, 57 years; range, years] and two men [average age, 62 years; range, years]). Patients with CT pulmonary angiography results that were positive for PE and subsequent MR imaging within 48 hours were collected over a period from August 2007 to July There were a total of 22 patients who fulfilled our inclusion criteria during this time period (average age, 61 years; range, years). The final study cohort consisted of 13 women (average age, Published online /radiol Radiology 2012; 263: Content code: Abbreviations: FISP = fast imaging with steady-state precession GRE = gradient echo PE = pulmonary embolism PIOPED = Prospective Investigation of Pulmonary Embolism Diagnosis 3D = three-dimensional Author contributions: Guarantors of integrity of entire study, B.K., H.D.K., Z.C., D.R.M.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; approval of final version of submitted manuscript, all authors; literature research, B.K., Z.C., D.R.M.; clinical studies, all authors; statistical analysis, B.K., Z.C., D.R.M.; and manuscript editing, B.K., J.R.C., Z.C., D.R.M. Potential conflicts of interest are listed at the end of this article. 272 radiology.rsna.org n Radiology: Volume 263: Number 1 April 2012

3 68 years; range, years) and nine men (average age, 53 years; range, years). Symptoms at presentation included chest pain (n = 9), dyspnea on exertion (n = 8), right upper quadrant abdominal pain (n = 1), syncope (n = 1), and leg swelling (n = 1), but two patients were asymptomatic. CT Pulmonary Angiography All CT pulmonary angiography scans were obtained following the standards of emergency medicine care, with the indication being a differential diagnosis of PE. CT pulmonary angiography was performed by using a 64-section CT scanner (Lightspeed VCT; GE Healthcare, Milwaukee, Wis) with a standardized PE protocol with the following parameters: pitch of 0.984:1; 40-mm coverage per rotation; tube rotation speed, 0.5 second; section thickness, mm; matrix, ; field of view, 388 mm 2 ; tube voltage, 120 kv; automated tube current, ma; noise index, 21. Low-dose scout imaging was performed, and the scan coverage was set to extend from the lung apices to the inferior costophrenic angle. Timing of the image acquisition was performed with a bolus tracking technique, with a bolus detection region of interest placed over the main pulmonary artery and the threshold for initiation of the CT acquisition set to HU. Contrast enhancement was achieved with 80 ml of 350 mg of iodine per milliliter iohexol (Omnipaque; GE Healthcare, Princeton, NJ) administered intravenously through an 18- or 20-gauge catheter in the antecubital vein at an injection rate of 4 ml/sec by using a power injector (Empower CTA; ACIST Medical Systems [Bracco], Eden Prairie, Minn). Pulmonary MR Imaging All MR imaging examinations were performed by using a 1.5-T MR system (Magnetom Avanto; Siemens Medical Systems, Erlangen, Germany) equipped with 45 mt/m gradients operating up to a 200 T/m/sec slew rate. Eight-channel anterior and six-channel posterior phased-array coils were used for signal reception from the chest. Our study was designed for analysis of three MR imaging techniques. Pulmonary vascular imaging was performed by using the following methods: (a) non breath-hold free-induction true FISP imaging, acquired without exogenous contrast material administration and with electrocardiographic and respiratory triggering; (b) gadolinium chelate based contrast-enhanced bolus-triggered breath-hold MR pulmonary angiography; and (c) contrast-enhanced recirculation-phase breath-hold 3D GRE imaging with a volumetric interpolated breathhold examination sequence (14). Triggered true FISP MR imaging. Free-breathing true FISP MR images were acquired in the axial plane using electrocardiographic and respiratory triggering, with the following parameters: repetition time msec/echo time msec, 3.3/1.5; flip angle, 70 ; field of view, 375 mm 2 ; matrix, 256; partial Fourier imaging; sections, 50; section thickness, 3 mm; no gap; and bandwidth, 1000 Hz per pixel. For electrocardiographic triggering, chest leads were used for detection. Respiratory triggering was performed with a two-dimensional navigator sequence, with the center of the trigger box placed over the dome of the right hemidiaphragmliver interface. MR pulmonary angiography. Contrast-enhanced MR pulmonary angiography was performed with gadobenate dimeglumine (Multihance; Bracco Imaging, Milan, Italy) administered intravenously at a dose of 0.1 mmol per kilogram of body weight and a flow rate of 2 ml/sec, followed by a 30-mL saline flush at 2 ml/sec by using a dual-chamber power injector (Spectris; Medrad, Warrendale, Pa). Optimum enhancement of the pulmonary arterial vessels was achieved by using a real-time bolus-tracking method. Coronal thickslab real-time images were acquired at one image per second, starting with the initiation of contrast material injection. When the bolus was detected at the subclavian vein, the patient was instructed to take in and hold a breath. The patient was given 3 seconds to follow the breathing instructions and to stabilize motion, at which point the MR pulmonary angiography acquisition was initiated in the coronal plane by using the following parameters: 3.3/1.1; flip angle, 30 ; field of view, 380 mm 2 ; matrix, 320; phase resolution, 90%; partial Fourier imaging; sections, 88; section thickness, 1.3 mm; bandwidth, 380 Hz per pixel; and acquisition time, 12 seconds. A second breath-hold MR pulmonary angiography acquisition with identical imaging parameters was obtained as soon as the patient was able to comply after the first acquisition. An MR pulmonary angiography data set acquired before the administration of contrast material served as a subtraction mask. The 3D MR pulmonary angiography coronal source data were reformatted in the axial dimension by using a 2-mm section reconstruction thickness with 50% overlap. In addition, multiplanar maximum intensity projections were obtained and rotated 360 with 3 increments around the coronal plane. 3D GRE MR imaging. Immediately following MR pulmonary angiography, axial breath-hold 3D GRE imaging (ivibe; Siemens Medical Solutions) was performed with the following parameters: 3.8/2.1; flip angle, 10 ; field of view, 380 mm 2 ; matrix, 288; phase resolution, 80%; partial Fourier imaging; sections, 96; section thickness, 2.8 mm; bandwidth, 350 Hz per pixel; and acquisition time, 15 seconds. The typical time delay from injection of the contrast material was 3 minutes. Image Analysis For both CT pulmonary angiography and MR imaging, findings were recorded on standardized reporting forms. A modification of a previously described (15) division of the lungs into vascular territories was used, which divided the pulmonary arterial tree into specified arterial branches (Table 1). Segmental and subsegmental emboli were assigned to the vascular distribution of the lobar pulmonary artery supplying the territory. This division of the pulmonary vascular tree approximates the method of image interpretation in which the interpreter tracks the course of each pulmonary vascular branch by scrolling through Radiology: Volume 263: Number 1 April 2012 n radiology.rsna.org 273

4 Table 1 Summary Statistics for MR Imaging Detection of Pulmonary Emboli according to Pulmonary Artery Segment Sequence and Statistics Right Right Pulmonary Artery Upper Lobe Middle Lobe Lower Lobe Upper Lobe Lingula Lower Lobe Left MR angiography Sensitivity 100 (1/1) 50 (6/12) 50 (5/10) 69 (11/16) 57 (4/7) 29 (2/7) 57 (8/14) Specificity 100 (21/21) 90 (9/10) 100 (12/12) 100 (6/6) 100 (15/15) 100 (15/15) 100 (8/8) Positive predictive value 100 (1/1) 86 (6/7) 100 (5/5) 100 (11/11) 100 (4/4) 100 (2/2) 100 (8/8) Negative predictive value 100 (21/21) 60 (9/15) 71 (12/17) 55 (6/11) 83 (15/18) 75 (15/20) 57 (8/14) 3D GRE MR imaging Sensitivity 100 (1/1) 67 (8/12) 70 (7/10) 81 (13/16) 57 (4/7) 43 (3/7) 93 (13/14) Specificity 100 (21/21) 100 (10/10) 100 (12/12) 100 (6/6) 100 (15/15) 100 (15/15) 100 (8/8) Positive predictive value 100 (1/1) 100 (8/8) 100 (7/7) 100 (13/13) 100 (4/4) 100 (3/3) 100 (13/13) Negative predictive value 100 (21/21) 71 (10/14) 80 (12/15) 67 (6/9) 83 (15/18) 79 (15/19) 89 (8/9) Triggered true FISP MR imaging Sensitivity 100 (1/1) 58 (7/12) 70 (7/10) 75 (12/16) 71 (5/7) 43 (3/7) 71 (10/14) Specificity 100 (21/21) 100 (10/10) 100 (12/12) 100 (6/6) 100 (15/15) 100 (15/15) 100 (8/8) Positive predictive value 100 (1/1) 100 (7/7) 100 (7/7) 100 (12/12) 100 (5/5) 100 (3/3) 100 (10/10) Negative predictive value 100 (21/21) 67 (10/15) 80 (12/15) 60 (6/10) 88 (15/17) 79 (15/19) 67 (8/12) P value* Note. Unless otherwise specified, data are percentages, and numbers in parentheses were used to calculate the percentages. No thrombi were present in the left pulmonary artery. * Comparison of 3D MR angiography versus 3D GRE MR imaging versus triggered true FISP MR imaging. multiple contiguous sections. Data sets from all CT pulmonary angiographic examinations and each separate MR imaging sequence were evaluated at separate times by using image review software (efilm, version 3.0; Merge Healthcare, Chicago, Ill) with identifying patient data removed prior to transfer to the review station. Two radiologists (D.R.M. and B.K., with 14 and 5 years, respectively, of subspecialty experience in vascular MR imaging) interpreted all examination findings. The reviewers were asked to record the presence of PE in each of eight different pulmonary arteries and/or vascular territories (right pulmonary artery, right upper lobe, right middle lobe, right lower lobe, left pulmonary artery, left upper lobe, lingula, and left lower lobe) for each examination. A PE was defined as a nonenhancing intraluminal filling defect within the specified pulmonary arterial branch (Fig 1). A negative result for PE was defined as the absence of a clot in all three imaging sequences. Results of image interpretation were recorded on standardized reporting forms. Figure 1 Figure 1: Pulmonary embolus (arrow) at bifurcation of the right pulmonary artery in a 67-year-old woman. (a) CT pulmonary angiogram. (b) Axial reformation of bolus-triggered 3D MR pulmonary angiogram (3.3/1.1; flip angle, 30 ). (c) Recirculation-phase axial 3D GRE MR image (3.3/2.1; flip angle, 10 ). (d) Axial freeinduction triggered true FISP MR image (3.3/1.5; flip angle, 70 ). 274 radiology.rsna.org n Radiology: Volume 263: Number 1 April 2012

5 Table 2 Summary Statistics for MR Imaging Detection of Pulmonary Emboli on a Per-Embolus Basis Statistic MR Angiography 3D GRE MR Imaging Triggered True FISP MR Imaging Combined MR Data* Sensitivity 55 (37/67 [43, 67]) 73 (49/67 [61, 83]) 67 (45/67 [54, 78]) 84 (56/67 [72, 91]) Specificity 99 (108/109 [94, 100]) 100 (109/109 [96, 100]) 100 (109/109 [96, 100]) 100 (109/109 [96, 100]) Positive predictive value 97 (37/38 [85, 100]) 100 (49/49 [91, 100]) 100 (45/45 [90, 100]) 98 (56/57 [89, 100]) Negative predictive value 78 (108/138 [70, 85]) 86 (109/127 [78, 91]) 83 (109/131 [75, 89]) 91 (108/119 [84, 95]) Note. Data are percentages. Numbers in parentheses were used to calculate the percentages, and numbers in brackets are 95% confidence intervals. * Includes each individual embolus identified on images generated with any of the three techniques. Figure 2 Figure 2: Segmental pulmonary embolus (arrow) in a 48-year-old man. (a) CT pulmonary angiogram shows focal thrombus in a right lower lobe segmental pulmonary artery. Thrombus was not identified during initial reading of the (b) MR pulmonary angiography (3.3/1.1; flip angle, 30 ) axial reformations. In retrospect, there is a subtle deformity of the vessel lumen correlating to the thrombus; lack of visualization of the vessel wall hinders embolus detection, with low signal intensity intrinsic to the thrombus showing no image contrast versus adjacent aerated lung. (c) Axial recirculation-phase 3D GRE (3.3/2.1; flip angle, 10 ) and (d) axial triggered true FISP (3.3/1.5; flip angle, 70 ) MR images show the thrombus, which contrasts well against the vessel wall. Since signal is generated from the vessel wall around the outer boundary of the intraluminal thrombus, there is improved conspicuity delineating the outer margin of the thrombus from the adjacent aerated lung. MR imaging. The reviewers (D.R.M. and B.K.) performed analysis for each sequence separately during independent reading sessions performed on different days with a minimum gap of 48 hours between reading sessions. The readers recorded the anatomic location of emboli on standardized forms while blinded to the CT results. Reviewers were also asked to assess the image quality for each sequence used for detection of PE (triggered true FISP MR imaging, 3D MR pulmonary angiography, and 3D GRE MR imaging). Specifically, wrap artifacts, motion artifacts, signal-to-noise ratio, and contrastto-noise ratio were assessed following the previously reported method used in the Prospective Investigation of Pulmonary Embolism Diagnosis (PIOPED) III study (13). CT pulmonary angiography. Each CT pulmonary angiography examination was retrospectively reviewed by a chest radiologist (S.T., with 15 years of experience in pulmonary imaging). The reviewer was asked to provide a detailed retrospective recording of the presence of a PE in each of the pulmonary arteries by anatomic division. The interpreting radiologist was blinded to the MR images and results but had access to the initial CT pulmonary angiography report with the objective of maximizing reader accuracy in order to use the CT pulmonary angiography results as the reference standard. Statistical Analysis Software (SAS, version 9.2; SAS Institute, Cary, NC) was used to perform all data analysis. Sensitivity, specificity, accuracy, and positive and negative predictive values were estimated as proportions, and 95% confidence intervals were calculated according to the efficient-score method, with a correction for continuity (14,15). The general linear model of the software was used to compare the accuracy among 3D MR pulmonary angiography, 3D GRE MR imaging, and triggered true FISP MR imaging within each lung territory. A t test was used to compare the accuracy of the total MR examination (3D MR pulmonary angiography, 3D GRE MR imaging, and triggered true FISP MR imaging combined) results in the lingula with every other lung territory. The significance level was set at.05 for all tests. An unweighted k coefficient was estimated as a measurement for the Radiology: Volume 263: Number 1 April 2012 n radiology.rsna.org 275

6 level of agreement between the two MR imaging reviewers (D.R.M. and B.K.). Figure 3 Results The average time between CT pulmonary angiography and MR imaging was 30.3 hours (range, 8 44 hours). A total of 67 pulmonary emboli were detected at CT, with 45% (30 of 67) of emboli detected in the lower lobes. The results of MR imaging interpretation categorized by vascular distribution are summarized in Table 1 and presented for each of the MR imaging techniques, with CT used as the reference standard. Table 2 summarizes the results on a per-embolus basis, taking into account all pulmonary emboli detected at CT in comparison with pulmonary emboli detected with each of the MR imaging techniques and also in comparison with pulmonary emboli detected with the combined MR imaging techniques. The result of combining the techniques improved the sensitivity and the negative predictive value versus any of the techniques taken alone (Figs 1 4). There was good overall agreement between the two MR imaging readers (k = 0.87) (16). Of 67 total pulmonary emboli detected at CT, 56 were detected with the overall MR examination. The 3D GRE sequence had the highest accuracy (89.8% [158 of 176]) in diagnosis, and the 3D MR pulmonary angiography sequence demonstrated the lowest sensitivity (55% [37 of 67]) for pulmonary emboli (Fig 2). The 3D MR pulmonary angiography sequence also produced the only false-positive result for thrombus, although this sequence did also demonstrate a specificity of 99%. No false-positive results were identified on either 3D GRE or triggered true FISP images. Differences in PE detection between 3D MR pulmonary angiography, 3D GRE, and triggered true FISP sequences were not significant (Table 1), with P values ranging from.13 to greater than.99 for the comparisons between each imaging technique individually for each vascular anatomic segment. Note that results from sequences that were judged to be technically Figure 3: Pulmonary embolus in a 60-year-old man with severe tachypnea and limited breath-hold ability. (a) CT pulmonary angiogram shows thrombus (arrow, also on d) in the left upper lobe pulmonary arterial branch. (b) MR pulmonary angiography (3.3/1.1; flip angle, 30 ) axial reformation and (c) axial 3D GRE MR image (3.3/2.1; flip angle, 10 ) are degraded by respiratory motion, limiting evaluation. (d) However, axial triggered true FISP MR image (3.3/1.5; flip angle, 70 ), which is acquired with respiratory triggering while the patient is permitted to breathe freely, shows the thrombus. This case shows the benefits of a combined strategy of sequences with complementary characteristics. Like CT, MR imaging provides an evaluation of other disease processes that may lead to chest pain and tachypnea. Also note large bilateral pleural effusions (arrowheads in a and d). inadequate were included in the statistical analysis. There were a total of 11 pulmonary emboli not identified with any of the MR imaging techniques. Eight were in a segmental pulmonary arterial branch, and three were present in a subsegmental branch. The overall sensitivity for the detection of thrombi within the pulmonary vasculature supplying the lingula was only 43% (three of seven) for all MR imaging techniques (Fig 4). In contrast, the sensitivity for embolus detection within the remainder of the lobar or segmental pulmonary vessels for the overall MR imaging examination ranged from 80% (eight of 10) in the right upper lobe to 93% (13 of 14) in the left lower lobe, with the highest sensitivities for embolus detection recorded for the lower lobes bilaterally. Difference in sensitivity for detection of emboli in the lingula compared with all other pulmonary vascular anatomic segments demonstrated a P value of.07. Assessment of image quality results are presented in Table 3. In both 3D MR pulmonary angiography and 3D GRE MR imaging, 14% (three of 22) of examinations were found to be inadequate secondary to motion artifact (Fig 3). However, in only one patient did these technical failures overlap. The triggered true FISP technique retained adequate image quality in this patient. The triggered true FISP technique was found to be inadequate in 4% (one of 23) of patients owing to failed respiratory triggering. 276 radiology.rsna.org n Radiology: Volume 263: Number 1 April 2012

7 Figure 4 Table 3 Technical Adequacy of MR Sequences Patient No. MR Angiography 3D GRE MR True FISP MR Figure 4: Lingular pulmonary embolus, which was not identified on MR images, in a 36-year-old woman. (a) CT pulmonary angiogram shows a mural thrombus (arrow, also on b d) in the pulmonary arterial branch supplying the lingula, in addition to a focal thrombus (arrowhead) at the origin of the left lower lobe pulmonary artery. (b) MR pulmonary angiography (3.3/1.1; flip angle, 30 ) axial reformation, (c) 3D GRE MR image (3.3/2.1; flip angle, 10 ), and (d) axial triggered true FISP MR image (3.3/1.5; flip angle, 70 ) show the focal clot (arrowhead) in the left lower lobe pulmonary artery, but the mural thrombus in the lingular branch was not identified. This case shows the potential for further improvements that may be derived from refinements in MR imaging techniques, focusing on spatial resolution and edge sharpness. In retrospect, there is a loss of smoothness to the posterior vessel wall of the lingual arterial branch on c and d. Although the lingula branch thrombus was not detected, the left lower lobe thrombus was seen with all MR imaging techniques, and on a per-patient basis, MR imaging would have yielded a similar diagnostic clinical result to CT pulmonary angiography. 1 Yes Yes No* 2 Yes Yes Yes 3 Yes Yes Yes 4 Yes Yes Yes 5 Yes Yes Yes 6 No No Yes 7 Yes No Yes 8 Yes Yes Yes 9 No Yes Yes 10 Yes Yes Yes 11 Yes Yes Yes 12 No Yes Yes 13 Yes Yes Yes 14 Yes Yes Yes 15 Yes Yes Yes 16 Yes Yes Yes 17 Yes No Yes 18 Yes Yes Yes 19 Yes Yes Yes 20 Yes Yes Yes 21 Yes Yes Yes 22 Yes Yes Yes Note. The true FISP MR sequence was performed with triggering. * Poor respiratory triggering. Motion artifact. Discussion Our study was designed to evaluate three MR imaging methods for PE detection, with CT pulmonary angiography as the reference standard. We found that a standard bolus-timed contrast-enhanced 3D MR pulmonary angiography technique does not produce the highest detection sensitivity, as compared with the 3D GRE or the triggered true FISP technique. The combination of all three of these MR imaging techniques generated higher sensitivity for PE than any of the techniques alone. The largest report on MR imaging detection of PE to date has been the PIOPED III study (13), which was designed as a prospective multicenter clinical trial with the objective of measuring PE detection sensitivity and specificity of conventional MR pulmonary angiography. The results indicated a sensitivity that ranged between 45% and 100% at different centers and averaging 78%. The limitations of the PIOPED III study included the use of only one MR imaging technique, a large range of results among centers, and the finding of a large number of technically failed MR imaging examinations that were not factored into the statistical analysis. Our interest in comparing different techniques is partly driven by the desire to examine the possibility of providing more robust methods. The technical failure rate of MR pulmonary angiography (14%) in our study is in keeping with the range reported in PIOPED III (17); however, there was no patient in which all three sequences failed technically. The non MR pulmonary angiography sequences that we used (3D GRE and triggered true FISP) provided image acquisition factors that were complementary to standard MR pulmonary angiography and, when combined into one patient examination, improved PE detection. The 3D GRE technique has the following beneficial attributes compared with standard MR pulmonary angiography: (a) There is no need for timing the acquisition in relation to the contrast material administration; and (b) the vessel walls can be visualized, providing contrast outlining the intraluminal PE. The latter characteristic compares favorably with features of conventional MR pulmonary angiography, which provides only visualization of the lumen a mural nonocclusive thrombus may be difficult to resolve and was a factor in diminished sensitivity in a number of cases in our study. Radiology: Volume 263: Number 1 April 2012 n radiology.rsna.org 277

8 The triggered true FISP technique also has beneficial attributes compared with standard MR pulmonary angiography, but these attributes are in contrast to those enumerated for 3D GRE: (a) There is no need for bolus timing, but this is a result of not having any need for gadolinium-based contrast material administration. (b) The resistance to motion degradation allows imaging in patients who are incapable of successful breath holding. (c) There is visualization of the vessel wall, allowing circumscribed visualization of a partial embolus. Our results also suggest that the location of the pulmonary embolus may have an important effect on the sensitivity for detection of PE and that sensitivity may be reduced for the detection of emboli within the lingula in all MR imaging techniques relative to other vascular distributions. We postulate that this may be related to the vascular geometry in this area, in addition to cardiac and respiratory motion effects. One limitation of our study was that all patients were known to have PE, as diagnosed with CT pulmonary angiography. Our study design limited measurement of detection sensitivity, and applicability of results to a PE-free population was not addressed. However, we used this study design as a controlled relative measure of sensitivity and specificity between MR imaging techniques. By using only patients with PE, we were able to markedly reduce the sample size of the study. Another potential limitation in our study was that there was a delay between CT pulmonary angiography and MR imaging of 30 hours, on average. During that interval, all patients had started receiving anticoagulation therapy, with the potential that a PE may have diminished by the time of the MR imaging examination. We also chose to use a per-embolus analysis of the statistical data rather than a per-patient analysis; while this may be a source of potential bias, this method helped reduce the sample size of the study and allowed for a meaningful comparison among MR imaging techniques. While there is a correlation between emboli of the same patient, given that each MR imaging technique depends mostly on the MR imaging machine (hardware and software elements) instead of judgment from the reader, the bias due to this correlation was expected to be minimal. In summary, our results lay the foundation for future clinical studies and technical development. We found that non bolus-timed recirculationphase 3D GRE and nonenhanced true FISP sequences with cardiac and respiratory triggering provided sensitive imaging for the detection of PE compared with standard bolus-timed MR pulmonary angiography. The combination of all three MR imaging techniques further improved detection of PE, providing a potential diagnostic alternative to CT without the associated risks of exposure to ionizing radiation and iodinated contrast material. Disclosures of Potential Conflicts of Interest: B.K. No potential conflicts of interest to disclose. P.S. No potential conflicts of interest to disclose. S.T. No potential conflicts of interest to disclose. G.L.R. No potential conflicts of interest to disclose. H.D.K. Financial activities related to the present article: none to disclose. Financial activities not related to the present article: employed by Emory Healthcare. Other relationships: none to disclose. J.R.C. No potential conflicts of interest to disclose. Z.C. No potential conflicts of interest to disclose. D.R.M. Financial activities related to the present article: none to disclose. Financial activities not related to the present article: institution received grant from Siemens Medical Solutions. Other relationships: none to disclose. References 1. Morpurgo M, Schmid C, Mandelli V. Factors influencing the clinical diagnosis of pulmonary embolism: analysis of 229 postmortem cases. Int J Cardiol 1998;65(suppl 1):S79 S Hampson NB. Pulmonary embolism: difficulties in the clinical diagnosis. Semin Respir Infect 1995;10(3): Nordenholz KE, Naviaux NW, Stegelmeier K, et al. Pulmonary embolism risk assessment screening tools: the interrater reliability of their criteria. Am J Emerg Med 2007; 25(3): Laack TA, Goyal DG. Pulmonary embolism: an unsuspected killer. Emerg Med Clin North Am 2004;22(4): Stein PD, Fowler SE, Goodman LR, et al. Multidetector computed tomography for acute pulmonary embolism. N Engl J Med 2006;354(22): Mayo JR, Aldrich J, Muller NL; Fleischner Society. Radiation exposure at chest CT: a statement of the Fleischner Society. Radiology 2003;228(1): Smith-Bindman R, Lipson J, Marcus R, et al. Radiation dose associated with common computed tomography examinations and the associated lifetime attributable risk of cancer. Arch Intern Med 2009;169(22): Richman PB, Courtney DM, Friese J, et al. Prevalence and significance of nonthromboembolic findings on chest computed tomography angiography performed to rule out pulmonary embolism: a multicenter study of 1,025 emergency department patients. Acad Emerg Med 2004;11(6): Herédia V, Ramalho M, Zapparoli M, Semelka RC. Incidence of pulmonary embolism and other chest findings in younger patients using multidetector computed tomography. Acta Radiol 2010;51(4): Mitchell AM, Kline JA. Contrast nephropathy following computed tomography angiography of the chest for pulmonary embolism in the emergency department. J Thromb Haemost 2007;5(1): Ersoy H, Goldhaber SZ, Cai T, et al. Time-resolved MR angiography: a primary screening examination of patients with suspected pulmonary embolism and contraindications to administration of iodinated contrast material. AJR Am J Roentgenol 2007;188(5): Gupta A, Frazer CK, Ferguson JM, et al. Acute pulmonary embolism: diagnosis with MR angiography. Radiology 1999;210(2): Kluge A, Luboldt W, Bachmann G. Acute pulmonary embolism to the subsegmental level: diagnostic accuracy of three MRI techniques compared with 16-MDCT. AJR Am J Roentgenol 2006;187(1):W7 W Rofsky NM, Lee VS, Laub G, et al. Abdominal MR imaging with a volumetric interpolated breath-hold examination. Radiology 1999;212(3): Teigen CL, Maus TP, Sheedy PF 2nd, et al. Pulmonary embolism: diagnosis with contrast-enhanced electron-beam CT and comparison with pulmonary angiography. Radiology 1995;194(2): Viera AJ, Garrett JM. Understanding interobserver agreement: the kappa statistic. Fam Med 2005;37(5): Stein PD, Chenevert TL, Fowler SE, et al. Gadolinium-enhanced magnetic resonance angiography for pulmonary embolism: a multicenter prospective study (PIOPED III). Ann Intern Med 2010;152(7): , W142 W radiology.rsna.org n Radiology: Volume 263: Number 1 April 2012