Multislice MRI in Assessment of Myocardial Perfusion in Patients With Single-Vessel Proximal Left Anterior Descending Coronary Artery Disease Before and After Revascularization
Background Our purpose was to use multislice MRI for detection of reversible myocardial ischemia and assessment of the effect of revascularization on tissue perfusion in patients with coronary artery disease.
Methods and Results Eleven patients with single-vessel proximal left anterior descending coronary artery disease were studied with MRI and thallium scintigraphy before and 3 months after revascularization. All patients had a reversible perfusion defect by scintigraphy before treatment. With a 1.5-T MR imager, IR-prepared turboflash images were acquired in three left ventricular short-axis planes during 0.05 mmol/kg Gd-DTPA bolus at rest and with dipyridamole-induced stress. Before treatment, stress increased enhancement slope in normal (6.4±4.4 to 7.4±5.0 s−1, P<.04) and decreased it in underperfused (5.4±3.7 to 2.6±1.4 s−1, P<.02) regions, resulting in a contrast-to-noise ratio of 6.87±3.09 in underperfused myocardium. Revascularization normalized enhancement patterns of the formerly underperfused myocardium and decreased defect size both in scintigraphy (66±53° to 8±12°, P<.001) and MRI sections (49±41° to 9±8°, P<.001). Agreement of 85% in detection and correlation of 0.86 (SEE, 21°, P<.001) in sizing perfusion defects was found between MRI and scintigraphy.
Conclusions Multislice contrast-enhanced MRI can be used to detect myocardial perfusion defects in patients with coronary artery disease and in assessment of the effect of treatment on myocardial perfusion.
Coronary artery disease is the major diagnostic challenge in cardiology, and the measurement of myocardial perfusion is the goal in many cardiac imaging techniques. Coronary x-ray angiography is considered the reference standard for the assessment of anatomic abnormalities within the coronary arteries, but the physiological significance of any particular lesion requires additional techniques for measurement of end-organ perfusion. Currently, several diagnostic tests, including stress echocardiography, thallium scintigraphy, and positron emission tomography, are used for detection of abnormalities in perfusion, metabolism, or function. Application of the ultrafast first-pass approach to MRI techniques1 offers the benefit of high spatial resolution and the possibility to image any desired plane with little invasiveness and without ionizing radiation.
Clinical cardiac MRI is used primarily to obtain anatomic information. In animal studies, both reversible and irreversible regional myocardial perfusion disturbances have been detected by combined use of exogenous contrast material and fast MRI.2 3 4 5 6 7 8 9 10 11 12 In patients with coronary artery disease, most MR perfusion studies have been performed by imaging one anatomic plane repeatedly during contrast agent injection.13 14 15 Coronary vasodilators, such as dipyridamole, have been used to induce a perceptible imbalance in perfusion between normal and hypoperfused myocardium.16 In MRI studies, pharmacological stress has been shown to increase differences in delivery of the contrast agent to myocardial regions perfused by normal and diseased arteries.2 5 6 11 14 15 17 18 19 Because low-molecular-weight MR contrast agents distribute to some extent from vascular to extracellular space during the first-pass circulation,20 only fast imaging sequences can capture regional myocardial perfusion differences. To be clinically useful in localization and sizing of perfusion defects, several LV planes must be covered during the same imaging session. If only one slice is imaged, 90% of the LV is neglected. If three slices are imaged, one third of the myocardial volume is encompassed.12 18 21 Acquisition of images of the entire heart with adequate temporal resolution will likely require echoplanar imaging.22 Our goal was to determine whether multislice first-pass MRI can detect myocardial perfusion defects during pharmacological stress in patients suffering from coronary artery disease and whether the effect of revascularization could be determined by this technique.
The study protocol was approved by the local ethics committee. Eleven consecutive patients with single-vessel proximal LAD stenosis ≥75% or occlusion detected in angiography and need of revascularization based on clinical findings and angiography were included. Thallium scintigraphy and MR perfusion studies were performed within 1 week. All patients underwent either balloon angioplasty (n=7) or bypass surgery (n=4) after imaging. Three months after revascularization, both thallium scintigraphy and MR perfusion study were repeated.
Informed consent was obtained from all 11 patients before they entered the study. Mean age of the patients was 56 years (range, 40 to 68 years); 6 were men. All patients belonged to group III according to the NYHA functional classification; the mean duration of symptoms before the study had been 5 months (range, 3 to 12 months). None of the patients had a history of myocardial infarction. In addition to coronary artery disease, the patients suffered from arterial hypertension (n=2), hypercholesterolemia (n=1), mitral insufficiency (n=1), and diabetes mellitus (n=1).
Coronary angiography was performed with 6F high-flow Judkins or Amplatz catheters and filming in multiple projections. LV cineangiography was documented in the 30° right anterior oblique projection. The studies were interpreted visually by experienced angiographers (K.S.V. and P.H.). Coronary artery luminal diameter reduction ≥75% was considered significant. LV ejection fraction was calculated by the single-plane ellipsoid area-length method.23
Multislice diagrams of gated single photon emission computed tomography 201Tl myocardial perfusion studies were obtained from all patients before and 3 months after revascularization with Elscint Apex 425 ECT equipment. Before the test, cardioactive drugs were discontinued for 7 plasma half-lives, and the patients were not allowed to drink caffeinated drinks. Dipyridamole was infused at a dose of 0.56 mg/kg IV over a period of 4 minutes, and 2 mCi [201Tl]thallium chloride was injected within 2 minutes of the end of the dipyridamole infusion. The acquisition for the first projection was started 5 minutes after the 201Tl injection.
Thirty views were acquired by 180° rotation at 6° intervals for 20 minutes with a single-head low-energy all-purpose collimator (Elscint APC-3). The redistribution images were obtained 3 hours later. Routine uniformity and radius-of-rotation checks were performed. A 64×64 matrix with pixel size of 0.6×0.6 cm and 0.6-cm slice thickness was used.
Images were reconstructed by a minicomputer (Apex F1, Elscint) from the early systolic and late diastolic phases of the gated frames. Three transaxial scintigraphy levels were selected to correlate the LV short-axis MR slices by determining their distance from the apex on LV long-axis MRIs and stacks of scintigraphy sections. Circumferential profiles of myocardial thallium uptake were generated by plotting transmural mean pixel intensity against the location of each 3° radius.24 Analysis of the sections and profiles was performed without knowledge of the presence or absence of perfusion defect on MR images. The profiles were normalized to the maximum pixels, and intensity <75% of the maximum was classified as reduced perfusion.25 The size of the perfusion defect was assessed for each section as degrees of the 360° circle.
MR perfusion imaging study was performed on all patients before and 3 months after revascularization. Cardioactive drugs and caffeinated drinks were discontinued as before scintigraphy. An 18-gauge catheter was inserted into the antecubital vein for injection of dipyridamole and contrast agent. Extension tubing was loaded with contrast agent to permit injections outside the bore of the magnet. Patients were positioned supine in the body coil of a 1.5-T Siemens Magnetom Vision imager. Imaging was performed with the body coil used as receiver. Coronal and LV long-axis scout images were obtained to determine the final imaging plane to the LV short axis.
To monitor the first transit of bolus-injected contrast agent, ECG-gated IR-prepared turboflash images were acquired on three LV short-axis planes with the following parameters: TR/TE, 3.3/1.4 ms; matrix, 128×128; field of view, 350 mm; and flip angle, 8°. An inversion time of 400 ms was chosen to provide null myocardial signal before contrast administration. Perfused with Gd-DTPA, myocardium would therefore show relatively increased SI because of the predominant T1 shortening.
LV short-axis orientation was selected for MRI to minimize partial volume effect, to allow comparison between myocardial regions perfused by different arteries and between MR and thallium scintigraphy sections. Images were acquired from three slices (1 cm thick, 0.5 cm apart); the basal slice was set at the level of the papillary muscles. Sets of 60 images were acquired, with the three slices imaged repeatedly. The set of three slices was obtained every four to six RR intervals (≈4 seconds), and one IR preparation pulse was followed by an imaging sequence for a single slice.
After the LV short-axis planes were localized, the transit of contrast agent was monitored at rest. During IR-prepared turboflash imaging, after the third set of the three slices, Gd-DTPA (0.05 mmol/kg) was injected intravenously at a speed of 5 mL/s. After the transit study, the washout of the contrast agent from the myocardium was monitored by repeated sets of IR-turboflash images every 5 minutes. As myocardial intensity came close to zero at 20 to 30 minutes after the contrast injection, the pharmacological stress was induced. The patient bed was withdrawn from the bore of the magnet, and care was taken not to move the patient. Dipyridamole (0.56 mg/kg) was infused over a period of 4 minutes during continuous monitoring of one ECG lead and blood pressure. Three minutes after the end of dipyridamole infusion, the patient was placed in the original imaging position, and a set of IR-prepared turboflash images was repeated during the injection of the second bolus of 0.05 mmol/kg Gd-DTPA. If chest pain appeared, the patient was treated with intravenous aminophylline immediately after imaging.
The effect of Gd-DTPA injection was quantified by measurement of SI changes from right ventricular and LV chamber blood, myocardial areas representing the perfusion beds of the three coronary arteries, subcutaneous fat, and the SD of the background noise.10 The ROI for the perfusion bed of the LAD was located at the anteroseptal, that for the LCx at the lateral, and that for the RCA at the posterior myocardial wall18 19 (Fig 1⇓). The ROI for the subcutaneous fat was placed at the anterior thoracic wall close to the heart, and the SD of the background noise was obtained from the ROI drawn outside the body along the phase-encoding direction. All three slice positions of both rest and stress image sets were analyzed with the three myocardial ROIs. Regional SI values (arbitrary units) were measured with the image analysis program NIH Image 1.59. Owing to the displacement of the heart during breathing, placement of ROIs was individually traced from image to image. ROI size and shape were kept constant throughout the analysis of each section and myocardial area. Values from the LAD ROI represented underperfused myocardium, and the mean of the LCx and RCA ROIs, normal myocardium. SI-time curves were generated for each section and used for further analysis.
Signal-to-noise ratio was calculated from the SI of normal myocardial region divided by the SD of background noise. The CNR of the underperfused myocardium was calculated from the formula CNR=(SI of normally perfused region−SI of LAD region)/SD of background noise. The SI change rate of normally perfused myocardium and the perfusion bed of diseased artery was assessed from the upslope of SI-time curves of each perfusion study and calculated as described by Matheijssen et al19 from the formula SI change rate=SI increase/time (ms). The ratio of SI change rate of underperfused to normal myocardium was calculated from these values. Heart rate during each imaging set was taken into account in analysis of the SI change rate and in the comparison of results between rest and stress studies.
The sizes of perfusion defects were assessed from stress images with the largest myocardial CNR. A circumferential ROI of three pixel thickness (7 mm) was drawn on the myocardium; care was taken to avoid pixels in the blood pool or epicardial fat. The SI profile was produced by plotting the mean SI of the three pixels against the angular position to correlate the profiles generated from the scintigraphy images. The mean and SD of SI of the normal myocardium in the posterolateral myocardial region were calculated for each section, and the pixels with SIs of 2 SD below the mean were classified to represent the areas with reduced perfusion. The size of underperfused region on each section was presented as degrees of the circumferential profile. Anterior and septal LV segments on scintigraphy sections25 were considered to correspond to the LAD ROI positioned on the anterior interventricular sulcus on MR images (Fig 1⇑). The sections were compared to determine the presence or absence and size of perfusion deficit. CNR was also calculated for residual deficits detected either on scintigraphy or MR sections after revascularization by use of values obtained from MR circumferential profiles.
All values were expressed as mean±SD. The significance of differences in SI-time curves was determined by repeated-measures ANOVA with multiple comparison of mean values by Scheffé’s F test. The paired two-tailed Student’s t test was used to determine the differences in CNR and SI change rate at rest and during stress before and after revascularization and perfusion defect sizes that were obtained from scintigraphy and MR images. Also, the paired two-tailed Student’s t test was used to compare baseline heart rate and mean blood pressure with the values obtained after dipyridamole injection and between scintigraphy and MRI. The unpaired two-tailed Student’s t test was used to compare patient groups with complete coronary artery occlusion and stenosis. The mean difference in perfusion defect sizes was obtained by subtracting defect size on MR sections from scintigraphy sections and calculating the mean of the differences, as suggested by Bland and Altman.26 Linear regression analysis was used to compare the perfusion defect sizes on MR and scintigraphy sections and to compare the myocardial SI change rates with the relative myocardial intensity on scintigraphy. A significance level of P<.05 was used.
Coronary Angiography and Thallium Scintigraphy
At coronary angiography, four patients had complete occlusion of the LAD. Seven patients had LAD stenosis with diameter reduction of 90±8% (range, 80% to 99%). All patients had normal LV function in cineangiography, and ejection fraction was 65±6% (range, 55% to 73%). Before treatment, at scintigraphy, all patients had reversible perfusion defects anteroseptally on at least one section. The size of perfusion defects was 66±53° (range, 0° to 166°) of the circumferential profile of each section.
All patients underwent revascularization in 2 to 8 weeks after coronary angiography. Four patients with a complete LAD occlusion were treated surgically with the native left internal thoracic artery used as bypass graft. In the rest, balloon angioplasty resulted in residual stenosis of 19±11% (range, 0% to 30%). In control scintigraphy, three months after revascularization, the defect sizes had decreased to 8±12° (range, 0° to 39°), P<.001 compared with the defect sizes before treatment. There were no resting deficits on scintigraphy sections before or after treatment.
Before revascularization, at rest, contrast injection was followed by rapid SI increase in chamber blood and slower enhancement of myocardium (Fig 2a⇓). Myocardial signal-to-noise ratio was 6±3 before contrast enhancement, 18±4 at the peak of the bolus, and 16±3 at the end of the image set. No statistical difference in SI curves representing areas perfused by diseased and normal arteries was detected (Fig 2a⇓). However, an increase in CNR and a difference in SI change rate between the myocardial regions was observed (Tables 1⇓ and 2⇓).
During pharmacologically induced stress, before contrast enhancement, there was no difference in the myocardial SI between the areas perfused by the diseased and normal arteries (Figs 2b⇑ and 3⇓). A significant difference in the myocardial regional SI lasted over seven images (≈28 seconds), starting from the second image of the upslope of the myocardial enhancement curve (Fig 2b⇑). CNR increased after enhancement and declined toward the end of the image set. The peak CNR was greater in image sets acquired at stress than at rest (Table 1⇑). The ratio of the SI change rates measured in the myocardial areas perfused by diseased to normal arteries was lower in the images acquired at stress than at rest (Table 2⇑). There was no statistical difference in CNR or SI change rate values in patients with complete occlusion compared with patients with coronary artery stenosis either at rest or at stress.
The effect of revascularization on myocardial MR perfusion images is illustrated in Fig 4⇓. After revascularization, the SI alterations, CNR, and SI change rates were not different in image sets acquired at rest and at stress. All these values were comparable to the values obtained from the images acquired at rest before revascularization (Fig 2c⇑, Tables 1⇑ and 2⇑). CNR was higher at the peak of the contrast compared with the end of the image set acquired at stress, but both values were within the range of noise level.
Before treatment, perfusion defect size on MRI sections was 49±41° (range, 6° to 144°) of the circumferential profile. Revascularization decreased the size to 9±8° (range, 0° to 22°), P<.001. No resting deficits were observed on MRI sections before or after treatment. The agreement on the presence or absence of perfusion deficits between scintigraphy and MRI sections was 85% (n=66 sections), and that between scintigraphy and MRI sessions, 95% (n=22 cases, 11 patients imaged before and after treatment). After treatment, three perfusion defects were detected only on scintigraphy and eight only on MRI sections. The three deficits seen only on scintigraphy had a size range of 6° to 39°, thallium count 61% to 74% of maximum, and CNR on MR images of 1.20 to 1.96. The eight deficits seen only on MR images had a size range of 4° to 22°, thallium count 76% to 85% of maximum, and CNR in the center of the deficit 2.46 to 5.97. In the rest deficit, size was 5° to 22° on MR and 3° to 39° on scintigraphy, thallium count 68% to 74% of maximum, and CNR in the center of the deficit 3.05 to 5.70. Correlation coefficient between the perfusion defect sizes measured with scintigraphy and MRI was.86, and SEE was 21.3°, P<.001 (Fig 5⇓). Defect sizes were 7±23° smaller on MR than on scintigraphy sections, P=.008.
CNR and SI change rates of the LAD ROIs were not different in sections with complete recovery compared with the sections with remaining small regional defects. The CNR, SI change rates, and size of residual perfusion defects were not statistically different between the patients who had been treated with balloon angioplasty and bypass surgery. Relative regional myocardial intensity on scintigraphy sections and SI increase rate on the corresponding regions on MRI had a correlation of r=.66 (Fig 6⇓).
Dipyridamole infusion increased heart rate from 56±7 to 71±10 bpm, and no significant change in mean blood pressure (105±14 mm Hg) was observed. There was no statistical difference in cardiovascular responses between scintigraphy and MRI. With slower heart rates, each MR slice was imaged every fourth heartbeat, and with faster heart rates (during pharmacological stress), every sixth heart beat, averaging into mean temporal resolution of 4.3±0.6 seconds (range, 3.4 to 6.0 seconds). There was no significant difference between rest (4.3±0.5 seconds) and stress (4.6±0.6 seconds) studies in temporal resolution, P=.06.
The purpose of this study was to use multislice pharmacologically stressed contrast enhanced IR-turboflash imaging (1) for detection and sizing of reversible myocardial ischemia in patients with proximal LAD disease and (2) in the assessment of the effect of revascularization on myocardial perfusion. Three slices were imaged repeatedly to cover a larger part of the myocardium than in single-slice studies. A body coil was used as receiver to allow the comparison of regional enhancement during the bolus injection of Gd-DTPA. Pharmacological stress was applied to increase the perfusion gradient between the myocardial areas. Data obtained from stress and rest studies were compared. The responses to bypass surgery and angioplasty were assessed and compared 3 months after revascularization.
Dipyridamole infusion was followed by increased CNR of myocardial perfusion defect and decreased ratio in SI change rates of underperfused to normal myocardial regions. The differential signal enhancement of normal and underperfused myocardial regions lasted >28 seconds, which allowed 85% agreement between scintigraphy and MRI on detection of the perfusion defects on all three myocardial sections. Relative regional myocardial intensity on scintigraphy sections correlated with SI increase rate on MRI. Revascularization decreased the size of the regional perfusion defects on scintigraphy and MRI sections. After revascularization, the enhancement patterns were identical in normal regions, in regions treated with angioplasty, and in regions treated with bypass surgery.
In conclusion, a close correlation was found between dipyridamole thallium scintigraphy and contrast-enhanced multislice MRI in detection and sizing of the regional myocardial perfusion defects. We suggest that multislice MR perfusion stress testing could be used to assess the effect of revascularization on regional myocardial perfusion.
In multislice myocardial perfusion imaging, good spatial resolution has been achieved with a surface coil as receiver, but uneven MRI signal has prevented quantitative analysis between anatomically different myocardial regions.18 We obtained images with the body coil as receiver, which allowed comparison both between the myocardial regions on the same MRI section and between the numerical results attained at rest and stress before and after revascularization. Also, sizing of perfusion defects was feasible from MRI sections. We selected a T1-weighted imaging sequence for the detection and sizing of myocardial perfusion defects, because myocardial signal-to-noise ratio on T2*-weighted images is low and the effect of bolus on myocardial signal is shorter.10 Multislice first-pass transit IR-prepared imaging has been able to locate acute myocardial ischemia in canine heart in vivo,12 and recent reports have outlined the feasibility of clinical MR imagers in multislice dynamic imaging of the human heart.17 18 21
The paramagnetic compound Gd-DTPA shortens T1 relaxation time, leading to increase in SI on T1-weighted MR images. Although in the first pass, 30% of the extracellular contrast agent enters the interstitium,20 27 myocardial Gd-DTPA concentration is determined primarily by coronary flow within 5 minutes of a bolus injection.28 Dipyridamole induces vasodilatation and flow in a normal more than in a diseased coronary artery,29 which generates contrast between normal and underperfused myocardial regions. After peripheral bolus injection, calculation of myocardial flow from the mean transit time curves is not feasible,30 but linear fit of SI increase is a sensitive parameter in discriminating underperfused from normally perfused myocardial regions.19 Consequently, in our study, the SI differences between the myocardial regions perfused by normal and diseased arteries were related to but did not exactly measure myocardial perfusion at the time of the bolus injection.
Before treatment, at rest, during peak enhancement of myocardium, we found an increase in contrast between regions perfused by diseased and normal coronary arteries and greater SI change rate in normal than in underperfused territory. In analysis of individual curves of each patient, baseline CNR was calculated from images with zero myocardial intensity (as a result of inversion time 400 ms). Peak CNR was derived from images with the largest contrast between the myocardial areas. In analysis of the mean curves of all patients, no statistical difference was found, possibly because the peak CNR appeared at divergent time points after the beginning of the enhancement and lasted only for two or three images. We suggest that the low peak CNR value indicates a small perfusion gradient between the myocardial areas at rest. Our finding is in agreement with the results of Manning et al,13 who used a single-slice IR-turboflash sequence and observed myocardial enhancement differences at rest in coronary artery disease patients. In multislice MR perfusion studies, comparison between regions has been prevented by either the use of surface coil17 18 or poor temporal resolution.17 In dogs, coronary artery occlusion but not stenosis has produced contrast between myocardial regions during bolus injection at rest.10 11 We suggest that in our patients, gradual development of coronary artery occlusion has induced collateral circulation to the underperfused area, which prevented differentiation between occlusion and stenosis.
Before treatment, during pharmacologically induced stress and peak enhancement of myocardium, we observed a significant increase in myocardial CNR. After dipyridamole infusion, the SI change rate of normal myocardium increased and that of underperfused region decreased (Table 2⇑), indicating a coronary steal phenomenon. This finding is in agreement with other reports on stress-test first-pass MRI in animal models2 5 6 11 and in coronary artery patients.14 15 17 18 21 The best CNR was detected during the upslope and peak of enhancement of normally perfused myocardium, which implies that with careful timing, the images with best contrast could be acquired during a single breathhold of 30 seconds.
After revascularization, contrast enhancement patterns in the myocardial regions were indistinguishable both at rest and at stress. Manning et al13 discovered a significant increase in SI peak but no change in SI increase rate in the revascularized myocardial segments at rest after bypass surgery. In our study, CNR and SI change rates did not allow any differentiation of the formerly underperfused and normal myocardial regions. Enhancement properties in the revascularized myocardium were not dependent on the method of the therapy.
We observed linear correlation in defect sizing between MRI and thallium scintigraphy. Discrepancy with a canine study, with histochemical morphometry for comparison, could be explained by homogeneous defect size induced by artificial coronary stenosis in dogs.11 Perfusion defects were 7° smaller on MRI than on scintigraphy sections, with considerable differences (SD, 23°) compared with scintigraphy values. Although treatment decreased perfusion defect sizes in our patients, small residual defects were identified, of which three were seen only on scintigraphy and eight only on MR sections. As suggested by the normal enhancement curves obtained from the myocardial LAD ROIs drawn on the revascularized regions, low-intensity areas with size <10° could represent the lowest 2.5% of the normal myocardial intensity variation. Second, discrepancy between the two imaging methods in detection of small and sizing of large perfusion defects could rely on different data collection procedures. In MRI, the defects were seen during the 20-second upslope of myocardial enhancement, which in scintigraphy images were obtained during 20 minutes. Third, myocardial enhancement during first pass on MRI is a result of inflow of the enhancing agent from vasculature to myocardial extracellular space28 and water exchange rate in the tissue.31 In comparison, myocardial 201Tl content is controlled by coronary blood flow and the ability of myocytes to extract 201Tl from the blood (Na,K-ATPase pump).32 We suggest that the relatively poor correlation between bolus tracking MRI and thallium scintigraphy in evaluation of myocardial blood flow could be attributed to the differences in uptake and distribution of the tracers.
Our data were based on a strictly limited patient population with a homogeneous disease pattern. All patients had LAD disease, which prevented blinded comparison of deficit localization between the imaging methods. No patients with less severe coronary artery disease, multiple stenoses, or previous myocardial infarction were included. Our goal, however, was to assess the feasibility of multislice MRI stress test on evaluation of reperfusion therapy. More extensive studies on larger patient populations with variable myocardial ischemic disease are needed before MRI can be used clinically for assessment of myocardial ischemia and reperfusion.
In conclusion, pharmacologically stressed multislice MRI has a potential in detection of reversible myocardial perfusion defects in patients with coronary artery disease and in assessment of the effect of revascularization on regional perfusion.
Selected Abbreviations and Acronyms
|LAD||=||left anterior descending coronary artery|
|LCx||=||left circumflex coronary artery|
|RCA||=||right coronary artery|
|ROI||=||region of interest|
This study was supported by grants from the Foundations of Ida Montin and P.O. Klingendahl, the Radiological Society of Finland, the Finnish Medical Society Duodecim, the Helsinki University Central Hospital Research Funds, the Paavo Nurmi Foundation, and the Juselius Foundation.
- Received April 8, 1997.
- Revision received June 11, 1997.
- Accepted June 14, 1997.
- Copyright © 1997 by American Heart Association
Arteaga C, Canet E, Ovize M, Janier M, Revel D. Myocardial perfusion assessed by subsecond magnetic resonance imaging with a paramagnetic macromolecular contrast agent. Invest Radiol. 1994;29:S54-S57.
Saeed M, Wendland M, Yu K, Lauerma K, Li H, Derugin N, Cavagna F, Higgins CB. Identification of myocardial reperfusion with echoplanar MR imaging: discrimination between the occlusive and reperfused infarctions. Circulation. 1994;90:1492-1501.
Matheijssen NA, Louwerenburg H, van Rugge F, Arens R, Kauer B, de Roos A, van der Wall EE. Comparison of ultrafast dipyridamole magnetic resonance imaging with dipyridamole sestaMIBI SPECT for detection of perfusion abnormalities in patients with one-vessel coronary artery disease: assessment with quantitative model fitting. Magn Res Med. 1996;35:221-228.
Tweedle MF. Physicochemical properties of gadoteridol and other magnetic resonance contrast agents. Invest Radiol. 1992;27:S2-S6.
Burow RD, Pond M, Schafer A, Becker L. ‘Circumferential profiles’: a new method for computer analysis of thallium-201 myocardial perfusion images. J Nucl Med. 1979;20:771-777.
Segall GM, Atwood E, Botvinick E, Dae M, Lucas J. Variability of normal coronary anatomy: implications for the interpretation of thallium-SPECT myocardial perfusion images in single-vessel disease. J Nucl Med. 1995;36:944-951.
Canty JM, Judd R, Brody A, Klocke F. First-pass entry of nonionic contrast agent into the myocardial extravascular space. Circulation. 1991;84:2071-2078.
Gehring PJ, Hammond PB. The interrelationship between thallium and potassium in animals. J Pharmacol Exp Ther. 1967;155:187-201.