| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Department of Diagnostic Radiology, Eberhard-Karls-University,
Hoppe-Seyler-Str. 3, Tuebingen 72076, Germany.
2 Department of Internal Medicine, Division of Cardiology,
Eberhard-Karls-University, Tuebingen 72076, Germany.
Received March 25, 2004;
accepted after revision October 26, 2004.
Address correspondence to M. Fenchel.
Abstract
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
SUBJECTS AND METHODS. Rest and stress perfusion MRI studies were performed in 22 patients with coronary artery disease at 1.5 T using a multislice saturation recovery true FISP sequence after the bolus injection of 0.025 mmol/kg of body weight of gadopentetate dimeglumine. The myocardium of each slice was divided into 12 radial segments with subdivision into subendocardial and subepicardial subregions. Myocardial perfusion was assessed semiquantitatively and independently for each subregion. The standard of reference for myocardial perfusion was SPECT. Delayed enhancement images were acquired after the injection of 0.15 mmol/kg of body weight of gadopentetate dimeglumine.
RESULTS. Sensitivity and specificity of perfusion MRI examinations for the detection of perfusion deficits were 81% and 89%, respectively, for the semiquantitative perfusion parameter upslope and 78% and 86% for the parameter peak signal intensity. More specifically, rest perfusion examinations were able to detect areas of infarction, whereas stress examinations increased the perfusion differences between normal and ischemic myocardial areas. Excellent correlation was observed between rest perfusion and late enhancement findings (r = 0.90).
CONCLUSION. In patients with single-vessel coronary artery disease, perfusion deficits can reliably be detected using a saturation recovery true FISP sequence. Semiquantitative perfusion parameters upslope and peak signal intensity yielded similar results.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
After the injection of contrast material, the identification of regional hypoperfusion is based on the detection of a relative reduction of signal intensity (SI) increase in the bed of the stenosed coronary artery during coronary vasodilation [5, 8-10]. This is accomplished by tracking a contrast agent bolus during the first pass through the heart before and after pharmacologically induced coronary vasodilation.
Several previous studies, using spoiled gradient echo pulse sequences, found good correlation of regional perfusion deficits detected by MRI compared with SPECT and invasive coronary angiography [8-11].
Although perfusion MRI continues to develop at a rapid pace, it still suffers from low signal-to-noise ratios [12-14]. Besides preparation pulses for stronger T1 weighting [14], alternative pulse sequences, such as fast imaging in steady-state precession (FISP), have the potential to significantly improve the signal-to-noise ratio of perfusion images [15-19].
|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Patients with unstable angina, diabetes mellitus, acute myocardial infarction (< 10 days), atrial fibrillation, or contraindications to dipyridamole (relevant obstructive pulmonary disease, asthma, sinoatrial or atrioventricular block > 1) or MRI examination were excluded.
Written informed consent was obtained from all patients before the MRI examination. The study protocol was approved by the local ethics committee.
Perfusion Sequence
The saturation recovery true FISP 2D perfusion sequence was prospectively
ECG-triggered. The sequence consisted of a non-slice selective 90°
saturation pulse and single-shot true FISP image acquisition with TR/TE,
2.2/1.1; inversion time, 110 msec; flip angle, 45-50° (depending on
specific absorption rate limits); bandwidth, 1,300 Hz per pixel
(Fig. 1). The matrix was 72
x 128; rectangular field of view, 225 x 300 mm; and trigger delay,
100 msec. Voxel size was 3.1 x 2.3 x 8.0 mm. Three representative
short-axis slices were acquired (left ventricular base, midventricular, and
apical). One image of each slice was obtained at every heartbeat.
MRI Protocol
Examinations were performed with a 1.5-T MRI system (Magnetom Sonata,
Siemens Medical Solutions) equipped with high-performance gradients (maximum
amplitude, 40 mT/m; slew rate, 200 mT/m/msec). An 18-gauge catheter was
inserted into an antecubital vein for injection of the contrast agent. MRI ECG
leads were placed on the patient's chest. Imaging was performed with a
phased-array surface coil as receiver.
Left ventricular short-axis orientation was used for perfusion imaging (to minimize partial volume effects). For myocardial perfusion measurement, three representative short-axis sections were placed in basal, midventricular, and apical regions of the left ventricle. In some patients, because of a fast heart rate, only two short-axis sections could be measured (midventricular and apical regions of the left ventricle). Rest and stress perfusion scanning acquired a series of 40 perfusion images for each slice. The patients were instructed to hold their breath as long as possible and to breathe shallowly thereafter. The perfusion images had the same orientation and position before and after the administration of dipyridamole. The acquisition of perfusion images was started simultaneously with the injection of gadopentetate dimeglumine (0.025 mmol/kg of body weight of Magnevist [Schering]) and the patient's breath-hold. The bolus injection of gadopentetate dimeglumine (flow rate, 5 mL/sec) was followed by a 20-mL flush of 0.9% NaCl (flow rate, 5 mL/sec). A delay of 15 min after the rest examination allowed residual gadopentetate dimeglumine to be washed out of the myocardium.
Pharmacologic stress was applied using dipyridamole according to a standardized protocol (0.56 mg of dipyridamole/kg of body weight over 4 min). Two minutes after the administration of dipyridamole, the perfusion MRI stress study was performed with identical parameters as before. Dipyridamole leads to reflex tachycardia. An increase of 20% or more in the patient's heart rate served as control for a sufficient pharmacologic stimulation of perfusion reserve.
Immediately after the stress perfusion scanning, 0.1 mmol of gadopentetate dimeglumine/kg of body weight was administered at 1 mL/sec, which, with contrast material applied for perfusion studies, totals 0.15 mmol of gadopentetate dimeglumine/kg of body weight. To visualize infarcted areas, delayed enhancement images were acquired after 15 min in several short-axis slices using an inversion recovery turboFLASH sequence. Sequence parameters were as follows: field of view, 300-340 mm; TR/TE, 9.56/4.38; inversion time, 200-260 msec; flip angle, 25°; matrix, 166 x 256; slice thickness, 5 mm.
Technetium-99m Tetrofosmin SPECT
SPECT examinations were performed within 4 weeks of MRI. No treatment
(e.g., balloon angioplasty) was done between the examinations. A standardized
1-day protocol was used. Pharmacologic stress was applied by injection of 0.56
mg/kg of body weight of dipyridamole over 4 min. A bolus of 800 MBq of
99mTc-tetrofosmin was injected in an antecubital vein and flushed
with 10 mL of 0.9% NaCl. SPECT image acquisition was performed with the
patients in the supine position using a double-headed rotating gamma camera
(Millennium VG, GE Healthcare) with a high-resolution collimator. All images
were reconstructed in short-axis, vertical, and horizontal long-axis views.
For data evaluation, short-axis sections corresponding to perfusion MRI slices
were used.
MR Data Analysis
A Matlab-based software (The MathWorks) was used for semiquantitative
analysis of myocardial perfusion. The myocardium of short-axis cross sections
was subdivided into 12 radial segments. Each transmural segment consisted of
approximately 40-50 pixels. The resulting transmural segments were further
subdivided into subendocardial and subepicardial subregions (
20-25
pixels), and for each subregion SI-time curves were calculated separately.
Baseline SI measured before contrast administration was used to correct
coil-induced inhomogeneous signal yield and potential differences in SI in
individual segments (SInorm = [SIcontrast -
SIbaseline]/SIbaseline). Time-intensity profiles were
calculated for each subregion in the left ventricular myocardium using the
normalized data points. A curve fit was performed using a cubic polynomial
function. The fitted curves were used to derive semiquantitative perfusion
parameters (maximum upslope and peak SI), which correlate with regional
myocardial perfusion as previously described
[8,
9,
20]. Segments with less than
mean - 1 times SD of all segments in the slice were considered to represent a
perfusion defect. A transmural segment was considered hypoperfused if the
subendocardial subregion alone or the subendo- and subepicardial subregions
were hypoperfused.
Perfusion MRI Score (PMRS)
A perfusion score was calculated for perfusion MRI parameters. A score of
n is assigned to segment k of each slice if
 |
where p(k) is the perfusion value for parameter upslope
or peak SI for segment k; p is the mean of all perfusion
values for one slice; 
is the SD of all perfusion values for one slice;
n = 1,2,3; m = number of segments per slice; and k
=1...m.
Calculations were performed for the perfusion parameters upslope and peak SI. Subsequently, MR segments were assigned to the "SPECT hypo" group if they were correlating with a perfusion defect in the SPECT images or to the "SPECT norm" group if the perfusion MRI defect was located in a remote area of the SPECT images. PMRS of individual segments were summed for each group (PMRSabs).
The relative PMRS for each group was calculated by dividing the absolute
PMRS by the number of segments:
 |
where Ngroup represents the number of segments in each group. These values represent the severity of perfusion defects for each slice and patient and the location relative to hypoperfused areas in SPECT images.
The perfusion score was calculated for each transmural segment from the hypoperfused subregions—that is, in the case of subendocardial perfusion deficits, from the subendocardial subregion alone; in the case of transmural perfusion deficits, from the mean of subendo- and subepicardial subregions.
Comparison of SPECT and MRI
The short-axis SPECT images were subdivided into the same number of
transmural segments corresponding to short-axis MR images. Each segment was
rated normally perfused (SPECT norm) or hypoperfused (SPECT hypo) using a
2-point scale in a consensus interpretation of two experienced observers.
Because rest examinations were performed after stress examinations in the
SPECT examination protocol, "SPECT hypo" in rest studies
represents fixed perfusion deficits, whereas "SPECT hypo" in
stress studies represents fixed and reversible perfusion deficits.
Corresponding segments were compared, using SPECT as standard of
reference.
Statistical Analysis
If not stated otherwise, data were given as mean ± 1 SD. A
p value of 0.05 or less was regarded as statistically
significant.
Sensitivity and specificity of rest and stress examinations, including 95% confidence intervals, were calculated for the perfusion parameters upslope and peak SI from the 2x2 tables comparing MRI and SPECT data. Overall accuracy was determined for both parameters.
Statistical difference of PMRS between rest and stress was assessed using a paired Student's t test. Differences of PMRS between normal and hypoperfused and between infarcted and noninfarcted regions were assessed using an unpaired Student's t test. These calculations were performed using the statistics software JMP, version 4 (SAS Institute). The number of segments manifesting hypoperfusion in rest examinations was compared with the number of segments showing late enhancement using Pearson's correlation coefficient.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Two-by-two tables are displayed providing numbers of hypoperfused and normally perfused segments on SPECT and MRI (Table 1). Sensitivities and specificities and overall accuracy for perfusion defects in MR examinations are summarized in Table 2 for the semiquantitative perfusion parameters at rest and after stress. The standard of reference is SPECT data from the same slice location. Sensitivity and specificity increased for perfusion parameter upslope (peak SI) from 63% (59%) to 81% (78%) and from 79% (81%) to 89% (86%), respectively, after administration of pharmacologic stress. Examples of rest and stress perfusion series are given in Figs. 2A, 2B, 2C, 2D, 3A, 3B, and 3C.
|
|
|
|
|
|
|
|
|
Numbers in the following paragraphs represent absolute PMRS/relative PMRS for the perfusion parameter upslope (peak SI). A total of 696 segments were evaluated. Segments were classified into hypoperfused (SPECT hypo) and normally perfused (SPECT norm) areas, according to SPECT findings. PMRS was 81/1.08 (73/0.97) in 75 hypoperfused segments and 408/0.66 (362/0.58) in normal perfused segments in SPECT images at rest, respectively. The latter represent potentially false-positive perfusion deficits detected with MRI relative to SPECT.
After the application of pharmacologic stress, PMRS increased significantly to 420/2.11 (338/1.70) (p < 0.05), whereas in normal perfused regions derived from SPECT stress images, PMRS decreased in stress MRI examinations to 281/0.56 (266/0.54) (Tables 3 and 4).
|
|
More specifically, the division of patients into two distinct subgroups with infarction and without infarction yielded an absolute/relative PMRS for upslope (peak SI) of 73/1.07 (68/1.00) in 68 hypoperfused segments and 200/0.63 (220/0.70) in normal perfused segments; PMRS increased significantly to 269/2.22 (165/1.36) (p < 0.05) in hypoperfused areas in the subgroup with infarction. In the subgroup without infarction, PMRS of 8/1.14 (11/1.57) for hypoperfused and 208/0.68 (142/0.47) for normal perfused regions were observed at rest. In stress examinations, PMRS increased to 151/1.94 (173/2.21).
Interestingly, a significantly reduced absolute PMRS and fewer hypoperfused segments were seen in rest examinations of patients without prior myocardial infarction.
Late enhancement was present in all patients with a history of myocardial infarction. In 12 patients, a total of 99 segments displayed late enhancement, whereas 88 segments were hypoperfused in rest perfusion examinations. The location with respect to the vascular territory was identical in all patients for resting hypoperfusion and late enhancement. Correlation of the number of segments with late enhancement and perfusion deficits at rest was excellent (r = 0.90, Pearson's correlation coefficient).
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Previous studies reported sensitivities of 64-91% and specificities of 76-94% relative to SPECT or conventional coronary angiography [1, 8, 11, 21-24]. Some of those studies used other sequence technologies based on turboFLASH or echo-planar imaging, or exhibited a decreased number of slices or temporal resolution. Our study yielded a sensitivity and specificity for detection of perfusion deficits of up to 81% and 89%, respectively, in stress perfusion examinations, which compares favorably with values reported in the literature.
Saturation recovery true FISP perfusion sequences have been shown to provide higher signal-to-noise ratio and contrast-to-noise ratio values than saturation recovery turboFLASH perfusion sequences in volunteer [16] and patient studies [18, 19]. Although patient studies were conducted with a limited number of subjects, the authors concluded that true FISP perfusion imaging may improve the visibility of perfusion deficits and therefore sensitivity for the detection of significant epicardial vessel stenosis.
A saturation recovery sequence was used in this study, although image quality of saturation recovery sequences is inferior to inversion recovery approaches because inversion recovery sequences yield good contrast between areas with different myocardial T1. However, the time between successive inversion pulses should allow longitudinal magnetization to recover almost completely. Therefore, multislice approaches with inversion pulse preparation most likely suffer from poor temporal resolution [7, 8, 10, 15, 25] because inversion pulse preparation and image data readout may last up to 650 msec. Preparation and readout of the saturation recovery true FISP sequence used in our study required 190 msec, which allows acquisition of multiple slices in a single heartbeat and reduces motion-induced blurring.
Likewise, SIs of inversion recovery sequences are dependent on the time interval between two successive data acquisitions, which can be troublesome in patients with only slight arrhythmia [23]. In addition, when using inversion recovery sequences, problems arise in maintaining steady-state conditions if the time between trigger pulses varies.
Absolute PMRS is the sum of all evaluated segments in one group (SPECT hypo or SPECT norm). Because there are fewer segments in the SPECT hypo group, especially in rest examinations, absolute PMRS is lower in this group. However, relative PMRS provides the mean perfusion score for the segments in each group that show decreased perfusion (i.e., higher relative PMRS values) in the SPECT hypo group.
In our study, PMRS increases significantly for the parameter upslope (peak SI) after the application of pharmacologic stress by 43% (37%). PMRS increased especially in areas with hypoperfusion in SPECT. Specifically, absolute and relative PMRS increased by 333% and 64%, respectively, (363% and 75%) for the parameter upslope (peak SI), reflecting a markedly decreased perfusion in MR stress examinations in areas that were rated hypoperfused in SPECT. This observation is in accordance with those of previous studies that reported a reduced perfusion reserve in the area of a stenosed coronary artery [5, 6, 9].
Interestingly, a decrease of absolute and relative PMRS occurs in stress examinations in normal perfused regions by 28% and 11%, respectively (27% and 7%) for the parameter upslope (peak SI). Although the decrease of PMRS in normally perfused regions in stress examinations is not statistically significant, it shows a trend of reduction of false-positive areas in stress examinations.
The PMRS, serving as an indicator of hypoperfusion, is increased in hypoperfused regions at rest compared with normal perfused regions, which is probably caused by regions of myocardial infarction. Twelve patients showed the presence of late enhancement in the area of the stenosed coronary artery. In addition, 8 of 12 patients with late enhancement in MR images displayed a fixed defect in SPECT images in the same myocardial region. In the regions with a fixed defect in SPECT, the relative PMRS was increased for the parameters upslope and peak SI (1.07 and 1.00) compared with normal regions in SPECT (0.63 and 0.70). The same relative PMRS, however, is also observed in resting scintigraphic perfusion defects in patients with no prior history of infarction. From our data, it is unclear whether SPECT and MR detected scar tissue in patients with an unapparent infarction or whether these are real false-positive results. Concerning the difference of absolute PMRS and the number of segments with resting perfusion defects in SPECT in patients with and without a prior history of infarction, there is evidence that, as previously shown, perfusion MRI at rest as well as a fixed perfusion defect in SPECT can detect regions with chronic infarction [26]. However, to reliably distinguish infarction from resting perfusion abnormalities on MRI, additional information, such as the presence of delayed enhancement, should be used [7, 23, 27].
Comparison of the relative PMRS in hypoperfused and normal perfused regions for the parameter upslope (peak SI) yields a ratio of 1.64 (1.67) for rest examinations and 3.77 (3.15) for stress examinations (PMRSHYPO/PMRSNORM). The further increase of the relative PMRS ratio after stress indicates the higher perfusion reserve in normal regions compared with areas in the perfusion bed of the stenosed coronary artery. The pathophysiologic correlate was previously described by Klocke [3].
To summarize these observations, rest MR examinations can detect scar tissue after myocardial infarction, whereas stress examinations increase the discriminatory power between hypoperfused and normal perfused myocardium by means of a reduced perfusion reserve in the area of the stenosed vessel.
Agreement between MRI and SPECT, which is still the gold standard for the assessment of myocardial perfusion, was only moderate for segmental perfusion scores with the patient at rest. This moderate agreement is consistent with a substantial number of segments found to be abnormal on MRI in normally perfused regions in SPECT. This might be caused by the fact that MRI was able to detect smaller changes in capillary perfusion than was SPECT. Most of the excess of abnormal segments showed only a mild perfusion abnormality on MRI, again suggesting that MRI was more sensitive to small changes in perfusion. Specifically, SPECT has a fairly poor spatial resolution of about 10 x 10 x 10 mm [28], which is comparable with the thickness of the heart wall. Therefore, subendocardial perfusion deficits, which involve only part of the myocardial wall, are below the spatial resolution of SPECT and can be merely appreciated by a reduction in SI. Contrariwise, perfusion MRI yields 3-4 pixels within the myocardial wall, enabling the detection of subendocardial perfusion deficits. Schwitter et al. [24] recently showed the importance of measuring myocardial perfusion in the subendocardial layer, which yielded a sensitivity of 87% in comparison with quantitative coronary angiography.
Another reason for variable findings between MRI and SPECT might be differences in the pathophysiologic approach to myocardial perfusion because technetium uptake is a function of blood flow and tissue viability, whereas T1 reduction in MRI, related to gadopentetate dimeglumine concentration, relies on myocardial perfusion, capillary permeability, and size of extracellular space. Therefore, a certain amount of discrepant results may be due to the fact that the two techniques measure different aspects of coronary artery disease.
A limitation of our study may be the highly selected patient population. Only patients with one- and two-vessel coronary artery disease were included in the study because semiquantitative approaches for the assessment of perfusion require the presence of at least one normally perfused myocardial region. Patients with three-vessel coronary artery disease pose problems in the evaluation process.
Baseline and peak myocardial signals in viable myocardium increased by similar amounts after the first five injections of 0.025 mmol of gadolinium/kg of body weight at 4-min injection intervals [7]. However, because of the persistent delayed uptake in areas of scar formation, there may be residual contrast agent in the myocardium at the time of stress perfusion examinations in patients with prior myocardial infarction. Distinction of hypoperfused and normally perfused regions—and therefore sensitivity—in stress examinations can be increased by performing the stress examination first. However, dipyridamole cannot be used in this examination protocol because its half-life is too long to perform the rest examination in the same study.
In conclusion, perfusion deficits can reliably be detected in patients with single-vessel coronary artery disease using multislice saturation recovery true FISP perfusion imaging. Furthermore, all patients with two-vessel disease displayed a perfusion deficit in both regions supplied by the stenosed epicardial vessels. Providing high signal-to-noise and contrast-to-noise ratios, rest perfusion MRI studies using the saturation recovery true FISP sequence were able to detect scar tissue of chronic myocardial infarction. With respect to the increased spatial resolution compared with SPECT, perfusion MRI may be advantageous for detection of subendocardial perfusion defects. Both semiquantitative perfusion parameters (upslope and peak SI) yielded comparable results for the detection of coronary artery stenoses.
Acknowledgments
We thank Carmel Hayes from Siemens Medical Solutions for the sequence
diagram of the saturation recovery true FISP perfusion sequence.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  R. Gebker, C. Jahnke, I. Paetsch, B. Schnackenburg, S. Kozerke, A. Bornstedt, E. Fleck, and E. Nagel MR Myocardial Perfusion Imaging with k-Space and Time Broad-Use Linear Acquisition Speed-up Technique: Feasibility Study Radiology, December 1, 2007; 245(3): 863 - 871. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. Merkle, J. Wohrle, O. Grebe, T. Nusser, M. Kunze, H. A Kestler, M. Kochs, and V. Hombach Assessment of myocardial perfusion for detection of coronary artery stenoses by steady-state, free-precession magnetic resonance first-pass imaging Heart, November 1, 2007; 93(11): 1381 - 1385. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
