TECHNOLOGY

FOCUS: Non-Invasive Coronary Imaging - MRI

MAGNETIC RESONANCE ANGIOGRAPHY IMAGING OF THE CORONARY ARTERIES

Yi Wang Ph.D., H. Dirk Sostman, M.D. Department of Radiology,, Weill Medical College of Cornell University, New York, N.Y. , USA

 ABSTRACT

 Introduction

 SUMMARY

 References

 A FOREST OF COLUMNS

ABSTRACT

Noninvasive magnetic resonance imaging (MRI) technology has the capacity to assess cardiac anatomy and function, promising significant advances in diagnosis and management of heart disease.
However, as yet the technical and clinical results from MR imaging specifically of the coronary arteries have not been sufficient to substantially impact clinical care of patients with ischemic heart disease.
Accordingly, an understanding of the imaging approaches, physical constraints and technical innovations, which contends with the physiologic and pathologic substrate of coronary atherosclerosis, is important.
This paper reviews the technical development of MRI of coronary arteries:
coronary magnetic resonance angiography (MRA).
When suitably developed, coronary MRA will be the major clinical application of cardiac MRI.
We will summarize the challenges that confront coronary MRA and the recent developments that make coronary MRA promising.
We hope to provide the reader an informed appreciation of how coronary MRA is performed and thus how it may be used, now and in the foreseeable future.
(Heart Views.2001;2(3):98-111) © 2001 Hamad Medical Corporation.

Key Words:

   magnetic resonance imaging    coronary arteries
   atherosclerosis    coronary artery disease    angiography

Introduction

Cardiovascular disease remains the leading cause of death in many developed countries.
In the United States, every year, more than 1.
5 million acute myocardial infarctions resulting in almost 600,000 deaths are attributed to coronary artery disease (CAD) (1).
Half of these infarctions occur suddenly in previously asymptomatic individuals (2).
The standard technology for evaluating CAD definitively is x-ray angiography (XRA), but it is invasive, expensive, and carries risk.
Cardiac catheterization is associated with 0.12-0.2% mortality, 0.03-0.2% cerebrovascular accident, 0-0.25% myocardial infarction, 0.57-1.6% local infarction, 0.3-0.63% arrythmias (3-6).
The cost for coronary XRA is high ($3,500-6,000 in the United States) (7-10).
These limitations make XRA inappropriate for use as a screening or discovery test.
Accordingly, physicians utilize multiple diagnostic procedures (including history, blood testing, stress ECG, coronary calcium scoring, stress ultrasound and stress perfusion scintigraphy) to assess the likelihood of CAD in patients.
Despite this typically extensive (and expensive) clinical, laboratory and imaging evaluation prior to catheterization, up to 20% of x-ray coronary angiograms demonstrate no significant coronary artery disease and therefore do not lead to further interventions (6).
Furthermore, these pre-angiography tests themselves carry small risks and expose patients to significant inconvenience as well as expense.
It would be very desirable to diagnose these stenoses in a noninvasive and cost effective manner using MRI.
Patients without evidence of stenosis on noninvasive imaging would avoid the cost and potential risks of x-ray angiography

A noninvasive and less expensive diagnostic modality would advance patient care and might reduce net healthcare costs for CAD diagnosis if it were sufficiently accurate and appropriately utilized.
Follow-up imaging of patients undergoing revascularization procedures could also be readily obtained and used to guide further therapy. Serial studies for progression of disease could be carried out, facilitating the development of drug or gene therapy for CAD (11).
MRI is a very promising approach for realizing the potential of noninvasive coronary imaging because of its high sensitivity to blood flow. Not only can the vessel lumen be visualized but also the flow velocity can be mapped directly and in principle plaque also can be visualized directly.
MR angiography (MRA) has the potential to image the proximal coronary arteries (3-5 mm diameter) and major branches and distal arteries (~2 mm). The 3D imaging of the coronary tree in MRA would overcome the vessel-overlapping problem in projectional XRA.
Clinically significant coronary stenoses (>50% diameter narrowing) are mostly located in the proximal portion of the major coronary arteries (left main, left anterior descending, left circumflex, right) (12-14). Arteries of this size and smaller in the brain and extremity are well imaged with MRA methods, and neuro and peripheral MRA have become valuable diagnostic tools that are routinely used in clinical practice (15-18). The major challenge for MRA of the coronary arteries is cardiac and respiratory motion.
Initial results using ECG triggering and respiratory gating or breath-holding demonstrate that coronary MRA is highly feasible and very promising.
However, as will be described below, further development is needed to optimize motion suppression and make the techniques robust for routine clinical application.
In addition to imaging coronary arteries, MRI is capable of comprehensive study of the anatomy and function of the heart, so-called “one-stop shopping” which might include assessing ventricular function, myocardial perfusion, and wall motion or thickening.
However, coronary MRA is one of the most important components of cardiac MRI, and recently has been identified as the No.
1 special area of emphasis for NIH research (19).
We organize this review article on coronary MRA development in the following order.

  1) Coronary motion

The motion of coronary arteries, due to respiration and cardiac contraction, is the major challenge to coronary MRA.
Study of the characteristics of coronary motion dissects this major challenge and lays the foundation for coronary MRA techniques.

Techniques that rely on the ECG signal to identify cardiac motion and on diaphragm navigator feedback or breath-holding to minimize respiration effects can depict coronary arteries.
However, image quality is variable and diagnostic accuracy is limited.

A new navigator approach is presented that can directly measure coronary motion and overcome the inaccuracy in detecting coronary motion from the ECG signal and the diaphragm navigator signal.

1. CORONARY MOTION

As mentioned above, the success of a coronary MRA technique is in large measure determined by its effectiveness in suppressing motion artifacts, so the value of studying the characteristics of coronary motion cannot be overemphasized.
First, we describe the coronary motion due to respiration.
Then, we examine the coronary motion due to cardiac contraction.
Of course, cardiac contraction and respiration occur concurrently and cannot be separated in motion measurement.
A useful and quantitative mathematical decomposition of coronary motion is presented.
The coronary motion consists of a dominant global component that can be measured in MRI and a small local deformation component.

1.1 Respiratory Motion (Fig.1)     1.2 Cardiac Motion    1.3 Total Global Motion Of Coronary Arteries

1.1 Respiratory Motion (Fig.1)

The heart is situated superior to the diaphragm, which during the respiratory cycle acts like a piston, pushing and pulling the heart in the superior-inferior (SI) direction.
Respiration amplitude varies on a 100 msec time scale and does not have a strictly periodical waveform, though there is a 2-4 sec, relatively regular cycle of inspiration and expiration.
The dominant component of coronary motion due to respiration is a linear translation along the SI direction, on the order of 10 mm (20).
Other components are much smaller, including translation along anterior-posterior (AP) and right-left (RL) directions on the order of 2 mm.
Finally, there is a dilation along the SI direction because the coronary roots are displaced less than the distal arteries at the bottom of the heart.
The coronary motion due to respiration is approximately correlated to the SI motion of the

Fig.1

Fig.1. An RCA section of the heart at three different respiration level: inspiration (left), middle and expiration (right). (Courtesy: MRM33:713-719, 1995 )

diaphragm, though there may be hysteresis during the respiratory cycle (21).
These characteristics imply that the respiratory component of coronary motion can be approximately monitored at the diaphragm.
However, for maximum accuracy motion measurement for coronary MRA should be performed directly on the heart.

1.2 Cardiac Motion

The coronary motion due to cardiac contraction and rotation (referred to as "cardiac motion" in this paper) varies on a 10 msec (or less) time scale with a fairly periodic waveform (~ 1 sec cycle).
Although the ECG signal is used to trigger CMRA acquisition, it is not a measure of coronary motion.
Cardiac motion can be measured quantitatively using invasive cine x-ray angiography method, and

Fig.2

Fig.2. Rest period versus heart rate in 13 patients. As heart rate increases, the duration of the rest period decreases. The rest periods for the right and left arteries differed substantially at low heart rate.

properties crucial to CMRA data acquisition have been identified (22).
The rest period or period of minimal cardiac motion (position variation < 1mm) varies substantially from patient to patient, and is not uniquely determined by the length of the cardiac cycle.
The rest period measured from a group of patients varied from 66 to 333 msec for the left coronary artery and from 66 to 200 msec for the right coronary artery.
The cardiac motion velocities measured immediately after the rest period are significantly larger than the velocities measured immediately before the rest period.
The transverse (RL and AP) motion range of the right coronary artery is greater than that of the left coronary artery.
When there is no respiratory motion in the thoracic trunk and no change in cardiac output, the heart should return to the same location from heartbeat to heartbeat.
Overall cardiac motion can be decomposed into two components:
1] global motion (translation, rotation about the long axis and dilation/contraction at the center of the heart) and 2] local deformation that is not accounted for in the global model, including bulging during contraction and twisting caused by non-uniform rotation.
A very useful and interesting finding is that most (~91%) of the 3D cardiac motion of coronary bifurcations can be represented by the global motion (23).
The global motion can be measured from its effects on MR signal.
The local deformation cannot be measured easily, but its magnitude is proportional to the magnitude of the global motion.
Accordingly, overall cardiac motion effects can be reduced by minimizing global motion within image data. This is the foundation for the use of navigator measurement of and correction for global translation, rotation and dilation in MRI.

1.3 Total Global Motion Of Coronary Arteries

As mentioned above, cardiac contraction and respiration occur concurrently and cannot be separated in motion measurement.
A useful and quantitative mathematical decomposition of coronary motion is available, however.
The discussion in the above three paragraphs leads to the following mathematical expression for the total global motion of coronary arteries.
Let x be the position vector, and Dr the displacement vector.
Then the respiratory motion component can be expressed in a matrix form:

#x'=Cr(x - Dr),[1]

where Cr is a diagonal matrix characterizing compression/dilation.
Cardiac motion can be expressed conveniently in a coordinate system XYZ, with Z the long axis of the heart and XZ in the plane of the interventricular septum.
The transformation from the laboratory xyz coordinate system to the cardiac major axes XYZ coordinate system consists of 1) rotating about the x-axis until the z-axis becomes the Z-axis and 2) rotating about the Z-axis until the x-axis gets to the septum (XZ) plane (both rotating angles can be easily measured from scout scans of the heart).
This transformation is, of course, intuitively familiar to any trained echocardiographer.
The Jacobian associated with this transformation (J) is the product of these two rotations and can be obtained from a scout scan.
The rotation and contraction associated with cardiac motion, adding the respiratory component, is composed of transforming from xyz to XYZ, displacement (Dc) contraction / dilation (Cc), rotation (Rc), and transforming from XYZ back to xyz:

x"=J-1(RcCcJx'-Dc)=A(x-D)[2]

2. EXISTING CORONARY MRA TECHNIQUES

Coronary arteries can be well visualized using MRI.
The cohort of existing coronary MRA techniques relies on the ECG signal to identify the period of minimal cardiac motion for data acquisition of coronary MRA.
Typically, a delay to mid-diastole estimated from the ECG is used to start MR data acquisition. However, the ECG signal bears neither direct nor quantitative correlation with coronary motion, and consequently there may be substantial residual cardiac motion to degrade image quality.
For suppression of respiration artifacts, breath-holding or navigator technique (see below for a discussion of navigator technique) that measures and corrects for diaphragmatic motion have both been used extensively.
Most coronary MRA techniques use gradient echo signal acquisition in order to acquire ‘bright blood’ image contrast. This has appeal due to similar image contrast with conventional angiography, and in addition it lends itself to further MR signal enhancement from the use of intravenous gadolinium-based MR contrast agents.
Satisfactory image contrast for coronary arteries when using “bright blood” technique depends upon suppression of the bright signal from epicardial fat surrounding the coronary arteries (24).
Image contrast for more distal coronary artery segments can be further enhanced by suppressing the MR signal of surrounding myocardium, which can be achieved using magnetization transfer pulses (25) or T2 preparation pulses (26).
Conventional rectilinear data sampling trajectories in k-space are standard and robust.
More efficient spiral trajectories (27) or multiple rectilinear lines (echo planar imaging) (28) have also been used for data sampling, but substantial diligence is required to minimize off-resonance artifacts.
In summary, a standard data acquisition sequence for coronary MRA consists of 1) magnetization preparation prior to data acquisition commencing at mid-diastole to suppress epicardial fat, and 2) using gradient echo sequence to acquire a few lines in k-space (over a duration of ~ 100 msec).
Depending on the method used to minimize respiratory artifacts, existing coronary MRA techniques can be grouped into breath-holding and navigator types.

2.1 Breath-Hold Coronary MRA            2.2 Navigator Coronary MRA

2.1 Breath-Hold Coronary MRA

Breath-Hold 2D Coronary MRA                              Multiple Breath-Hold 3D Coronary MRA Using A Respiratory Feedback Monitor                      Breath-Hold Contrast Enhanced 3D Coronary MRA

Breath-Hold 2D Coronary MRA

The 2D acquisition can be completed within a breath-hold of 20 heartbeats using a gradient echo sequence of TR ~ 10 msec and acquiring 8 k-space lines per heartbeat.
This 2D breath-hold technique was first to demonstrate the feasibility of coronary MRA (24).
However, clinical trials up to date have yielded undesirably variable accuracy (Table 1).
Limitations intrinsic to the breath-hold 2D approach include inconsistency between breath-holds, poor resolution or missing artery segments in the ‘section’ plane (due to large section thickness and non-isotropic spatial resolution, and poor SNR.
These limitations of 2D data acquisition could be overcome by using a 3D acquisition approach, which provides higher SNR and higher spatial resolution than 2D acquisition and natural 3D display for the tortuous coronary artery tree.

Multiple Breath-Hold 3D Coronary MRA Using A Respiratory Feedback Monitor

A typical 3D coronary MRA acquisition requires many breath-holds to complete.
It is essential to maintain the heart in the same position during all breath-holds for a multiple breath-hold acquisition.
A respiratory feedback monitor can aid the subject to hold his/her breath consistently (29,30).
A measure of respiration level such as the diaphragm position can be visually displayed to the subject.
While the respiratory feedback can reduce the variation among breath-holds and permit high quality multiple breath-hold 3D coronary MRA (31) (Fig.3), substantial cooperation and comprehension, in addition to simple breath holding, is required from the patient.
This makes such technique a difficult and not very robust choice for coronary imaging in clinical practice. To date, there is no reported clinical result from this technique.

Breath-Hold Contrast Enhanced 3D Coronary MRA

The 3D acquisition can be made fast enough to allow single breath-hold 3D coronary MRA (32).
This is accomplished by using a high sampling rate and fast gradient system, but it carries with it the trade-off of poor SNR.
The relatively poor SNR intrinsic to this approach can be alleviated by using MR contrast agents that enhance blood signal substantially.
This approach is used in the vast majority of clinical MRA in all parts of the body (33).
One specific implementation of such a contrast enhanced breath-hold 3D coronary MRA technique is illustrated in Fig.4.
Disdaqs (executing RF and gradient pulses with data acquisition disabled) and fat saturation were used to establish spin equilibrium.
All background tissues are well saturated, generating bright contrast for T1 shortened blood (Fig.5).
During the acquisition window in the cardiac cycle, an edge-center-edge view order was used such that the center of k-space was acquired in mid-diastole when motion of the coronary arteries was minimal.
As is well known, the central k-space views govern the overall image contrast and provide most of the signal, so making these correspond to the motionless portion of the cardiac

Fig.3

Fig. 3. Proximal right coronary artery and left coronary artery demonstrated in an oblique section reformmatted from a 3D data set acquired with multiple breath-holds aided with a respiratory feedback monitor.

cycle has obvious advantages (Fig.6).
The whole 3D data acquisition in this specific example was completed in a 32-heartbeat breath-hold.
Data acquisition began when the contrast bolus arrived at the ascending aorta.
Coronary arteries are well delineated with the breath-hold contrast enhanced 3D coronary MRA using motion-matched view ordering, and the background suppression in the contrast enhanced acquisition allows easy display of the 3D coronary data set.
Very likely, contrast enhancement will play a significant role in improving coronary artery contrast.
Initial clinical evaluations of the breath-hold 3D contrast enhanced MRA techniques yielded encouraging results, but the diagnostic accuracy is not yet satisfactory (Table 1).
The data acquisition matrix has to be increased to provide sufficient image resolution, and residual cardiac and respiratory motion effects (diaphragm can drift during breath-holding) have to be reduced.

 2.2 Navigator Coronary MRA

Breath-hold acquisition has fundamental shortcomings including the necessity for cooperation in maneuvers that may be beyond the physiologic capability of certain patients; limited spatial resolution or SNR due to short available scan time; and unanticipated or residual motion during attempts of breath holding.
These drawbacks can be overcome with the navigator approach.

Fig.4

Fig.4. Timing schematics for contrast enhanced breath-hold 3D coronary MRA using motion matched view order. (Courtesy: Radiology 215:600-607, 2000) Fig. 5
Fig. 5. (a) an oblique section reformatted from a breath-hold contrast enhanced 3D acquisition to demonstrate LAD. (b) is non-enhanced breath-hold 2D acquisition of the LAD in the same subject. Compared to (b), the contrast enhanced acquisition in (a) provided much better background suppression and correspondingly substantially improved image coronary artery contrast. (Courtesy: Radiology 215:600-607, 2000)
Fig. 6
Fig.6. Comparing centric (a) to motion matched (b) view order in contrast enhanced 3D CMRA, motion-matched view order substantially sharpens the coronary artery (arrows). (Courtesy: Radiology 215:600-607, 2000) Fig. 7
Fig.7. Timing schematics for navigator gated coronary MRA acquisition.
Fig.8
Fig.8. 3D coronary MRA acquired under breathing (a) without navigator gating and (b) with navigator gating. Motion effects (ghosting and blurring) are substantially reduced by navigator gating. Images are reformatted to show the proximal left anterior descending artery (LAD, arrow). (Courtesy: Radiology 198:55-60, 1996)                     Navigator Concept                              Navigator Motion Correction In MRI                      Retrospectively Gated 3D Coronary MRA                      Prospective Or Real-Time Gated 3D Coronary MRA                      Adaptive Correction                                                  Navigator Concept The concept of navigator motion reduction was introduced in the early 1950’s for correcting the distorting effects of the Earth’s atmosphere in ground-based astronomical imaging (34). In ground-based optical astronomy, resolution is limited not only by the size of telescopes, but importantly by atmospheric turbulence, which deforms the image on a millisecond time scale. Adaptive optical systems consisting of natural or artificial guide stars, wavefront sensors and real-time phase delay corrections compensate for the atmospheric distortions and significantly improve image resolution (35). This navigator motion reduction idea has also been very successfully applied in camcorders and other imaging technologies (36). The navigator approach offers a tool to deal with motion problems in data acquisition generically. The key concept is that effects of motion can be suppressed to a degree determined by the accuracy achieved in measurement of the motion. Navigator Motion Correction In MRI Spatial resolution of MR images of the moving coronary artery tree is limited by physiological motion, not by the intrinsic spatial resolution of clinical MR imaging (which is determined by the capability of the MR scanner in sampling high spatial frequencies). Navigator based motion gating and motion correction techniques were introduced in MRI in the late 1980’s (37,38). The beauty of navigator technique in MRI is that the motion to be corrected can be detected from the MR signal itself, with no additional hardware needed for motion monitoring. Currently navigator motion Fig.9

correction techniques are widely used to correct undesired motion effects in diffusion imaging, fMRI, body MRA and spectroscopy.
The initial results from our group and others vindicate the utility of these navigator methods for reducing motion effects in CMRA.
The widely used navigator to monitor respiration is the 2D RF excitation pulse (39-41) used to acquire a longitudinal cylinder of tissue through the posterior part of the diaphragm (29).
The diaphragm position can be extracted accurately from the pencil beam navigator echo using edge detection algorithms such as the least squares algorithm (42).
The SI respiratory motion of the heart is then estimated from the SI motion of the diaphragm.
Such a pencil-beam navigator echo can be flexibly positioned away from the heart and thus will not affect the cardiac magnetization.
In addition to monitoring the SI position of the diaphragm, the pencil beam navigator echo can also be used to monitor the AP oriented motion of the chest wall and the upper abdominal wall.
Once the motion of interest is detected, navigator methods to suppress the motion’s undesirable effects on the MR image are applied.
Those explored for CMRA include gating and correction.
The gating method acquires image data only when the amplitude of the detected motion falls within the gating window.
Gating both retrospectively and prospectively has been implemented.
Linear motion can compensated for by phase shifts.

Retrospectively Gated 3D Coronary MRA

Retrospective navigator gating consists of oversampling data (~5X) and then selecting the set of data with associated diaphragm positions within or closest to the gating window (43).
Because certain data points may lay outside the gating window, this retrospective gating method may be not effective, but it is easy to implement.
Several clinical evaluations of the retrospective gated 3D coronary MRA technique yielded limited accuracy (Table 1).
The effectiveness of navigator gating can be improved using the real time gating, which is the topic of the remaining part of this section and the next section.

Prospective Or Real-Time Gated 3D Coronary MRA

Real time gating requires computer capability that can process navigator data and communicate with data acquisition within one R-R interval or less (44,45). . In one implementation of real-time navigator gating for 3D coronary MRA, the pencil beam navigator echoes are acquired immediately before and after the image data in every cardiac cycle (Fig.5).
The SI positions of the diaphragm are extracted from the two navigator echoes in real-time (< 2 ms).
The decision to control data acquisition is made at the beginning of the data acquisition in the next heartbeat.
When both positions are within the gating window, image data collected between the navigator echoes is accepted and image data acquisition is advanced to the next segment of k-space.
Otherwise, the same segment of k-space is reacquired in the next heartbeat.
Such an implementation rejects data if motion occurs during the image data acquisition period.
The location of the gating window can be optimally selected using the diminishing algorithm (46).
The effects of residual motion within the gating window can be minimized by smoothly distributing residual motion in the k-space through view ordering (47).
Real-time navigator gating (at a gating window of 3 mm SI range of the diaphragm) provides significant reduction of respiration effects (Fig.8), and can be used to
follow up patients after revascularization (Fig.9).
Clinical evaluations of the preliminary implementations of the real time navigator gated 3D coronary MRA technique demonstrated variable accuracy (Table 1); the inconsistency in image quality be may caused by the fundamental limitation of the diaphragm navigator echo as discussed in the next session on current developments in coronary MRA techniques.

Adaptive Correction

Effective real-time navigator gating requires a narrow gating window, which leads to undesirable increase in scan time.
To shorten scan time, a wide gating window can be used to increase the data acceptance rate and to correct for the residual displacements within the wide gating window (48-50).
The displacement effects on MRI data are phase shifts, which can be corrected for by multiplying a phase factor:

Fc(k) = F(k) exp[-ik D].[3]

The SI displacement of the diaphragm dSI has to be scaled to estimate the displacement of the heart DSI = a dSI.
The scaling factor a can be searched in the range (0,1] for the best correction (20).
This combination of real-time navigation gating and adaptive motion correction can reduce motion effects without the need for extensive patient cooperation and without an undesirable increase in scan time (Fig. 10).
However, the effectiveness of motion correction based on phase shift is quite limited, because the coronary motion is more complex than the linear motion estimated from the diaphragm.

3. CURRENT DEVELOPMENTS IN CORONARY MRA TECHNIQUES

The inability to achieve consistent image quality through use of existing coronary MRA techniques (as reviewed above) may be due to errors associated with the attempts to estimate the cardiac and respiratory motion components.
The ECG signal is used to minimize cardiac motion, but the ECG has only an indirect and inconsistent correlation with coronary motion.
Breath holding imposes significant limitations on image resolution and/or SNR, and there may still be residual motion despite conscientious effort by the patient .
Diaphragm navigator echo may not accurately monitor cardiac motion due to respiration when there is hysteresis between the diaphragm and heart positions.
Accordingly, most efforts to improve the accuracy of coronary MRA are focusing on achieving more accurate measurement of coronary motion or suppression of motion effects on the MR image.
Pencil beam navigator echo can be positioned directly on the heart, rather than on the diaphragm.
In this way, it is possible to monitor the motion of the heart in a manner similar to an M-mode ultrasound.
Such ‘M-mode’ pencil beam navigator can be used to identify the period of minimal cardiac motion (51).
The motion of coronary arteries is very difficult to extract from such a signal, because it is a mixture of chamber blood, myocardium and arterial blood.
However, if only the coronary artery or the myocardial wall is excited and contributes to the MR signal, then there is no need to use spatial resolution to identify
the heart, and the motion of coronary arteries can be estimated directly and conveniently from the k-space data (52).
An exciting development towards this goal of fast direct and accurate coronary motion measurement is a recently presented cardiac navigator echo using epicardial fat (53).

3.1 Epicardial Fat Navigator         3.2 Real Time Volumetric Orbital Navigator Approach

3.1 Epicardial Fat Navigator

The heart is comprised of chamber blood, myocardium, and epicardial fat. The motion of coronary arteries correlates directly with the motion of the myocardium and epicardial fat, but not the motion of the chamber blood. Myocardium and blood have the same resonance frequency and therefore it is difficult to excite myocardium only. However, the epicardial fat can be selectively excited using a spatial-spectral selective pulse (53,54). Human coronary arteries are surrounded with plentiful epicardial fat. There is sufficient fat signal for monitoring the coronary motion without affecting coronary blood signal. Preliminary data demonstrate the feasibility of the volumetric fat navigator echo for monitoring cardiac motion (53). This fat navigator echo work suggests the following general volumetric orbital navigator approach to motion issues in coronary MRA.

3.2 Real Time Volumetric Orbital Navigator Approach

Note that the global motion of coronary arteries can be characterized by a few parameters (~ 10, Eq.2), and that many points (~ 200) are sampled in an echo in MRI.
So it is possible to solve all global motion parameters from a single navigator echo by designing k-space sampling trajectories that sensitize all motion components.
This leads to the development of orbital navigator echo.
The general k-space expression of the motion effects fm(x) = f(A(x_D)) is:

Fm(k) = eik DF((A-1)Tk) [4]

The six motion parameters in matrix A (the rotation angle in Rc, two scaling factors in Cc, and three scaling factors in Cr) can be determined by the least squares minimization, min{Ýk(|Fm(ATk)|-|F(k)|)2}.
This requires sampling a k-space volume where ATk might be located around k, and a search for the minimal by resampling data. Under certain situations, trajectories may be simplified.
In the 2D case, the k-space trajectory is a circle, which allows complete determination of rotation and translation in a plane (52). A trajectory consisting of 3 axes can be used to simultaneously detect 3D translation and dilation (53).
Once the global motion is known, its complex effects on MR signal in Eq.4 can be simultaneously corrected by adjusting the gradients and RF.
The gradient adjustment according to the following k-space vector:

SUMMARY

Coronary magnetic resonance angiography has advanced substantially in the ten years since it was first attempted. The progress in coronary MRA has been realized through increased imaging speed due to more powerful gradient hardware, implementation of blood signal enhancement through use of Gadolinium contrast agents, better understanding of coronary motion resulting in development of techniques for more effective suppression of motion effects in coronary MRA data acquisition.
Coronary MRA has been found useful clinically in checking the patency of coronary by-pass grafts (55,56) and identifying anomalous coronary arteries (57,58). In spite of these advances, consistent and accurate imaging coronary lesions of clinical significance remains elusive, as reviewed in this paper. The fundamental limitation common to all existing coronary MRA techniques is that the motion of coronary arteries during data acquisition is not measured accurately enough (or not measured at all). The errors in motion measurement are the bottleneck for effectiveness of motion suppression in coronary MRA. Once coronary motion is accurately measured, the real-time navigator approach can suppress motion effects in coronary MRA, limited primarily by the accuracy of the motion measurement.

Current work on directly measuring the motion of coronary arteries may lead to accurate coronary motion measurement (53) and may provide a potential solution to the fundamental motion problem of coronary MRA.
On-going work in fast imaging, such as true FISP imaging (59), may further increase data acquisition speed and signal-to-noise ratio.
These current developments in coronary motion measurement and fast imaging are encouraging. They certainly will bring us closer to the goal of accurate imaging of coronary lesions, and may allow us to reach this goal in the foreseeable future.

References

1.    American Heart Association. Heart and stroke facts: statistics. 1993;8. Dallas,
       Texas.

2.   Fuster V, Badimon L, Cohen M, et al. Insights in to the pathogenesis of acute
      ischemic syndromes. Circulation 1998;77:1213-1220.

3.   Davis K, Kennedy JW, Kemp HG, Judkins MP, Gosselin AJ, Killip T. 
      Complications of   coronary arteriography from the collaborative study of
      coronary artery surgery  *CASS). Circulation 1979;59:1105-1112.        

4.   Kennedy JW. Complications associated with cardiac
      catheterization and anigraphy. Cathet Cardiovasc Diagn 1982;8:5-11.

5.   Wyman RM, Safian RD, Portway V, Skillman JJ, MacKay RG, Baim DS. Current
      complications of diagnostic and therapeutic cardiac catheterization. J Am Coll Cardiol       1988;12:1400-1406.

6.    Johnson LW, Lozner EC, Johnson S, et al. Coronary arteryography 1984-1987: a
       report of the Registry of the Society for Cardiac Angiography
       and Interventions. I.Results and complicatoins. Cathet Cardiacvasc
       Diagn.1989;17:5-10.

7.    Hansing CE. The risk and cost of coronary angiography.I. Cost of coronary
       angiography in Washington State. JAMA. 1979;242(8):731-4.

8.    Thomson PD. Cost of coronary angiography. JAMA.1980;243(12):1232-3

9.    Keavney B. Haider YM. McCance AJ. Skehan JD. Normal coronary angiograms:
       financial victory from the brink of clinical defeat? Heart. 1996;75(6):623-5.

10.  Manning WJ, Edelman RR. Magnetic resonance coronary angiography. Magn.
      Reson. Quarterly 1993;9:131-151,.

11.  Mack CA. Magovern CJ. Budenbender KT. et al.Salvage angiogenesis induced by
       adenovirus-mediated gene transfer of vascular endothelial growth factor protects
       against ischemic vascular occlusion. J Vasc Surg. 1998;27(4):699-709.

12.  Vieweg WV. Alpert JS. Johnson
       AD. Dennish GW. Nelson DP. Warren SE. Hagan AD. Distribution and severity of
       coronary artery disease in 500 patients with angina pectoris. Catheterization &
       Cardiovascular Diagnosis. 1979;5(4):319-30.

13.  Vieweg WV. Alpert JS. Hagan AD. Caliber and distribution of
       normal coronary arterial   anatomy.
       Catheterization & Cardiovascular Diagnosis.1976;2(3):269-80.

14.  MacAlpin RN. Abbasi AS. Grollman JH Jr. Eber L.
       Human coronary artery size during   life. 
       A cinearteriographic study. Radiology 1973;108(3):567-76.

15.  Potchen, E.M. Haacke, J.E. Siebert, A. Gottschalk, “Magnetic resonance
       angiography: concepts & applications”, Mosby-Year Book Inc., St. Louis, 1993.

16.  Anderson, R.R. Edelman, P.A. Turski, “Clinical magnetic resonance angiography”,
       Raven Press, New York, 1993.

17.  Baum RA. Rutter CM. et al.
       Multicenter trial to evaluate vascular magnetic resonance  angiography of 
       the lower extremity. American College of Radiology Rapid
       Technology Assessment Group. JAMA. 1995;274(11):875-80

18.  Owen RS. Carpenter JP. Baum RA. Perloff LJ. Cope C.
       Magnetic resonance imaging of angiographically occult runoff vessels in peripheral
       arterial occlusive disease. New England Journal of Medicine. 1992.326(24):1577-81..

19.  Budinger TF, Berson A, McVeigh ER, Pettigrew RI, Pohost GM,
       Watson JT, Wickline SA.
       Cardiac MR imaging: report of a Working Group sponsored by the 
       National Heart,  Lung, and Blood Institute. Radiology 1998;208:573-576.

20.Wang Y, Riederer SJ, Ehman RL, Respiratory motion of the heart: kinematics and the      implications for the spatial resolution in coronary imaging, Magn Reson Med
    1995;33:713-719,.

21.  Nehrke K, Bornert P. Study of the respiratory motion of the heart using multiple
       navigator pulses. ISMRM, 404,200.

22.  Wang Y, Vidan E, Bergman GW. Cardiac motion of coronary arteries: variability in
       the rest period and implications for coronary MR angiography.Radiology.
       1999; 213    (3):751-8.

23.  Potel MJ, Rubin JM, Mackay SA, et al. Methods for evaluating cardiac wall motion in
       three dimensions using bifurcation points of the coronary arterial tree. Invest Radiol
       18: (1) 47-57 1983.

24.  Edelman RR, Manning WM, et al, Coronary arteries: breath-hold MR angiography.
       Radiology. 1991;181:641-643.

25. Wang Y, Grist TM, Mistretta CA, Dispersion in magnetization transfer contrast at a
      given specific absorption rate due to variations of RF pulse parameters in the
      magnetization transfer preparation, Magnetic Resonance
      in Medicine  .1997;37:957-962.

26.  Brittain JH, Hu BS, Wright GA, Meyer CH, Macovski A, Nishimura DG. Coronary
       angiography with magnetization-prepared T2 contrast. Magn Reson Med.
      1995;33(5):689-96.

27.  Meyer CH, Hu BS, Nishimura DG, Macovski A. Fast spiral coronary artery imaging.
       Magn Reson Med. 1992;28(2):202-13.

28.  Wielopolski PA, Manning WJ, Edelman RR. Single breath-hold volumetric imaging of
       the heart using magnetization-prepared 3-dimensional segmented echo planar
       imaging. J Magn Reson Imaging.1995;5(4):403-9.

29.  Liu Y, Riederer SJ, et al, A monitoring, feedback, and triggering system for
       reproducible breath-hold MR imaging. Magn Reson Med 1993;30:507-511.

30. Wang Y, Christy PS, Korosec FR, Alley MT, Grist TM, Polzin JA, Mistretta CA.
      Coronary MRI with a respiratory feedback monitor: the 2D imaging case.
       Magn Reson Med. 1995;33(1):116-21.

31.  Wang Y, Grimm RC, Rossman PJ, Riederer SJ, Ehman RL, 3D coronary MR
       Angiography in multiple breath-holds using a
       respiratory feedback monitor, Magnetic Resonance in Medicine. 1995;34, 11-16.

32.  Wielopolski PA, van Geuns RJ, de Feyter PJ, Oudkerk M. Breath-hold coronary MR
       angiography with volume-targeted imaging. Radiology. 1998;209(1):209-19.

33.  Prince MR, Grist TM, Debatin JF. 3D contrast MR angiography. Springer, -Verlag,
       Berlin, 1999.

34.  HW Babcock, The possibility of compensating astronomical seeing, Publ. Astron.
       Soc. Pac. 65, 229,1953.

35.  Thompson LA, Adaptive optics in astronomy. Physics Today. 1994;27(12):24-31.

36.  Shake-free video, Popular Science, p48, September 1998..

37.  RS Hinks, Monitored echo gating (mega) for the reduction of motion artifacts, Proc.
       6th SMRI, p.48, 1988.

38.  RL Ehman, JP Felmlee, Adaptive technique for high-definition MRI of moving
       structures. Radiology 1989;173:255-263,.

39.  Pauly J, Nishimura D, Macovski A. A k-space analysis of small-tip-angle excitation.
       J.  Magn. Reson. 1989;81, 43-56.

40.  Pauly J, Nishimura D, Macovski A. A linear class of large- tip-angle selective
       excitation pulses. J. Magn. Reson. 1989;82, 571-587.

41.  Hardy CJ, Cline HE. Broadband nuclear magnetic resonance pulsese with two-
       dimensional spatial selectivity. J. Appl. Phys. 1989;66, 1513-1516.

42.  Wang Y, Grimm RC, Felmlee JP, Riederer SJ, Ehman RL. Algorithms for extracting
       motion information from navigator echoes. Magn Reson Med. 1996;36(1):117-23.

43.  Li D. Kaushikkar S. Haacke EM. Woodard PK. Dhawale PJ.
       Kroeker RM. Laub G. Kuginuki Y. Gutierrez FR.Coronary arteries:
       three-dimensional MR imaging with
       retrospective respiratory gating. Radiology. 1996;201(3):857-63.

44.  TS Sachs, GH Meyer, et al, Real-time motion detection in spiral MRI using
       navigators. Magn Reson Med 1994;32:639-645.

45.  Y Wang, PJ Rossman, et al, Navigator-based real-time respiratory gating and
       triggering for reduction of respiration effects in three-dimensional coronary MR
       imaging. Radiology 1996;198:55-60,.

46.  Sachs TS, Meyer CH, Irarrazabal P, Hu BS, Nishimura DG, Macovski A. The
       diminishing variance algorithm for real-time reduction of motion artifacts in MRI.
       Magn Reson Med. 1995;34(3):412-22.

47.  Jhooti P, Gatehouse PD, Keegan J, Bunce NH, Taylor AM, Firmin DNPhase ordering
       with automatic window selection (PAWS): a novel motion-resistant technique for 3D
       coronary imaging. Magn Reson Med. 2000;43(3):470- 80.

48.  Oshinski JN. Hofland L. Mukundan S Jr. Dixon WT. Parks WJ. Pettigrew RI. Two-
       dimensional coronary MR angiography without breath holding. Radiology.
       1996;201(3):737-43.

49.  P.G. Danias, M.V. McConnell, V.C. Khasgiwala, M.L. Chuang, R.R. Edelman, W.J.
       Manning. Prospective navigator correction of image position for coronary MR
       angiography. Radiology 1997;203, 733-736.

50.  Wang Y, Ehman RL. Retrospective adaptive motion correction for navigator-gated
       3D coronary MR angiography. J Magn Reson Imaging. 2000;11(2):208-14.

51.  Wang Y, Watts R, Mitchell I, Nguyen TD, Bezanson JW, Bergman GW, Prince MR.
       Coronary MR angiography: selection of acquisition window of
       minimal cardiac motion   with electrocardiography-triggered navigator cardiac
        motion prescanning--initial  results. Radiology. 2001;218(2):580-5.

52.  Fu ZW, Wang Y, Grimm RC, Rossman PJ, Felmlee JP, Riederer SJ, Ehman RL.
       Orbital navigator echoes for motion measurements in magnetic resonance
       imaging.Magn Reson Med. 1995;34(5):746-53.

53.  Nguyen TD, Wang Yi, Prince MR, A Novel Navigator Technique for Fast and Direct
       Detection of 3D Displacement of the Coronary Arteries, ISMRM 173,2001.

54.  Meyer CH. Pauly JM. Macovski A. Nishimura DG. Simultaneous spatial and spectral
       selective excitation. Magnetic Resonance in Medicine. 1990;15(2):287-304.

55.  Wintersperger BJ, Engelmann MG, von Smekal A, Knez A, Penzkofer HV, 
       Hofling B,  Laub G, Reiser MF. Patency of coronary bypass grafts: 
       assessment with breath-hold    contrast-enhanced MR angiography--value 
       of a non-electrocardiographically   triggered technique. Radiology.
       1998;208(2):345-51.

56.  Langerak SE, Kunz P, de Roos A, Vliegen HW, van Der Wall EE. Evaluation of
       coronary artery bypass grafts by magnetic resonance
       imaging. J Magn Reson Imaging.1999;10(3):434-41.

57.  Duerinckx AJ, Bogaert J, Jiang H, Lewis BS. Anomalous origin of the left coronary
       artery: diagnosis by coronary MR angiography. Am J Roentgenol.
       1995; 164(5):1095-7.

58.  McConnell MV, Stuber M, Manning WJ. Clinical role of coronary magnetic resonance
       angiography in the diagnosis of anomalous coronary arteries. J Cardiovasc Magn
       Reson. 2000;2(3):217-24.

59.  Deshpande VS, Shea SM, Laub G, Simonetti OP, Finn JP, Li D. 3D magnetization-
       prepared true-FISP: A new technique for imaging coronary arteries.
       Magn Reson Med. 2001;46(3):494-502.

60. Manning WJ, Li W, Edelman RR. A preliminary report comparing magnetic
      resonance coronary angiography with conventional angiography. N Engl J Med.
      1993; 328(12):828-32.

61.  Duerinckx AJ, Urman MK. Two-dimensional coronary MR angiography: analysis of
       initial clinical results. Radiology. 1994;193(3):731-8.

62.  Pennell DJ, Bogren HG, Keegan J, Firmin DN,Underwood SR. Assessment of
       coronary artery stenosis by magnetic resonance imaging. Heart. 1996;75(2):127-33.

63.  Post JC, van Rossum AC, Hofman MB, de Cock CC,Valk J, Visser CA.
       Clinical utility  of two-dimensional magnetic resonance angiography
       in detecting coronary artery  disease. Eur Heart J. 1997;18(3):426-33.

64.  Watanuki A, Yoshino H, Udagawa H, Yokoyama K, Nitatori T, Hachiya J, Ishikawa K.
       Quantitative evaluation of coronary stenosis by coronary magnetic resonance
       angiography. Heart Vessels. 2000;15(4):159-66.

65. Kessler W, Laub G, Achenbach S, Ropers D, Moshage W, Daniel WG. Coronary
      arteries: MR angiography with fast contrast-enhanced three-dimensional breath-hold
      imaging--initial experience. Radiology. 1999;210(2):566-72.

66.  Regenfus M, Ropers D, Achenbach S, Kessler W, Laub G, Daniel WG,
       Moshage W.   Noninvasive detection of coronary artery stenosis
       using contrast-enhanced three-  dimensional breath-hold magnetic resonance
       coronary angiography.  J Am Coll Cardiol. 2000;36(1):44-50.

67.  van Geuns RJ, Wielopolski PA, de Bruin HG, Rensing BJ, Hulshoff M, van Ooijen
       PM, de Feyter PJ, Oudkerk M. MR coronary angiography with breath-hold targeted
       volumes: preliminary clinical results. Radiology.2000;217(1):270-7.

68.  Kessler W, Achenbach S, Moshage W, Zink D, Kroeker R, Nitz W, Laub G,
       Bachmann K. Usefulness of respiratory gated magnetic resonance coronary
       angiography in assessing narrowings > or = 50% in diameter in native coronary
       arteries and in aortocoronary bypass conduits. Am J Cardiol. 1997;80(8):989-93.

69.  Muller MF, Fleisch M, Kroeker R, Chatterjee T, Meier B, Vock P.
       Proximal coronary  artery stenosis:
       three-dimensional MRI with fat saturation and navigator echo. 
       J Magn    Reson Imaging. 1997;7(4):644-51.

70.  Woodard PK, Li D, Haacke EM, Dhawale PJ, Kaushikkar S, Barzilai B, 
       Braverman AC,   Ludbrook PA, Weiss AN, Brown JJ, Mirowitz SA, 
       Pilgram TK, Gutierrez FR. Detection of  coronary stenoses on source and 
       projection images using three-dimensional MR
      angiography with retrospective respiratory gating: preliminary experience.
       Am J Roentgenol. 1998;170(4):883-8.

71.  Schmidt M, Theissen P, Crnac J, Jochims M, Voth E, Baer FM, Schicha H, Erdmann E.
       [Clinical value of magnetic resonance tomography in imaging coronary stenoses.
       A comparison with coronary angiography and myocardial scintigraphy].
       Dtsch Med Wochenschr. 1999;124(44):1294-300. German.

72.  Sardanelli F, Molinari G, Zandrino F, Balbi M. Three-dimensional, navigator-echo MR
       coronary angiography in detecting stenoses of the major epicardial vessels, with
       conventional coronary angiography as the standard of reference. Radiology.
       2000; 214(3):808-14.

73.  Lethimonnier F, Furber A, Morel O, Geslin P, L'Hoste P,Tadei A, 
       Jallet P, Caron-Poitreau  C, Le Jeune JJ. Three-dimensional coronary artery 
       MR imaging using prospective real-time  respiratory navigator and linear phase 
       shift processing: comparison with conventional  coronary angiography.
       Magn Reson Imaging.1999;17(8):1111-20.

74.  Stuber M, Botnar RM, Danias PG, Sodickson DK, Kissinger KV, Van Cauteren M, De
       Becker J, Manning WJ. Double-oblique free-breathing high resolution three-dimensional
       coronary magnetic resonance angiography. J Am Coll Cardiol. 1999;34(2):524-31.

A FOREST OF COLUMNS

Interior of The Great Mosque of Cordoba Cordoba, Spain   Supported in part by USPHS Grants R01HL60879, R01HL62994, R01HL64647 Correspondence to: Yi Wang, Ph.D., Radiology/MRI Research 515 East 71st Street, Suite S120, New York, NY 10021, USA. Email:yiwang@med.cornell.edu