| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Editorials |
1 From the MR Laboratory, Mayo Clinic, 200 First St SW, Rochester, MN 55905. Received January 8, 2004; accepted January 12. Address correspondence to the author (e-mail: riederer@mayo.edu).
Index terms: Angiography, technology • Editorials • Magnetic resonance (MR), vascular studies
Magnetic resonance (MR) techniques have been used to study the physical aspects of the vasculature for more than 40 years, with early studies of blood flow estimation (1) even preceding the initial description of MR imaging techniques by Lauterbur (2). Over the past 20 years, the field of MR angiography has risen from research nascence to everyday clinical reality. Among the critical advances has been the identification of the means for generating the angiographic signal. These have included differential in-phase versus out-of-phase imaging (3), phase contrast MR angiography (4), three-dimensional MR angiography with time-of-flight enhancement (5), two-dimensional imaging with time-of-flight enhancement (6), and contrast material–enhanced techniques (7). In the initial description of each of these, the vessels studied were generally relatively large in diameter or not prone to significant motion (eg, the aorta and carotid arteries or the peripheral vessels). However, for all of these techniques, the coronary arteries were identified as a targeted site of application—and justifiably so, given the significance of coronary artery disease. There has been a never-ending quest to find a reliable, noninvasive method for imaging the coronary arteries. In this current issue, the article by Spuentrup et al (8) presents one look at the contemporary status of coronary artery MR angiography.
The technical obstacles to be surmounted in obtaining high-quality images of the coronary arteries are well known. These principally include motion of the heart itself during the cardiac cycle, as well as respiratory motion, and the difficulty of achieving adequate spatial resolution and an adequate overall signal-to-noise ratio. Other challenges include accommodation of the tortuous geometry of the coronary vascular tree and generation of adequate contrast between the vessel lumen and surrounds, such as epicardial fat. These hurdles and many of the methods developed to address them have been reviewed in detail in a recent review article (9).
Technique selection for MR coronary angiography requires making many choices, including the mechanism for generating the vascular signal, the strategy for covering k space, the use of multiple single-section or volume acquisitions, breath holding or extended acquisition time, and the means for suppressing the signal in fat. Over the past 15 years, numerous specific techniques and their variants have been developed to address one or more of these choices. These techniques include segmentation for measuring multiple k-space lines per cardiac cycle (10), spiral k-space acquisitions for improved speed (11), real-time navigator echoes to monitor respiratory phase (12), and T2 preparation techniques to suppress the signal in surrounding tissues (13). More recently, investigators have studied means for selecting the admissible cardiac phase on a per-patient basis (14). An additional characteristic of MR imaging versus computed tomography (CT) is that it is "undemocratic" in that all (phase encoding) views are not equal. Specifically, the fact that the central k-space lines dictate the overall appearance of the image can be used advantageously. This means that the centric ordering of data for k-space acquisition (15), as well as, for example, the placement of more exacting criteria on navigator echoes for central k-space lines (16), can be effective in providing improved image quality.
As to the specific work by Spuentrup et al, the investigators studied two different means for generating contrast and two different methods for covering k space. Regarding the generation of contrast, the time-of-flight–enhanced signal in blood on standard gradient-echo images was compared with the blood signal on steady-state free precession images, and the latter was found to be superior. Regarding the coverage of k space, a conventional three-dimensional Fourier transform sequence was compared with a hybrid sequence involving Fourier encoding in one dimension and projection reconstruction in the other two, and the hybrid sequence was found to be superior. In addition, the authors carefully integrated into the acquisition many other techniques such as those mentioned earlier, as well as the use of oblique planes of section, in an attempt to further reduce the amount of data that must be acquired.
Compared with many applications of MR imaging, coronary artery MR angiography is one of the more sophisticated means for image acquisition. The pulse sequence diagram in Spuentrup et al (8; fig 1a) shows no less than four different constituent pulse sequences applied within a single cardiac cycle. Furthermore, the interrogated regions are not the same for each of these four, and, even within a given region, fat and water may be treated differently. The acquisition is further complicated by the hybrid nature of the k-space coverage, although this is not apparent in the diagram. Many other coronary artery acquisition strategies have a similar level of sophistication, but there are subtle and possibly important differences in the constituent pulse sequences used. The fact that such complex sequences can be performed to generate meaningful data is a manifestation of the high versatility of MR imaging, the technical capabilities of contemporary MR imaging systems for executing multiple pulse sequences at the submillisecond level of precision, and the ingenuity of MR pulse sequence designers.
In parallel with the continuous and significant developments in MR imaging over the past two decades, many of which are applicable to coronary artery MR angiography, there have been major developments in other imaging modalities, as well—most notably, in CT. The first such development was helical scanning (17), with continuous rotation of the x-ray tube and detector system while the patient was transported through the scanner bore. The second was a further improvement with the use of multiple rows of detectors (18). The first development allowed continuous imaging of an arbitrarily long field of view, while the second has provided a four- to 32-fold improvement in the parallelism of data collection. The application of initial multi–detector row CT technology to coronary angiography has been reported (19). With modern CT imaging systems, the spatial resolution attainable in coronary artery imaging is firmly within the submillimetric isotropic realm, but this is achieved with drawbacks, such as radiation exposure and the possible nephrotoxicity of iodinated contrast material.
It can be instructive to compare CT and MR imaging from the standpoint of the speed with which data are acquired—an important factor when fast scans are desired, such as those done during breath holding. With modern CT, the time per rotation may be 500 msec or less, in the course of which several hundred projections may be acquired. Thus, the time per projection could be estimated as 1–2 msec. In MR imaging, the time per projection can generally be approximated as the repetition time of the pulse sequence used; and with modern machines, the minimum times are typically in the 3–6-msec range, although shorter times are possible. (In the study by Spuentrup et al, the repetition time used was intentionally chosen to be 6.1 or 6.5 msec.) Thus, at the outset, CT confers a one- to twofold advantage in speed. However, this comparison applies to a single transverse section. In coverage of a volume along the long axis of the patient, the improved parallelism in data collection with multi–detector row CT in effect reduces the average time per projection by the number of detector rows, decreasing the figure of merit to the realm of 0.1 msec per projection—10 to 30 times less than with MR imaging.
Granted, MR advocates can respond by saying that there are several methods in the MR imaging research pipeline that might reduce the time per projection or, equivalently, allow fewer measured projections for the same spatial resolution. "Parallel" MR imaging uses signals from multiple receiver coils to reduce the number of projections actually measured (20,21). For N coils, up to an N-fold improvement in speed is possible. However, the quality of the results is highly dependent on the dissimilarity of the coil responses, and it remains to be determined whether improvements greater than two- to fourfold are possible, particularly for anatomic regions deep in the chest. Another method for speed improvement is undersampled projection reconstruction. This method has been applied in MR angiography of the carotid arteries and peripheral vasculature (22). A requirement in this application is that the object of interest, such as the coronary vascular tree, be small and bright in comparison with its surrounds. Undersampling would not work effectively for CT because of bright objects, such as the ribs, that are located at the periphery of the large field of view. Yet another approach is to target the MR angiographic acquisition to a narrowly focused zone, on a vessel-by-vessel basis (23). Improved speed of acquisition at MR imaging would allow improved vessel sharpness, as well as improved depiction of distal coronary vessels.
In many centers, MR angiography is well accepted as the principal diagnostic imaging modality for many vascular territories, such as those in the head and neck, the abdomen (including the renal arteries), and, increasingly, the peripheral vasculature. CT angiography can also be used in a number of these regions. The choice of whether to perform MR imaging or CT often depends on the version of the MR or CT equipment available at a given center, as well as on local imaging expertise and referral patterns. However, the point is that, with regard to these vascular regions, MR angiography and CT techniques have evolved so that their technical reliability is very high for obtaining a diagnostic-quality study with adequate signal-to-noise ratio and spatial resolution and without artifacts and other limitations. Although it is not always necessary to rigorously demonstrate this, in one assessment of the technical performance of triggered contrast-enhanced MR angiography, the technical reliability exceeded 98% for multiple noncoronary vascular areas (24). In spite of innovations by multiple investigators in the past 20 years, coronary artery MR angiography has not yet evolved to this level of reliability. Research studies such as that by Spuentrup et al continue to advance this important field, and additional work is highly warranted.
FOOTNOTES
See also the article by Spuentrup et al in this issue.
REFERENCES
Related Article
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
