Dr. Daniel Parker is in his third year of Radiology residency at the Baylor University Medical Center in Dallas, TX. He graduated from the University of Utah School of Medicine and will begin a Body Imaging fellowship at the University of Utah in 2003.

Dr. Dennis Parker is a Professor of Radiology and Medical Informatics at the University of Utah School of Medicine.

Cardiac gating is used in a variety of magnetic resonance imaging applications. Its role has evolved significantly over the last several years, with advances in imaging hardware and software. The accuracy of gated sequences depends on an accurate representation of the cardiac cycle. New techniques, including modified electrocardiogram (EKG) detection algorithms and vector electrocardiography, have improved detection of the R wave. Gating techniques are useful for reducing motion artifact in the thorax, especially in cardiac imaging, and for optimizing blood signal in time-of-flight (TOF) and phase-contrast imaging. Selectively acquiring k-space at discrete phases in the cardiac cycle may offer increased signal in small vessels in TOF imaging.

Cardiac gating can be used in magnetic resonance imaging (MRI) applications to evaluate the heart and great vessels ^1-3 and to help maximize inflow signal in time-of-flight (TOF) and phase-contrast (PC) magnetic resonance angiography (MRA). ⁴ Cardiac gating is a valuable tool, especially in cardiac imaging applications in which the complex motion of the heart would preclude imaging with traditional methods. The role of MRI in cardiac imaging is expanding. ⁵ Cardiac MRI can provide useful information regarding cardiac and vascular anatomy, ⁶ myocardial function, and myocardial perfusion and viability. ⁷ Ultimately, real-time cardiac cine is expected, which could obviate the need for cardiac gating, though at the expense of signal-to-noise ratio (SNR). This would likely increase the clinical utility of cardiac MRI, as gating methods are imperfect and less time-efficient.

Cardiac gating still has a role in thoracic applications where it can reduce artifacts caused by great vessel pulsation. It may be useful in evaluating the anterior mediastinum, which is also subject to motion artifacts. Contrast-enhanced magnetic resonance angiography (CEMRA) has in many instances supplanted traditional TOF and PC applications in the body where cardiac gating has been used. ⁸

Gating techniques have been explored in TOF applications in an effort to exploit phases of the cardiac cycle with highest segmental blood velocities. Modified techniques have attempted to capitalize on increased blood velocity as a function of time in the cardiac cycle and yet maintain imaging times at acceptable levels. Given the limitations of intracranial CEMRA, gating applications may help to provide increased SNR, especially in small vessels. A brief description of some ongoing research evaluating the pulsatile effects of blood flow in intracranial MRA is outlined in this article.

Multiple cardiac gating techniques have been developed to overcome observed pitfalls. Special algorithms have been developed that aid in the detection of the "true" R wave even with a limited EKG signal. ⁹ Techniques such as vector gating provide a trigger without using EKG leads. ¹⁰ One weakness all cardiac gating sequences share is the requisite increased acquisition time. Modified MRA gating techniques that selectively acquire the center of k-space near systole offer a compromise between temporal resolution and scanning time.

Cardiac gating techniques

Cardiac gating applications depend on an electric trigger received by the magnetic resonance scanner that corresponds with the patient's heart cycle. An EKG voltage source, specifically the R wave, is commonly used as the trigger. ^3,11 Most detecting algorithms use a set threshold voltage to determine the cardiac trigger. The R wave is a useful trigger because the voltage peak is usually much higher than the other points of the EKG waveform, allowing for easier detection (figure 1). Imaging sequences allow acquisitions at desired phases in the cardiac cycle relative to the cardiac trigger.

In cardiac MR, gating is used to depict phases of systole and diastole. The QRS waveform signals ventricular contraction. This typically occurs approximately 160 ms after the initial P wave. An image acquired at the same time as the R wave would depict the heart in end diastole, because mechanical systole occurs approximately 200 ms after electrical systole. An image obtained 200 ms after the R wave would depict mechanical systole.

Velocity-dependent sequences (eg, TOF and PC MRA) can be optimized by acquiring data at intervals in the cardiac cycle where there is maximum arterial velocity. A TOF MRA sequence of the abdominal aorta, obtained entirely in early systole, should yield much higher signal than one obtained in diastole. Magnetic resonance angiography of the coronary arteries can be optimized by imaging at select phases in the cardiac cycle with maximum blood flow (ie, diastole).

For a conventional MRI sequence, imaging time is the product of the radio frequency repetition time (TR), the number of phase-encoding steps, and the number of excitations. In conventional MR, the phase encoding steps (usually 128 to 512) progress without a delay and the total imaging time is therefore relatively short. In a gated sequence, phase encoding steps advance with sequential R waves. Acquiring 256 phase-encoding steps would require 256 separate R waves, in the case of conventional cine imaging. In the case of segmented k-space techniques, in which multiple radiofrequency (RF) pulses and phase-encoding steps are obtained within a given "segment" of the cardiac cycle, fewer R waves are required to fill k-space.

Accurate gating depends on an accurate gating signal. As depicted in figure 2, if the trigger threshold is set too low, or there is an inadequate R wave, the cardiac cycle will be misrepresented. An amplified T wave, arrhythmias, or premature ventricular contractions can cause inaccurate gating. Newer gating algorithms analyze the slope of the waveform as well to determine the R wave. ⁹

Quality control can greatly aid in producing a reliable and accurate EKG waveform. EKG leads should be placed either on the front or the back of the thorax following Enthoven's Law. ¹² The lead with the highest R wave peak voltage is selected. When using an EKG trigger, the skin lead interface should be maximized. Oils and dead dermis can cause increased conduction impedance, which can be resolved with an abrasive gel. Sometimes it is necessary to shave excessive hair to provide better lead-skin contact.

EKG leads in MRI have a few unique potential problems. Radio frequency pulses used in signal generation can cause increased lead impedance and a noisy waveform. The large magnetic fields inherent in MRI may generate induction currents within EKG leads that could potentially burn a patient. The lead wires should exit the bore of the magnet parallel to the magnetic field to reduce artifacts in the lead signal.

Vector triggering methods in cardiac MRI have demonstrated the potential of depicting the cardiac cycle more accurately. In vector cardiac gating a trigger is produced that depends on heart motion. Many of the deficiencies of traditional EKG gating are overcome with this technique, even in problematic patient populations (ie, patients with cardiac arrhythmias). ¹⁰

Less precise cardiac gating can be done with a peripheral pulse oximeter. A pulse oximeter detects blood flow with an optical trigger. Blood flow is monitored on a finger, toe, or ear. In contrast to EKG gating, in which cardiac systole happens shortly after the R wave, in peripheral gating the signal is a delayed representation of systole.

Cardiac gating in thoracic imaging

Magnetic resonance imaging is establishing itself as a key resource in cardiac imaging, although development of cardiac MRI applications has been limited by the heart's complex motion. Cardiac gating is still a valuable tool in the majority of cardiac imaging sequences. Magnetic resonance sequences are being developed and used to evaluate heart morphology, myocardial wall motion, ¹³ perfusion and viability, ⁶ and coronary artery caliber. ⁷ Both acquired and congenital heart disease can be evaluated. Magnetic resonance imaging has the potential to provide superior pericardial and mediastinal imaging. ¹⁴ The ability of MR to acquire a three-dimensional volume data set, its superior intrinsic soft tissue contrast, and ability to quantify blood velocity are indicators of its great potential. Advances in imaging software and hardware have made MRI of the heart realistic.

Basic cardiovascular imaging sequences can be classified as "white blood" and "black blood." White-blood techniques use fast gradients to obtain cine type images and are useful for wall motion evaluation. Key white-blood cardiac MR sequences include steady-state free precession (SSFP) or true fast imaging with steady-state precession (true FISP) ^1,15 and fast gradient echo. Advances in SSFP in recent years have markedly improved the ability of MRI to evaluate cardiac anatomy and function. Steady-state free precession sequences offer much more uniform blood signal even in slow flow regions. The increased temporal and spatial resolution of SSFP is becoming a mainstay in cardiac imaging.

Black-blood sequences include traditional spin-echo, and more recently double and triple inversion recovery fast spin-echo techniques. ¹⁶ The double inversion recovery fast spin-echo sequence removes the signal of flowing blood by applying an RF inversion pulse. This technique is limited in regions with slower blood flow, such as in aneurysms and dissections, and long vessel segments. In triple inversion recovery sequences, an additional inversion RF pulse is added to null fat signal. This may lead to problems when trying to differentiate hematoma from fat as methemoglobin may also be suppressed by this sequence. ¹⁷

Cardiac wall motion can be evaluated with "tagging" sequences. These techniques evaluate wall motion and wall thickening during contraction. A grid is formed in the imaging plane by applying RF pulses. This grid allows accurate quantification of myocardial characteristics. Potentially, MRI can differentiate viable from nonviable myocardium by comparing enhancement characteristics. ^5,7 Healthy myocardium enhances as contrast extends from the capillary bed into the interstitial space. Nonviable myocardium demonstrates decreased early phase enhancement, but shows increased delayed enhancement as infarcted tissue allows more gadolinium to diffuse.

Magnetic resonance imaging is becoming a useful tool for evaluating the coronary arteries. New techniques in coronary MRA include FISP and double-inversion recovery black-blood methods. True FISP often suffers from heterogeneous signal and requires careful shimming before an examination. Coronary MRA is currently useful in evaluating the proximal coronary arteries, including variant anatomy (eg, origins of the right and left coronary arteries), aneurysms and pseudoaneurysms, and in the evaluation of bypass grafts. ^6,7

Cardiac gating in intracranial MRA

In current intracranial TOF techniques, large blood vessels with high flow rates are visualized routinely. The smallest blood vessels, which are routinely visualized on X-ray angiography, are not visualized routinely by MRA. There is evidence that this loss in visibility is due to signal saturation experienced by the slow-moving blood in the smallest vessels. ^18,19 There may be blurring and signal loss due to the pulsatile nature of blood flow in the brain. ²⁰ The cardiovascular system is dynamic, and a wide range of velocities is experienced throughout the system as a function of time in the cardiac cycle. Pulsatile blood flow is hypothesized to cause variation in signal intensity as a function of time in the cardiac cycle.

Cerebral MRA imaging was performed at 1.5 T with actively shielded gradients. A commercially available cardiac gated sequence (2D Fastcard, General Electric Medical Systems, Milwaukee, WI) was modified to perform EKG-synchronized high-resolution three-dimensional (3D) acquisition. Cardiac gating was performed using peripheral pulse oximetry. Images were obtained using a four-element bilateral phased-array coil. ^21,22 For all studies, imaging parameters were: TR/TE = 25 ms/4.0 ms; tip angle = 30š; imaging field-of-view (FOV) = 220 mm (x), 165 mm (y), and 32 mm (z) with nx, ny, nz = 512, 192, 32, giving voxel dimensions of 0.43 (x), 0.86 (y) and 1.0 (z). The phase-encoding technique of the modified Fastcard pulse sequence is illustrated in figure 3.

In this technique, the cardiac cycle is divided into multiple time intervals or "phases." Phase 1 (*1) is the first interval after the cardiac trigger. By sharing views, images at intermediate time points are also obtained (ie, the first n/2 samples of the second phase (*2) are combined with the last n/2 samples of phase 1 (*1) assuming n ky phase-encodings). Thus, images for seven cardiac phases are obtained from the four time intervals shown in figure 1. After the same set of ky phase-encodings has been obtained for each cardiac phase, and for each kz phase-encoding, a subsequent set of ky phase encodings is used and the process is repeated for subsequent cardiac cycles.

Six volunteers were imaged. Images were acquired using: 1) at least one cardiac gated seven-phase acquisition, 2) one ungated 3D TOF acquisition, and 3) two or three cardiac gated single-phase acquisitions where the ky phase encodings were obtained with a centric order in each phase (see below). For the cardiac gated acquisitions, a user-specified delay from the trigger was chosen in terms of a number of repetitions (TR times) from the cardiac trigger (pressure pulse). For all seven-phase protocols, the cardiac cycle was divided into four intervals. After reconstruction and combining of overlapped phases, images are obtained in seven overlapping intervals, with temporal spacing of T/2 ms.

The 3D Fastcard sequence was modified to allow centric encoding of the ky phase-encodings as a function of time during the cardiac cycle. The lowest ky encodings are acquired early in each cardiac cycle progressing to the highest ky phase-encodings at the end of the cardiac cycle. To obtain a single phase, approximately 32 views per segment were acquired. The time for the centric acquisition is only slightly higher (about 5%) than that of the nongated 3D TOF acquisition.

Blood signal was measured using a computer program, which estimates the signal-difference-to-noise ratio (SDNR) along a vessel segment. ^23,24 In a given volunteer, the blood signal in the same vessel segments was measured for each acquisition.

The third, fifth, and seventh images from a seven-phase study obtained from Volunteer A are shown in figure 4A through 4D. The seventh image in the sequence appears to have significantly more vascular detail than any of the previous images in this sequence (figure 4C). The seventh image represents flow acquired in a 200-ms window centered at 1200 ms after the initial pressure wave. The first image was acquired in a 200-ms window beginning after the initial pressure wave.

Images obtained using the centric technique for Volunteer A are shown in figure 5. In this case, offsets of 500, 750, and 1000 ms (20, 30, and 40 X TR) were used.

Figure 6 demonstrates the average peak SDNR averaged over 94 vessel segments for Volunteer A for two seven-phase studies (the first initiated immediately and the second 500 ms after the pressure pulse), a 3D TOF study, and three single-phase centric acquisitions (at various delays). The measurements are plotted as a function of effective time in the cardiac cycle where the center of k-space was sampled.

The percent difference of the average SDNR measurement for the seven-phase and centric acquisitions from each volunteer compared with a nongated 3D TOF sequence was computed. The maximum (most positive) percent difference results for all six volunteers are summarized in Table 1.

This work shows that a significant increase in intracranial small vessel visibility and SDNR can be achieved by selectively acquiring images in early systole. The multiple-phase image studies clearly demonstrate that blood signal intensity fluctuates as a function of time in the cardiac cycle (figure 6). This fluctuation is likely caused by a fluctuation in the blood velocity and in the number of excitation pulses experienced by the blood between the time that it enters the slab and the time of arrival in the smallest arteries.

A technique that acquires the central lines of k-space in early systole was developed and tested. This cardiac centric technique shows the potential of significantly increasing small vessel visibility and SDNR with very little change in acquisition time. Although these new techniques appear promising, further studies will be necessary before their full clinical utility can be determined.

Summary

Cardiac gating is becoming a key tool in cardiovascular applications. Such applications are currently used to suppress artifacts caused by the heart's complex motion and to exploit signal in velocity sensitive sequences. This recent research illustrates how cardiac gating and modified k-space data collection may be used in intracranial TOF imaging. As MRI hardware and software evolve, cardiac gating is establishing a major role in cardiovascular imaging.