Dr. Prince is a Professor of Radiology at Weill Medical College of Cornell University and Columbia College of Physicians and Surgeons, and Chief of Magnetic Resonance Imaging of New York Hospital, New York, NY. Dr. Zhang and Dr. Ersoy are Research Fellows in the Department of Radiology at Weill Medical College of Cornell University. Dr. Dong is a Research Associate from the Department of Radiology, University of Michigan, Ann Arbor, MI.

Magnetic resonance angiography (MRA) (Figure 1) and magnetic resonance imaging (MRI) of the liver, breast, uterine fibroids, cardiac perfusion, brain perfusion, and other applications may require the injection of contrast dynamically during scanning. Gadolinium (Gd) increases image signal-to-noise ratio, thereby enhancing quality. This enhancement is generally greatest during the first pass of the bolus through the tissues of interest. First pass, or dynamic, MR imaging adds additional complexity to preparing and performing the MR examination. An intravenous (IV) line with the proper fittings for the pressure of injection must be started, which allows the contrast to be followed by a saline flush. In addition, the injection must be properly timed to capture the arterial and venous phases of contrast enhancement.

This article reviews the basic steps for dynamic contrast injections in the MR scanner. Also discussed are methodological details relating to selecting IV sites, angiocatheter size, power versus hand injection, selection of IV tubing, starting/testing IV lines, and timing injections for breast, liver, renal artery, peripheral artery, and uterine MR.

Dynamic MR contrast injection technique

Right versus left arm

The right arm has the most direct venous path to the heart for achieving a tight predictable bolus. Left-arm injections are less desirable because contrast must cross through the brachiocephalic vein to reach the superior vena cava. This may cause delay and dilution of the bolus in older patients with atherosclerotic, ectatic aortas that press against the sternum and pinch the left brachiocephalic vein. Selection of the right arm is especially important when imaging the aortic arch in order to avoid overlapping enhancement in the left brachiocephalic vein. However, when there has been a right mastectomy, recent right-arm surgery, local right-arm infection, right subclavian venous obstruction, other local contraindication, or failed attempts at IV placement in the right arm, then the left arm is suitable. Try to avoid the legs, since irritation of leg veins by contrast agents can precipitate thrombophlebitis. If a leg vein is used, it must be flushed generously afterwards with at least 100 mL of normal saline to ensure that no irritating contrast remains stagnant in the veins.

Often, the antecubital fossa is chosen because a large antecubital vein is easily accessible. This vein can handle large caliber IV catheters suitable for the high injection rates required for computed tomography (CT) and during emergencies that can occur from an angiographic procedure. But MR does not require a high injection rate; thus any vein that will take a 20- or 22-gauge angiocatheter is acceptable. It can be placed in the forearm, wrist, or even the back of the hand when more proximal veins are not available. Often, patients can quickly point out the best sites, thereby saving the difficulty of finding them.

What size angiocatheter?

Angiocatheter size for a particular patient is generally chosen based upon the viscosity of what will be injected and the required injection rate. Angiocatheter size is indicated by the gauge; the larger the gauge, the smaller the caliber of the angiocatheter and the greater the resistance to the injection. High viscosity or fragility of the fluids (ie, blood) and necessity of high injection rates require large caliber angiocatheters (smaller gauge: 16 or 18). But smaller caliber angiocatheters (larger gauge: 20 or 22) are easier to insert. For dynamic MR contrast injections, generally 20-gauge is a good compromise--it provides sufficient caliber for fast injections but is still small enough to be easy to insert. In a pinch, a 22-gauge angiocatheter is also acceptable, especially if you warm the Gd contrast to body temperature to lower its viscosity.

Power versus hand injection

Hand injection gives the operator more control over bolus administration and allows early detection of contrast reactions and IV problems, such as blockage or extravasation. When standing next to the patient for hand injection, the operator can elicit maximum patient cooperation by more effectively communicating breath-holding instructions and assuring the patient that everything is going smoothly. For these reasons hand injection is preferred.

With conventional angiography and CT, the contrast dose/injection rate may be too great for hand injection. In addition, power injectors are used to avoid radiation exposure to personnel. However, with MRI, there is no radiation danger to staff performing hand injection and the dose/injection rate is substantially lower. Thus, power injection is not necessary with MR studies. Especially for babies and small children, patients with central lines, and patients with tenuous IVs, power injection can be dangerous and should be avoided.

Power injection has the advantage of delivering contrast at a consistent and predictable injection rate. It also allows a single operator to run the MR examination from the control room without having to enter the scanner room. However, power injectors can add additional complications. The power injector user interface is not always intuitive, and accidental contrast injections occur periodically. Since the IV is tethered to the power injector, sliding the patient into the scanner occasionally pulls out the IV line when the tubing is not long enough or gets caught. The high pressure of a power injector can cause loose IV connections to pop open, spraying contrast all over the magnet bore. Battery and mechanical failure of the power injector can ruin studies as well. For all of these reasons, we generally prefer hand injection for MR examinations.

Prepare the intravenous tubing

It is important to use IV tubing that allows simultaneous attachment of separate syringes for the Gd contrast injection and a subsequent saline flush. The mechanics of these connections must be set up to prevent reflux of the Gd contrast or saline flush. The IV tubing must be long enough to extend from the IV site on the patient inside the magnet to the operator performing the injection standing outside the magnet. This requires at least 1 m--preferably 2 m--of length. But longer tubing should be avoided because it may have a large dead space and also higher capacitance, which can alter bolus timing and injection rate. The male end of the tubing that plugs into the angiocatheter should have a locking mechanism to prevent separation under the pressure of the injection. The female end, which extends outside the magnet, must have a similar locking mechanism that allows attachment of Gd and saline flush syringes. Ideally, the tubing should be sufficiently thick-walled to prevent kinking, expanding, or bursting under the pressure of fast injection and there should be a minimum of connections. A clamp is necessary to prevent fluid dripping when the IV is removed.

One example is the SmartSet (TopSpins, Inc., Ann Arbor, MI) (Figure 2), which was developed specifically for performing dynamic contrast injections for MRA. It has one-way valves that allow switching between the contrast injection and saline flush. In this way it is easy to have one continuous bolus followed by flush without any gaps. The SmartSet is made of thick-walled tubing that does not expand or kink. Its coil design adjusts to the necessary length up to nearly 2 m. It also has a clamp and normally closed check valves to prevent fluid from dripping when it is removed from the patient. An optional side port allows injection of medication close to the IV site in the event of an adverse event.

It is important to use the same tubing for all patients receiving dynamic Gd injections to become familiar with the details of performing the injections: resistance to injection, volume of contrast necessary to prefill the dead-space, mechanism of switching from contrast to flush, etc. In this way, the operator can concentrate on the more important issues of bolus timing and coordination with patient breath-hold without having to think about distracting IV details.

Prepare the arm and puncture vein

Apply a tourniquet high on the upper arm and search for a suitably distended subcutaneous vein. If no veins appear, you can sometimes feel them by palpating the arm. Having the patient squeeze his/her hand to make a fist several times helps to maximize venous engorgement. If you still cannot find any veins, it may be useful to wrap the arm in a warm compress for 5 to 10 minutes to stimulate peripheral vasodilatation and venous distension. If no veins are found after a meticulous search, then move the tourniquet to the forearm and search in the distal forearm, wrist, and hand, or move to the other arm.

Once a suitable vein is found, clean and disinfect the area by swiping several times with alcohol wipes (Figure 3A). If the arm has too much hair, it may be necessary to shave the IV site with a disposable razor for cleanliness and to simplify securing the IV line with tape.

Select a 20- or 22-gauge angiocatheter. A 20-gauge angiocatheter allows a faster injection rate; 22-gauge is easier to insert into small veins. Take the angiocatheter apart and put it back together to experience how it works and to get a sense of how much force is required to slide the plastic catheter over the metal stylet.

To puncture the vein, use one hand to apply counter tension against the skin while the other hand advances the needle (Figure 3B). The hand applying the counter tension will be pulling skin toward the wrist opposite from the direction the needle will be advancing. When applying this counter tension, be careful not to compress inflow to the vein, which may cause the vein to collapse. First, advance the angiocatheter through the skin overlying the vein or adjacent to the vein. Use a quick jabbing motion to minimize patient discomfort. Then advance the angiocatheter well into the vein and look for the dark red flashback of blood at the angiocatheter hub.

If this first pass is unsuccessful, slowly withdraw the angiocatheter without pulling it all the way out of the skin and watch carefully for the flashback to occur. If the catheter is still not within the vein, then advance it again for a second attempt. While withdrawing the catheter, always stop before pulling it out completely to avoid repeating the skin puncture. If after several attempts the vein is never entered, consider it a failure, release the tourniquet, place gauze over the skin puncture site, withdraw the angiocatheter, and tape down the gauze. Then move on to a more distal vein, which can be distended when the tourniquet is placed below the failed site.

Once the angiocatheter is well seated within the vein, slide the plastic angiocatheter forward deeper into the vein over top of the needle (Figure 3C). The hub of the angiocatheter should advance all the way to the skin puncture site. The plastic catheter should slide forward easily.

Then release the tourniquet, apply gentle pressure over the vein to collapse it so that blood will not pour out of the angiocatheter when the stylet (needle) is removed. Once you remove the stylet, dispose immediately in a sharps container.

Attach SmartSet

Attach and lock the IV tubing (SmartSet) to the angiocatheter and secure with tape (Figure 4). Taping is one of the most important tasks, because this is what prevents you from having to repeat the IV insertion in the event of an inadvertent tug on the IV tubing.

Test the IV

Inject saline to test the IV line; there should be no resistance. If there is any difficulty, try repositioning the patient's arm to make it straighter. External rotation may also be helpful since internal rotation sometimes pinches the antecubital vein. Sometimes an IV will begin to work if it is withdrawn slightly so the tip of the IV sits in a better position within the vein. The angiocatheter tip may also not work if it is pressed up against a venous valve; this is also alleviated by withdrawing the angiocatheter a few millimeters. If necessary, you may also test the IV by removing the entire Y check valve assembly and aspirating (Figure 5) until blood is seen entering the IV tubing. Alternatively, you can test the line by engaging the clamp and injecting or aspirating via the side port.

Dynamic injection for breast, liver, angiography, and uterus

There are several important details to be aware of to ensure optimal dynamic contrast injections during MR scanning. Once the scanner is prepped and ready to scan, it is necessary to prime the IV tubing before starting the injection. After the IV is verified to work by testing with normal saline, prime the tubing by advancing sufficient Gd contrast into the tubing to fill the dead space. It is important to know the priming volume of the IV tubing in advance. The SmartSet priming volume is
6 mL. As you prime the tubing with Gd, try to get a sense of how much force is required to advance the syringe plunger for a 2 mL/sec injection rate. This way you will know how much force to use once it is time to begin the main injection.

It is also important to know how k-space is mapped during the MR pulse sequence. Central k-space dominates image contrast. The location of the Gd bolus when central k-space is acquired will determine what is bright and what is dark on the image. With centric or elliptical centric mapping of k-space, the center of k-space is at the beginning of the scan. With the more common sequential ordering, the center of k-space is in the middle of the scan. Sequential ordering produces fewer artifacts. Elliptical centric, however, sometimes simplifies timing, especially for MRA sequences.

Breast

One approach to dynamic contrast-enhanced MRI of the breast is to acquire a series of 60- to 90-second three-dimensional (3D) spoiled gradient-echo volumes with fat saturation at multiple time intervals, including precontrast, arterial phase, and then every 60 to 90 seconds for 10 minutes. To minimize artifact, k-space should be ordered sequentially (ie, do not use centric or elliptical centric ordering of k-space). After checking the precontrast volume to be sure the fat saturation is working properly and the breast anatomy of interest is included in the imaging volume, begin the arterial-phase scan and contrast injection simultaneously. Make sure to finish the Gd injection and saline flush before the midpoint of the arterial phase scan.

Differentiation of malignant tumors from normal breast parenchymal enhancement requires temporal postprocessing on a computer workstation. Cancer tissue shows sharp enhancement during the early arterial phase while benign tissues, such as normal breast parenchyma, enhance more gradually (Figure 6). Malignant tissue may also show washout of contrast in later phases.

Liver

Dynamic enhancement of the liver is typically performed as an axial two-dimensional (2D) or 3D spoiled gradient-echo acquisition with fat saturation (Figure 7). Spoiling helps sensitize the sequence to paramagnetic contrast (ie, Gd). Repeat the scan multiple times with breath-holding as follows: pre-injection, arterial phase, immediate post-arterial phase, and delayed (if a cirrhotic liver, then a delay of 2 to 3 minutes is adequate; if there is a lesion that might be hemangioma or cholangiocarcinoma, then repeat every 2 to 3 minutes up to at least 10 minutes). To capture the arterial phase, it may be helpful to time the acquisition with a test bolus, automatic triggering, or fluoro-triggering. Peak arterial phase enhancement of the liver occurs 8 to 10 seconds after contrast is detected in the midabdominal aorta. Alternatively, a reasonably accurate method is to inject the Gd (2 mL/sec) followed immediately by a 20-mL saline flush. Just as the saline flush is finishing, the scan is begun. With axial 3D imaging, fat suppression is essential to prevent excessive wrap-around of fat in the slice direction. However, to keep the sequence from becoming too long for breath holding, fat suppression can be performed with a single inversion pulse per slice loop instead of being applied every repetition time (TR).

Renal artery

MR angiography is performed with a 3D spoiled gradient pulse sequence using the shortest possible TR and echo time (TE). Arterial-phase timing is essential and can be accurately accomplished using a test bolus, MR fluoroscopy, or automated triggering. When using elliptical centric ordering of k-space, it is useful to use CENTRA or to recess the absolute center of k-space 3 to 4 seconds from the beginning of 3D data acquisition to avoid ringing artifact and excessive venous enhancement. If recessing the absolute center of k-space from the beginning of the scan is not possible, then there should be a 6- to 8-second image acquisition delay between detecting the arrival of Gd and beginning elliptical centric data acquisition. A dose of 0.1 to 0.2 mmol/kg Gd contrast (20 to 30 mL) injected as fast as possible, ~2 to 3 mL/second, is recommended. When Gd infusion is complete, flush with at least 20 mL normal saline. At the end of the arterial-phase scan, have the patient take 3 to 4 quick breaths and then scan again to catch the portal venous phase. This 3D MRA data is then transferred to a computer workstation to perform volume rendering, maximum intensity projections, and reformations.

Pulmonary artery

Bolus timing for the pulmonary arterial phase is critical. Since many vascular structures in the thorax enhance at different time points, it is best to order k-space sequentially for minimal artifact. Optimal timing for a 20- to 30-second sequentially ordered scan is to begin injecting at a rate of 2 mL/second and then to give a delay of 5 to 10 seconds before starting the scan. Use a 5-second delay for healthy patients with an IV in the antecubital fossa and 10 seconds for patients with pulmonary hypertension or an IV in the hand. Be sure to follow the contrast injection immediately with saline flush at the same rate. Use at least 20 mL of saline flush to completely purge the IV tubing and also to help flush contrast through the arm veins. A large contrast dose, 40 mL, helps to obtain enough signal-to-noise to evaluate the small pulmonary arteries and also to compensate for bolus timing errors. A large contrast dose at a high injection rate, >4 mL/sec, may demonstrate perfusion defects that accompany occlusive emboli.

3D bolus-chase peripheral MRA

Use time-resolved 2D projection or 3D MRA to calculate the contrast travel time to the calf (Figure 1). For fast or average flow rates, use short first and second stations (ie, 10 to 20 seconds). Generally, it is easier to make the thigh faster than pelvis because only 20 slices are necessary to completely image the superficial femoral artery. For patients with slower flow, use longer first and second stations to avoid getting ahead of the bolus. The scan delay = time to fill pelvis ­ (scan time)/2 + 5-second safety margin; bolus duration = 3/2 pelvis scan time ­ 1/3(time to feet) (about 25 seconds). Typically, the injection rate is 1.5 mL/sec. For 40 mL at an injection rate of 1.5 mL/sec, the bolus duration is 26 seconds. First, perform a mask run without contrast at all 3 stations. Quickly check the mask to be sure the anatomy is adequately covered. It is useful if the first and second stations both cover the common femoral bifurcation so that in the event of a timing error, it will be well visualized on at least one station. The actual arterial-phase run is complicated because you have to coordinate injection, breath-holding, and scanning. Getting all this to work optimally requires standing in the scanner room next to the patient during the hand injection. Try to perform the arterial-phase scan as fast as possible after the mask run to minimize the chance of intervening motion.

Uterus

Prior to fibroid embolization, MRI is performed to confirm the diagnosis and to identify contraindications, including pedunculated fibroids and auto-infarction (absence of fibroid enhancement). Acquire a 2D or 3D gradient-echo sequence with spoiling and fat saturation in the sagittal plane precontrast, during the arterial phase, and at 1-minute and 5-minute delays. This same sequence is repeated at 3-month postembolization follow-up to verify infarction of the leiomyomas. Infarction is characterized by absence of enhancement (Figure 8). This dynamic Gd-enhanced sequence may also be helpful for assessing endometrial carcinoma, and it also has been described for identifying placenta accreta, increta, and percreta.

Conclusion

Imaging during contrast injection can dramatically enhance MR image quality and can be especially important for MR angiography, organ perfusion, and discriminating malignant from benign lesions. Optimizing IV site selection, IV tubing, hand injection, timing, and patient cooperation helps maximize quality of dynamic contrast-enhanced MR.

Acknowledgments

The authors acknowledge Michelle L. Moore for her assistance in preparing the manuscript. Special thanks to Evelyn Pence and Patricia Ferrer for medical illustration.