High-Field versus Low-Field Scanners

Scanners come in a variety of field strengths, and options are available for various gradient strengths and slew rates intended to optimize scanning. Software options are often available at additional cost. These issues can cause confusion when selecting a scanner for purchase. In addition, different manufacturers have different terminology for sequences, magnet homogeneity, and other physical features that make it difficult to perform an “apples to apples” comparison. However, rarely does musculoskeletal imaging come into consideration when purchasing a scanner. Typically, neurologic and body imaging applications are those that guide scanner selection and purchase, and musculoskeletal imaging is a secondary consideration. A basic understanding of MR physics principles is generally all that is needed to optimize musculoskeletal imaging protocols, no matter what machine is used. The differences between high-field and low-field scanners are extremely important in musculoskeletal imaging; advantages and disadvantages are summarized in Table 1-1.

Table 1-1 Low-Field MRI: Advantages and Disadvantages

Low-field scanners, that is, lower than 0.7 tesla (T), have difficulty performing standard presaturation-type frequency selective fat suppression because the signal peaks for the protons in water and fat are closely approximated. This is very important because musculoskeletal radiologists generally prefer to apply fat suppression on T2-weighted images to highlight fluid and edema. In addition, because fat is ubiquitous in the musculoskeletal system, fat suppression is preferred when gadolinium is injected intravenously or into a joint. As such, this can be a significant limitation of low-field scanners. Instead of T2-weighted fast spin-echo fat-suppressed imaging, which is a standard in musculoskeletal radiology, short tau inversion recovery (STIR) imaging is used to achieve fluid conspicuity, but at the expense of lowered resolution or a larger field of view. Alternatively, fluid conspicuity can be achieved without fat suppression by increasing echo time (TE), about 100 to 120 msec. However, this is at the expense of the SNR, which decreases as TE is raised. This creates difficulty in visualizing small cartilage lesions. One should not expect to consistently see small cartilage defects on a low-field scanner.

Without fat suppression on T1-weighted images, visualization of intravenously or intra-articularly administered contrast can be difficult, limiting applications such as tumor/infection imaging and MR arthrography. Also, there is relatively lower signal overall. To compensate, this requires increasing the number of signal averages to increase signal at the expense of increased imaging time, which can increase motion artifact in the extremities. Length of examinations is generally much greater on low-field scanners and may not be as useful for uncooperative patients. Coil options are often limited on low-field scanners, with a small variety of multipurpose surface coils available for imaging various musculoskeletal structures. This can result in an inappropriately large field of view and low resolution.

Nevertheless, there are some advantages to musculoskeletal imaging on low-field scanners. T1 contrast is actually superior to that of high-field scanners, although this is a relatively minor advantage in practice; in fact, improved T1 contrast can be disadvantageous. Consider evaluation of the knee on a low-field system; on a T1-weighted sequence fluid may appear black, blending with signal of the menisci (Fig. 1-1). Because of lower field strength, artifact from metal may be decreased compared with high-field scanners, and imaging patients who have prostheses, screws, or other orthopedic hardware can actually be improved by directing these patients to a low-field scanner. Keep in mind that advances in gradient technology at high fields have offset many of these advantages for high-field scanners. Chemical shift artifact is also decreased at low field.

Figure 1-1 0.3 T MRI of the knee. Low-field scanners have better T1 contrast than high-field units. However, this can have a detrimental effect on joint imaging.

Moreover—and what is possibly the most important consideration—is the gantry size and table weight limit of low-field scanners, which generally offer an open environment and a weight limit of up to 500 pounds. For obese patients, no other imaging options may be available. This is mainly an issue in the United States. Regarding fat suppression, many low-field scanners offer a software option based on the Dixon technique, which acquires an in-phase and out-of-phase image, and through subtraction post-processing, obtains a fat-suppressed image (Fig. 1-2). If the subtraction is performed from images acquired in the same series no subtraction errors occur, and the images and degree of fat suppression are generally excellent. In fact, if the radiologist’s intention is to perform MR arthrography on a low-field scanner, strong consideration should be given to acquiring this post-processing software. Newer generations of low-field scanners can actually separate the fat and water peaks, performing true fat saturation, but this has been suboptimal compared with the Dixon technique.

Figure 1-2 MR arthrogram on a 0.3 T system. Dixon technique for fat-water separation. An in-phase and out-of-phase image is acquired (useful for marrow evaluation) as well as a fat-suppressed and a water-suppressed image. The fat-suppressed image is ideal for use in MR arthrography on low-field scanners. However, as a gradient-echo sequence, it is prone to susceptibility artifact from metal, air, blood products, and calcium.

Extremity scanners are also available, which are generally low field at about 0.2 T; although they are low cost and provide a high degree of patient comfort, low image quality corresponds to the low strength. Also, the narrow bore of the magnet limits scanning to wrists/hands, elbows, ankles/feet and knees. However, a 1.0 T extremity scanner is available that provides good image quality comparable to that achieved with closed 1.5 T scanners (Fig. 1-3). Other open configuration 1.0 T scanners are also available.

Figure 1-3 Coronal gradient-recalled echo image of the wrist acquired with a 1.0-T dedicated extremity scanner.

(Courtesy of ONI and Joel Newman, Boston, MA.)

Advantages and Disadvantages of 3 T Scanners for Musculoskeletal Imaging

There are some distinct advantages of 3 T MRI for musculoskeletal imaging. The high field strength provides high SNR over all imaging sequences, allowing an increase in the matrix, a decrease in slice thickness, and a decrease in field of view, providing increased resolution; ultimately, high resolution is a major key to success in musculoskeletal radiology (Fig. 1-4). Alternatively, one can use the signal surplus to decrease number of excitations (NEX), thereby shortening examination time and increasing patient throughput. However, protocols must be altered somewhat to account for the different physical properties of the 3 T environment. For example, the high field strength at 3 T accentuates susceptibility artifact from metal, limiting evaluation of orthopedic hardware. In addition, chemical shift artifacts may be increased, resulting in black-white effect at fat–water interfaces. This can be eliminated if fat suppression is used; alternatively, one can make the pixels smaller to reduce this artifact by increasing matrix or decreasing the field of view. Manipulation of the receiver bandwidth is a third strategy that can be used to address the problem.

Figure 1-4 High-resolution 3 T imaging of the thumb using a small solonoid coil. Improved knowledge of anatomic detail and microstructural pathology will be required once this degree of resolution is routinely achieved.

(Courtesy of Ivan Dimitrov, Best, the Netherlands, Philips.)

Coils

Although an understanding of MR imaging physics and basic sequence optimization is essential, purchasing a wide range of coils is the easiest way to optimize musculoskeletal imaging (Fig. 1-5), regardless of scanner type. Unlike coils designed for the abdomen, head, and spine, there is a tremendous degree of variation in configuration and quality of various surface coils used for extremities. A rule of thumb is that the smallest coil possible should be used for imaging an extremity to achieve the desired field of view (Box 1-1). Basically, the coil is analogous to an ear trying to hear a faint noise coming from the body part being imaged. The farther the ear is from the body part, the softer the noise (i.e., weaker the signal) will be. Therefore, the more closely the coil fits against the body part, the better the images will be. It follows that an extremity coil used for the knee will not be as optimal for imaging the ankle, which is smaller and not cylindrically shaped.

Figure 1-5 Optimizing coil selection and placement is a major factor in optimizing image quality. This knee of an obese patient was imaged on a 0.3 T scanner; the knee did not fit in the extremity coil. The flex coil (left) yielded unacceptable signal-to-noise ratio (SNR). Use of a quadrature cervical coil (right) improved overall signal. However, note asymmetric distribution of signal on the sagittal image.

BOX 1-1 COIL BASICS

Try to use smallest coil possible, closest to area of concern (maximize signal)

Prefer phased array coil

Prefer send/receive coil

Small parts: Try taping 3-inch temporomandibular joint coils or 5-inch surface coil to the area

Coil must be positioned correctly! (Look at scout to check)

If patient cannot tolerate position, it may be best to go with second option to avoid motion

One common mistake is to attempt to image both ankles at the same time within a head coil, which provides inadequate imaging for both ankles. A similar error is to attempt to image the foot and ankle at the same time without moving the coil (although newer coil designs have a chimney for the forefoot and provide satisfactory images). A send-receive coil is preferred to a receive-only coil; a receive-only coil requires the body coil to send the radiofrequency signal, so a larger area is exposed to the energy, which can create issues of specific absorption rate (SAR) (body energy deposition) and also aliasing artifact. Quadrature or phased-array coils are also preferred, as are multichannel coils. When a particular coil does not fit a body part, a combiner box can be used to link two coils positioned around the anatomy, but this partitions signal from each coil such that signal returning from each coil is reduced. This can be a limitation if the body part being imaged is thick, leading to signal drop-off in the center of the image.

Occasionally, patient limitations require an alternate coil configuration. For example, it is optimal to image the elbow above the head in an extremity coil, but this position may be difficult for patients to achieve, particularly those who have shoulder problems. A reasonable alternative would be imaging the arm at the side wrapped in a flex coil; at least the patient will be able to tolerate the examination with less motion. Similarly, some newer shoulder coil configurations are fitted to shoulder shape, but this can also create drawbacks. A shoulder coil design that is cuplike may not fit all patients, resulting in field cut-off (Fig. 1-6). This may also not allow positional imaging, such as the abduction–external rotation (ABER) position.

Figure 1-6 Optimizing coil placement is also essential. Here, the shoulder coil is malpositioned, leading to dropoff of signal medially; the labrum is barely visible. Coil position is easily gleaned from the scout image (right). If the image is inadequate, the technologist should reposition the coil or use a better-fitting coil.

Spatial Resolution versus Contrast—Which to Choose?

Sequences that offer high spatial resolution are useful for evaluating small structures, such as interosseous ligaments of the wrist and the labrum of the hip, or for evaluation of the articular cartilage. These sequences require high SNR, allowing the radiologist to decrease field of view and increase matrix size. Gradient-echo imaging (see Fig. 1-3) can provide high in-plane and out-of-plane resolution; in-plane resolution is related to SNR, field of view, and matrix (i.e., discrimination of adjacent structures in the X/Y plane), and out-of-plane resolution is related to slice thickness and interslice gap. Three-dimensional gradient sequences can be acquired at very narrow slice thickness, without a gap. Proton density imaging provides high SNR, since fat and fluid both have high signal (Fig. 1-7). This added signal can be used to decrease field of view and increase image matrix size, thus increasing spatial resolution (Fig. 1-8).

Figure 1-7 Sequence emphasizing spatial resolution. Proton density (intermediate-weighted) sequences are very useful in musculoskeletal imaging; high signal allows alteration of other parameters such as acquisition of thinner slices or higher matrix in order to achieve high resolution. Gradient-echo sequences can also be used to provide high resolution.

Figure 1-8 If the image has adequate signal, increasing the matrix and/or decreasing the field of view improves resolution. Note improved detail of trabeculae as well as soft tissue structures as the matrix is increased. However, as the voxel size decreases, signal-to-noise ratio decreases, so this must be titrated. FSE, fast spin-echo; PD, protein density; TE, echo time; TR, time to repetition.

Sequences with high contrast resolution provide a large difference in signal between the anatomic structures being imaged and the surrounding tissues. For example, a fat-suppressed T2-weighted image of the wrist has high contrast resolution, with fluid being very bright and the triangular fibrocartilage being very dark. STIR imaging has high contrast, but may be limited by lower spatial resolution. One way to increase contrast resolution is to perform MR arthrography using fat-suppressed T1-weighted images (Fig. 1-9), which, when performed with spin-echo or gradient-echo imaging, also can be optimized for spatial resolution. This is one advantage for performing MR arthrography for internal derangement. Although it may be desirable to combine high contrast and spatial resolution, often these cannot be achieved in the same sequence, and protocols should be set up with some sequences optimized for spatial resolution (high matrix, small field-of-view proton density for gradient-echo imaging) and others designed for high contrast (fat-suppressed T2-weighted imaging or STIR). Newer scanners provide an estimate of SNR when changing sequence parameters from the standard settings, allowing the technologist to alter those parameters to adjust SNR to a satisfactory level. This can take some of the guesswork out of protocol management.

Figure 1-9 High-contrast sequence. Intravenous contrast dye was administered with T1-weighted fat-suppressed spin-echo imaging performed after a 30-minute delay (indirect MR arthrogram [MRA]). High-signal joint fluid provides high contrast from the adjacent intermediate-signal hyaline cartilage and low-signal fibrocartilage.