Inflammatory and noninflammatory arthropathies are a common cause of disability in modern society. MRI offers a noninvasive means of assessing the degree of damage to ligaments, muscle, bone, and articular cartilage. The keys to high-quality musculoskeletal MRI are as follows: (1) knowledge of tissue types and local joint anatomy, (2) selection of appropriate radiofrequency (RF) coils for a given clinical problem, and (3) selection of optimum pulse sequences to generate image contrast between normal or abnormal structures. With modern MRI methodology, it is possible to have both high spatial resolution and exquisite image contrast.


In the imaging of any body area, knowledge of tissue types and anatomy is critical. Musculoskeletal structures are composed of bone (cortical and cancellous), bone marrow, cartilage, ligaments, tendons, muscles, vessels, and nerves. Because of different proton densities and relaxation times, these tissues are well delineated with many MR pulse sequences, and combinations of sequences are usually used to cover a spectrum of image contrast regimens.

In general, tissues with a high amount of collagen and low water density have short T2 relaxation times and appear dark on common MRI methods that use an echo time (TE) in the millisecond range. These tissues include cortical bone, the knee meniscus, the glenoid and acetabular labrum, and most ligaments and tendons. It is not unusual to see a slight amount of intermediate signal within these structures on short-echo TE pulse sequences; however, it is abnormal to see significant high signal with these structures on long TE (T2-weighted) sequences.

Arthritis causes predictable signal alterations on MRI. Inflammatory arthropathies cause bone marrow erosions and edema, which typically have long T2 relaxation times and enhance after administration of gadolinium. Synovial enhancement is also common in inflammatory disease but may also be seen in osteoarthritis. Degeneration of fibrocartilage, such as tendons and menisci, often appears as a T2 signal of intermediate intensity. Frank tearing or fissuring of structure appears as linear high T2 signal intensity on a background of short T2 or intermediate T2 tissue. To evaluate these changes, a mixture of contrast mechanisms is often helpful.

Protocols for evaluation of arthritis often include T1-weighted, intermediate-weighted, and T2-weighted images. Each of these weightings has strengths and weaknesses in the evaluation of certain joint structures. These different weightings are combined with fat suppression and contrast enhancement to give optimum images for each joint structure. Evaluation of a particular joint requires knowledge of the specific anatomy and disease process to design the best protocol to show abnormalities.


The fundamental tradeoffs in MRI involve image signal-to-noise ratio (SNR), acquisition time, contrast, and resolution ( Fig. 1-1 ). In designing protocols for imaging of arthritis, a compromise much be reached between spatial resolution and keeping the examination times reasonable. For example, the Osteoarthritis Initiative decided on a maximum of 60 minutes of examination time for both knees in this large multicenter study (Available at ). This resulted in compromises in the spatial resolution, acquisition plane, and contrast mechanisms used.

Figure 1-1.

The relationship between signal-to-noise ratio (SNR), voxel size, and scan time in MRI. The signal in an MRI acquisition is proportional to the main field strength (B0) and the size of the voxel in each dimension (delta x, delta y, and delta z). The SNR may be increased by averaging and increases with the square root of the acquisition time. SNR is also a function of the contrast mechanism chosen and the T1 and T2 of the tissues being imaged.

Protocols will vary on a particular MRI system, depending on field strength, coils, and imaging gradient hardware. Resolution that can be achieved depends greatly on the field strength, coil, and anatomy imaged ( Fig. 1-2 ). For example, a 3.0T system with modern coils and high-performance gradients will be able to image faster or with higher resolution than a 1.5T or lower field strength system.

Figure 1-2.

Example images showing the relationship between imaging time and signal-to-noise ratio (SNR). Two sagittal proton-density images are shown with identical scan parameters, except for the number of signal averages. A, Image acquired with one signal average has poor SNR. B, Same image acquired with four signal averages (4× the imaging time), with improvement in SNR by a factor of 2.

In terms of protocol selection, the field strength of the system will have a big impact on overall SNR. In general, a 3.0T system provides roughly twice the SNR of a 1.5T system; this means that images can be twice the resolution in one direction or up to four times faster and have the same SNR. Systems with stronger gradients will be able to image with longer echo train lengths with the same blurring, increasing imaging speed. Finally, good RF coils are essential at any field strength.


Selection of appropriate RF coils is the most important aspect of optimizing image quality in musculoskeletal MRI. The reason for this is that the signal seen by a coil is relatively constant, but the noise increases as the size of the coil increases. Hence, larger coils such as the body coil have a worse SNR than smaller coils such as a surface coil. So, in selecting the coil for a given body part or imaging area, one should select the smallest coil that will have sufficient field of view (FOV) and penetration for the area of interest.

Phased-array coils are combinations of surface coils that enlarge the effective FOV of the coil by combining the signal from many small coils (up to 32 coils). Because each small coil sees only a limited area of tissue, the SNR remains as high as when using a surface coil, but because the signals from all coils are combined to form the image, the FOV is larger than that from one single surface coil. These coils are useful for parallel imaging, which can be used to decrease imaging times or shorten echo train lengths ( Fig. 1-3 ). In general, the best SNR possible is achieved with a phased-array coil; however, if the area of interest is not deep within the body, high-quality images are achievable with a surface coil. General guidelines with descriptions of relative advantages and disadvantages of various coils are presented in the following sections.

Figure 1-3.

Parallel imaging. To speed up data acquisition, the object is placed in a phased-array coil with multiple elements. The object is then undersampled, leading to aliasing. The parallel imaging algorithm uses the coil sensitivity locations to create the final unaliased image.

(Courtesy of Brian Hargreaves, PhD.)

Body Coil

Although the body coil is commonly used as the RF transmitter, its use as a receiver is rarely ideal, because better local coils exist for almost all body parts. For large FOV survey imaging, such as initial identification of areas of signal abnormality in an area such as the bilateral thighs or calves, the body coil is a good choice.

Torso Phased-Array Coil

The torso phased-array coil is the ideal coil for imaging of the pelvis and hips and can in some circumstances be effectively used for more peripheral extremity imaging. Image SNR for this coil is considerably higher than that for the body coil, enabling improved spatial resolution. For patients with relatively normal abdominal girth, this coil has sufficient penetration of RF to provide uniform signal intensity in the hips. For extremity imaging, the torso coil can be used over the bilateral thighs, for example. Be aware that as the two elements of the coil get closer together, coil performance will diminish; similarly, as the coil becomes curved, imaging may degrade. This causes the coil’s performance to be somewhat unpredictable in areas such as the bilateral calves or when placed off center in a smaller patient for imaging structures such as an upper extremity. If the torso coil is chosen for larger FOV imaging of an upper extremity such as a humerus, the coil should not be folded around the sides. Also, the coil should be kept relatively well centered in the magnet and the patient shifted eccentrically if possible. If the torso coil is not working well, these applications are probably best done initially with body coil imaging, followed by targeted imaging using a more dedicated coil.

As a rule of thumb, at 1.5T, the torso coil is adequate for 512-frequency matrix imaging with FOVs larger than 20 to 24 cm. At this large FOV the voxel size should allow ample SNR for these images; however, when one gets an FOV below this, consider switching back to a 256-frequency matrix. At FOVs below 20 cm, it is often possible to use a smaller coil with better intrinsic SNR.

Extremity Coil

This is a workhorse coil for standard imaging of the knee, ankle, and foot. In addition, this coil is used for imaging of extremities such as the forearm, wrist, and hand when a larger FOV is required, provided the patient is capable of positioning the arm above the head. Many extremity coils are transmit/receive coils that are mounted in the center of the scanner gantry and cannot be moved off center. To use an extremity coil, it must be closed, so those patients with extremely large extremities have to be imaged in an alternative coil. Phased-array coils are also available for the knee, foot, and ankle. The maximum FOV coverage for the extremity coil is approximately 24 cm. The extremity coil is a reasonable choice for imaging of the elbow if a dedicated small extremity phased array is not available.

Wrist Coil

A dedicated wrist coil is optimal for imaging the hand and wrist and in some circumstances can be used for imaging the distal aspect of the foot. These are typically phased-array or quadrature receive-only coils that can be placed either in the center of the magnet or off center. In the off-center situation, the most challenging aspect is to get good fat saturation. Coils may be oriented either with the long side parallel or perpendicular to the table. Note, however, that the coil should not be rotated with respect to the B0 axis because image quality will markedly degrade. Also, when 3D acquisitions are desired, remember that oblique prescriptions may not be allowed, so that having the wrist either perfectly parallel or perpendicular to the floor is necessary.

The wrist coil is also appropriate for use in imaging of the fingers and toes. Especially useful is high-resolution imaging of tissues such as the first metatarsophalangeal joint for sesamoid injury. Alternatively, a small surface coil can be used for these areas, with somewhat less signal uniformity.

Shoulder Coil

Shoulder coils are usually limited to use in the shoulder. Two different configurations exist. One is a traditional loop-type design. This is a single-element receive-only coil. The critical factor with this coil is that it needs to be positioned as perpendicularly to the B0 field as possible. Any flux from the main field traveling through the coil in the z-direction will decrease SNR. More recently, high-quality phased-array shoulder coils that have a fixed orientation relative to the scanner have become available, thus removing a variable that might degrade image quality.

Small Extremity Phased-Array Coil

This is a miniature version of an extremity coil that is ideal for imaging the elbow or for smaller extremities such as children’s knees or ankles. This coil can also be placed off center in the magnet, which distinguishes it from the routine extremity coil.

General Purpose Surface Coils

Three-inch or 5-inch surface coils can be used alone or in pairs using a phased-array adapter. Such coils can be useful for imaging toes if they do not fit into the wrist coil. These coils are also useful for supplementary high-resolution imaging of joints. In general, superior dedicated coils exist for most situations.

Flex Coil

Flex coils can be used to image the elbow at the patient’s side in the absence of a dedicated coil. Currently, most of the previous applications of the flex coil are better performed with more dedicated coils.


Pulse Sequence Selection

One of the major advantages of MRI is the ability to manipulate contrast to highlight tissue types. The common contrast mechanisms used in MRI are T1-weighted, proton-density (PD), and T2-weighted imaging ( Figs. 1-4 and 1-5 ). The appearance of these has changed over time with the introduction of fast or turbo spin-echo imaging and the use of fat saturation.

Figure 1-4.

Image contrast generated by changing the repetition time (TR). The image on the left has a short TR, and cartilage and muscle ( dashed line ) are relatively bright compared with joint fluid ( solid line, arrows ). The image on the right is acquired with a long TR, allowing recovery of signal from joint fluid ( arrows ). The joint fluid is somewhat brighter than cartilage in this image owing to higher proton density (PD).

(Courtesy of Brian Hargreaves, PhD.)

Figure 1-5.

Image contrast generated by changing the echo time (TE). The image on the left has a short TE ( dotted line ), and the image on the right, a long TE ( dashed line ). In the short echo-time proton-density (PD) image, cartilage ( arrow ) and muscle are relatively close in signal compared with joint fluid. In the long echo-time T2-weighted image, the cartilage ( arrow ) and muscle are decreased in signal relative to joint fluid owing to shorter T2 relaxation times.

(Courtesy of Brian Hargreaves, PhD.)

Spin-Echo and Fast Spin-Echo Imaging

T1- and T2-weighted spin-echo MRI allows depiction of joint structures and can demonstrate defects and gross morphologic changes. However, T1-weighted imaging does not show significant contrast between joint effusion and the cartilage surface, making surface irregularities difficult to detect. Fat suppression adds dynamic range and reduces the effects of chemical shift artifacts. T2-weighted conventional spin-echo imaging has largely been replaced by faster techniques.

Fast spin-echo (FSE) or turbo spin-echo (TSE) imaging is a technique that uses multiple echoes per repetition time (TR) to acquire data faster than conventional spin-echo imaging. Proton-density FSE (PD FSE) has a short TE, and the higher SNR echoes are placed at the center of k-space. T2-weighted FSE, with a longer TE, has the higher SNR echoes placed at the edges of k-space. One disadvantage of PD FSE is blurring of short-T2 species owing to acquisition of high spatial frequencies late in the echo train ( Fig. 1-6 ). The appropriate echo train length using FSE requires some knowledge of details about the given scanner in question. In general, increasing the echo train length decreases the imaging time, but the edge blurring effect increases in the images with short effective TE ( Fig. 1-7 ). In longer TE (i.e., T2-weighted) sequences, there is actually an edge enhancement effect that occurs with FSE or TSE. The amount of blurring that occurs in the short-TE scans is a function of both the echo train length and inter-echo spacing, which depends on gradient hardware.


Full access? Get Clinical Tree

Get Clinical Tree app for offline access