Several MR systems are commercially available for use in the assessment of OA in a clinical and research setting. Most widely applied are 1.5 T large-bore MR systems. The knee is usually imaged using a dedicated knee coil. The patient is supine as static images are acquired. Examination times vary depending on the purpose of the exam but usually last between 20 and 40 min including patient positioning. Novel coil technology such as commercially available 8-channel multi-array coils improves image quality for 1.5 T systems.

High-field MRI of 3 T was introduced for clinical application several years ago and ­experience has shown that these systems can be applied to knee OA in a clinical and research environment (Fig. 5.16). A higher signal-to-noise ratio is its definitive advantage, but disadvantages include increased artifacts, high costs, and limited commercial availability of coils at present. Due to their wide availability and reliable image quality, most OA studies that include MRI are using 1.5 MR systems. The OAI is one of the few large studies using a 3 T system.

The role of contrast-enhanced MRI (CE-MRI) in clinical and research settings remains to be fully established. Visualization of synovitis in OA is superior on contrast-enhanced scans using the intravenous paramagnetic agent gadolinium (Grainger et al. 2007) and recent studies have shown that CE-MRI-detected synovitis is associated with knee pain (Guermazi et al. 2011c).

Since different tissues are involved in OA and both morphologic and quantitative analyses are needed, a variety of different sequences have been developed for “whole-organ” assessment of OA. It is critically important to select the appropriate MR pulse sequences to study specific features in OA. In general, fluid-sensitive fat-suppressed sequences (e.g., T2-weighted, proton density-weighted, or intermediate-weighted spin-echo sequences) are useful to evaluate cartilage, bone marrow, ligaments, menisci, and tendons (Link 2009). It is of particular importance to use these sequences for the assessment of focal cartilage defects (Hayashi et al. 2010a) and bone marrow lesions (BMLs) (Crema et al. 2011a). Gradient-recalled echo (GRE)-type sequences (e.g., 3D spoiled gradient echo at steady state (SPGR), double echo steady state (DESS), and fast low-angle shot (FLASH)) are not suitable for marrow or focal-defect assessment as they are prone to susceptibility artifacts, which hinder the accurate interpretation of images (Peterfy et al. 2006). Also, GRE-type sequences usually require a fairly long imaging time, and motion artifacts can degrade image quality (Link 2009). A recent study demonstrated that focal cartilage lesions were larger and more conspicuous on the intermediate-weighted fast spin-echo sequence with fat suppression compared to the DESS sequence (Roemer et al. 2011a) (Fig. 5.16). It was also shown that GRE-type sequences demonstrate limited sensitivity to BMLs compared to fast spin-echo sequences (Hayashi et al. 2011b) (Fig. 5.16). For the assessment of synovitis, CE-MRI is preferable to non-CE-MRI, but if only non-CE-MRI is available, gradient echo-type sequences again should be avoided because they are prone to chemical shift artifacts, making accurate assessment of synovium difficult (Roemer et al. 2009b).

In contrast, GRE-type sequences provide high spatial resolution, excellent contrast of cartilage to subchondral bone and are well suited for quantitative measurement of volume and thickness (Eckstein et al. 2005, 2008; Link 2009). Clinicians embarking upon clinical OA research with MRI as an outcome parameter should consult experienced musculoskeletal radiologists for guidance on which pulse sequences are best suited to their study.

Semiquantitative whole-organ scoring has been applied to a multitude of OA studies (Figs. 5.17 and 5.18). Analyses based on semiquantitative scoring have added greatly to the understanding of the pathophysiology and natural history of OA as well as the clinical implications of structural changes. Examples are the associations of subchondral BMLs with cartilage loss in the same subregion cross-sectionally and longitudinally (Felson et al. 2007).

To date three semiquantitative scoring systems based on non-CE-MRI for the assessment of knee OA have been published and have been applied in epidemiological studies: the Whole-Organ Magnetic Resonance Imaging Score (WORMS) (Peterfy et al. 2004), the Knee Osteoarthritis Scoring System (KOSS) (Kornaat et al. 2005), and the Boston-Leeds Osteoarthritis Knee Score (BLOKS) (Hunter et al. 2008).

Inter- and intra-reader reliability results have been published for all three systems. However, one should be aware that these numbers are highly dependent on the specific MRI protocol used and on the experience of the individual readers. When estimating the time needed to apply any of these semiquantitative systems, the imaging protocol and the quality of images as well as the degree of joint abnormalities due to OA have to be taken into account. Lastly, reader experience and manual or electronic documentation of the scoring are very important. Useful MRI protocols for whole-organ assessment of knee OA have been published (Peterfy et al. 2006).

Recent studies compared the performance of the WORMS and BLOKS scoring systems (Felson et al. 2010; Lynch et al. 2010) both cross-sectionally and longitudinally, using a subcohort from the OAI. In this comparative study, neither system showed definite superiority over the other. Based on the strengths of both systems and to alleviate their weaknesses, a novel scoring system called the Magnetic Resonance Imaging Osteoarthritis Knee Score (MOAKS), was ­developed (Hunter et al. 2011). Notable changes include that “synovitis” and “effusion” were renamed “Hoffa synovitis” and “effusion synovitis,” as current data suggests that effusion assessment on water-sensitive sequences always reflects a combination of both synovial thickening and joint fluid (Roemer et al. 2009c, 2010b, 2011c).

Semiquantitative MRI scoring systems are also available for OA of the hip (Hip Osteoarthritis MRI Scoring System, HOAMS) (Roemer et al. 2011b) and the hand (Oslo Hand Osteoarthritis MRI score, OHOA-MRI) (Haugen et al. 2011). Although the use of such systems is so far not as common in the hip or the hand as the knee, OA researchers can now utilize these potentially useful tools for clinical or epidemiological studies involving hip and hand OA.

One of the promising aspects of MRI for OA research and clinical practice is its capability to assess biochemical properties of different joint ­tissues and thus to be very sensitive to early, pre-morphologic changes. The vast majority of studies applying compositional MRI have focused on cartilage, although there also seems to be a possible role for compositional techniques in the assessment of other tissues such as the meniscus or the ligaments (Krishnan et al. 2007; Rauscher et al. 2008; Li et al. 2011).

Possibly the most studied parameter for molecular imaging of cartilage is T2 mapping. T2 reflects the interactions between water molecules and surrounding macromolecules which are affected by the many physiologic and pathophysiologic processes within cartilage. Focal increase in T2 relaxation time has been associated with cartilage matrix damage, in particular with loss of collagen integrity and an increase in water content (Dunn et al. 2004; Menezes et al. 2004). In contrast, other data demonstrate unchanged or decreased T2 with in vitro degeneration (Li et al. 2007) and in clinical T2 images (Burstein 2006). These studies were limited by reporting of a bulk T2 value from heterogeneous areas of T2 in cartilage. Thus, more sophisticated T2 analysis tools, as well as larger and long-term studies, will be needed to better evaluate the overall diagnostic and prognostic power of T2.

A second parameter that has been investigated in depth for biochemical imaging is T1rho MRI, or T1 in the rotating frame. T1rho is similar to T2 in that it is sensitive to interactions of water with macromolecules. T1rho has been shown to correlate with the proteoglycan concentration in cartilage (Borthakur et al. 2006) and is also sensitive to collagen (Menezes et al. 2004). Although T1rho has been shown to have a larger dynamic range than T2 (Li et al. 2007), it is more complex to implement and is limited by radiofrequency power deposition. A recent study demonstrated thatT1rho and T2 values show different spatial distributions and may provide complementary information regarding cartilage degeneration in OA (Li et al. 2009). Current pulse sequence development is ongoing to enable more widespread implementation (Pakin et al. 2006), allowing larger and longer-term studies to better delineate its utility in the study of OA.

Sodium MRI and the dGEMRIC technique are designed to measure fixed charge density in cartilage, adapting established biochemical and histological approaches for measuring proteoglycan content (Gray et al. 2008) (Fig. 5.19). In the MRI implementation of this principle, the mobile ions are the naturally abundant sodium or the MRI contrast agent Gd(DTPA)²⁻ (Magnevist, Berlex, NJ).