T_1ρ MR Imaging of Human Musculoskeletal System

Abstract

Magnetic resonance imaging (MRI) offers the direct visualization of human musculoskeletal (MSK) system, especially all diarthrodial tissues including cartilage, bone, menisci, ligaments, tendon, hip, synovium etc. Conventional MR imaging techniques based on T₁- and T₂-weighted, proton density (PD) contrast are inconclusive in quantifying early biochemically degenerative changes in MSK system in general and articular cartilage in particular. In recent years, quantitative MR parameter mapping techniques have been used to quantify the biochemical changes in articular cartilage with a special emphasis on evaluating joint injury, cartilage degeneration, and soft tissue repair. In this article, we will focus on cartilage biochemical composition, basic principles of T_1ρ MR imaging, implementation of T_1ρ pulse sequences, biochemical validation, and summarize the potential applications of T_1ρ MR imaging technique in MSK diseases including osteoarthritis (OA), anterior cruciate ligament (ACL) injury, and knee joint repair. Finally, we will also review the potential advantages, challenges, and future prospects of T_1ρ MR imaging for widespread clinical translation.

Introduction

More than half of the individuals aged over 65 years show symptoms of osteoarthritis (OA) (1, 2). OA severely affects the quality of life of elderly people and is increasingly regarded as a metabolically active joint complaint of multi-factorial etiologies (3, 4).

Normal articular cartilage is composed of a small portion of chondrocytes, which are dedicated cells, within a large extracellular matrix (ECM). Cartilage ECM is mainly composed of water (60%–80%), type II collagen (15%–20%) and negatively charged proteoglycans (PGs) (3%–10%) that facilitate the binding of water molecules. The structure of cartilage varies across its depth and consists of four histological zones or layers. These are the superficial (tangential) layer, the transitional (intermediate) layer, the radial (deep) layer, and the calcified layer. Underneath these layers are subchondral bone and cancellous bone, respectively (5, 6). It is generally recognized that the collagen fibrils are oriented perpendicular to the articular surface in the radial layer, randomly distributed in the transitional layer, and parallel to the articular cartilage surface in the superficial layer (6). Articular cartilage contains a high concentration of PGs which are complex macromolecules consisting of protein and polysaccharides (6, 7). Aggrecan, a high-molecular-weight PG, making up 80–90% of the total cartilage (dry weight), is comprised of a bottle brush-shaped protein core with a long extended domain to which many glycosaminoglycan (GAG) side chains are attached. The distribution of the PG macromolecules within the collagen fibre network contributes to the resistance of the matrix to the passage of water molecules and therefore affects the mechanics of the cartilage in this fashion (5, 7). Many different types of collagen are found in cartilage, but the most common is type II collagen (90–95% of the total volume of collagen in cartilage). The main role of the type II collagen fibers is to offer a tensile force opposing the tendency of the PGs to expand the cartilage and serve to immobilize the PGs as well. The concentration of collagen and PGs varies with the depth from the cartilage surface to the radial layer. The PG concentration at the cartilage surface is at its lowest, rising to a broad peak in the middle layers (7, 8). Water is the most abundant component of cartilage with the concentration ranging from 80% of the gross mass on the surface to 65% in the deep layer. Only a small fraction of the water is bound to collagen molecules (5, 7).

A high PG content is necessary to the load-bearing requirements of joints for it confer low hydraulic permeability and high swelling pressure upon the cartilage tissue. Cartilage degeneration with a loss of PG from the ECM is thought to be an initiating event of early OA, and the fatigue and breakdown of collagen fibre network may be the earliest event in a degenerative chain of OA process (7, 8, 10). The biochemical changes of the degenerative cartilage generally include the reduction of PG concentration, the elevation of water content, and the increased rate of synthesis and degradation of matrix macromolecules (5, 10). All these changes in the macromolecular matrix lead to alteration of the mechanical properties of the cartilage, which consequently becomes unable to serve as an effective load-bearing matrix.

Over the last two decades, there have been endeavors in developing new techniques for the treatment of OA such as chondroprotective drugs and repairs of the cartilage defects through autologous chondrocyte implantation (ACI) or allograft with subsequent regeneration of the damaged cartilage (11–14). Most recently, the clinical outcomes of the particulated juvenile articular cartilage (PJAC) technique showed a significant improvement over baseline, with histologically favorable repair tissue 2 years postoperatively (15). Matrix-associated autologous chondrocyte transplantation (MACT) has become an established articular cartilage repair technique which provides good short-term and midterm results (16). At long-term follow-up, microfracture technique (MF) and osteochondral autologous transplantation (OAT) mosaicplasty have no significant differences in treating focal chondral lesions of the knee (17). The cell seeded collagen membrane method was used to treat the localized cartilage defects (18). A novel functionalised hydrogel with an integrated drug delivery system was developed to stimulate articular cartilage regeneration (19). The hyaluronic acid-salmon calcitonin conjugate was used for the local treatment of osteoarthritis and had chondro-protective effect in a rabbit model of early OA (20). Although the development of drugs in animals and humans is promising in preventing the cartilage macromolecules from breakdown and effectively halting further progression of OA, the long natural history of OA (approximately 10–20 years in humans) validating the efficacy of these drugs requires a noninvasive technique directly evaluating their effect on molecular changes in association with early stages of cartilage degeneration prior to the emergence of distinct morphological changes (13, 21, 22). The reliable and non-invasive measurements of molecular (PGs, collagen, and water) and morphological (joint space width, tissue surface area and volume) changes in cartilage enable the detection of OA in its early stages. These measurements can monitor the disease progression, evaluate the potential scenarios for disease management, and validate the efficacy of the disease-modifying drugs (13).

In current clinical settings, the diagnostic imaging techniques available to examine the progression of cartilage loss in OA mainly include plain radiographs, microfocal radiographs, radionuclide imaging, arthrography, computed tomography (CT), ultrasound, arthroscopy, and MR imaging of articular cartilage. Arthroscopy is an invasive method that can cause discomfort to patients, although it is a proclaimed gold standard of joint examination. Plain and microfocal radiographs are the simplest means of joint evaluation. Radiography along with radionuclide scans, arthrography, CT, and ultrasound are limited in their clinical use because they cannot detect early cartilage abnormalities. In contrast, MR imaging (MRI) has the advantages of multi-planar imaging, excellent soft tissue contrast, non-invasiveness, and no ionizing radiation (22). The role of MR imaging of cartilage is evolving and articular cartilage can be evaluated by various MRI techniques. In recent years, new MR imaging techniques of being able to detect changes in the ECM have been developed, among which T_1ρ mapping is most sensitive to the changes of PG content in the ECM compared to proton-based methods such as T₂ relaxation mapping, delayed gadolinium-enhanced MR imaging of cartilage (dGEMRIC), chemical exchange-dependent saturation transfer (gagCEST) (23), and the direct MR imaging of sodium (5, 9). T₂ mapping is effective primarily in quantifying changes associated with collagen component of the cartilage ECM. Sodium MR imaging of cartilage and dGEMRIC have been shown to be useful mainly in quantifying changes of PG content (4, 24). However, dGEMRIC is invasive and patients have to wait for a relatively long time before they are scanned; Sodium MR of cartilage has the drawbacks of low image contrast and resolution and the need of extra hardware of a sodium coil. T_1ρ imaging technique has the comparative advantage of direct visualization of PG content changes in cartilage non-invasively without the addition of any extra hardware provided that the limit of specific absorption rate (SAR) is satisfied (25, 26).

In the following sections of this article, we will review the basic principles of T_1ρ MR imaging, implementation of different T_1ρ pulse sequences, biochemical validation and outline the potential applications of T_1ρ MR imaging technique in MSK diseases such as OA, anterior cruciate ligament (ACL) injury, knee joint repair etc. We will summarize the potential advantages, challenges, and future prospects of T_1ρ MR imaging for widespread clinical translation as well.

Basic Principles of T_1ρ MR Imaging

Spin-lattice relaxation in the rotating frame is characterized by the time constant (T_1ρ) that defines the magnetic relaxation of spins under the influence of a radio-frequency field (27–30). A spin-locking (SL) experiment was described by Redfield (31), and it has been used extensively to investigate low-frequency interactions between macromolecules and bulk water (28, 30). T_1ρ MR imaging has been shown to be sensitive to changes in PG of cartilage (27). The T_1ρ relaxation is often measured using the spin-lock technique. T_1ρ is somewhat of a hybrid between T₁ and T₂. In fact, the boundaries for T_1ρ are always between T₂ and T₁. Normally, magnetization is rotated into the transverse plane along one axis, immediately followed by a spin-locking pulse applied along the same axis. The spin-locking pulse is an on-resonance and continuous wave radiofrequency pulse with long duration and low energy. Because the magnetization and radiofrequency field are along the same direction, the magnetization seems to be “spin-locked” provided the locking condition is satisfied, namely, instead of allowing the transverse magnetization to decay unperturbed, a low amplitude RF pulse is applied for a prolonged period of time to force the transverse magnetization to relax while under the influence of this applied pulse. If the RF pulse is appropriately aligned with the transverse magnetization, the magnetization no longer relaxes according to T₂, but instead relaxes according to T_1ρ. The corresponding RF pulse that accomplishes this is called a spin-locking pulse termed, say, B_SL as shown in Fig. 1. The T_1ρ relaxation parameter provides unique biomedical information in the low-frequency regime. The frequency range is generally from a few hundred hertz to a few kilohertz (32, 33).

Fig. 1.

Application of a spin-locking pulse in the rotating frame. Initially, longitudinal magnetization (a) is flipped into the transverse plane (b). A spin-locking pulse, which by definition is oriented along the magnetization, is immediately applied (c). After some time, the magnetization decays under the influence of the spin-locking pulse (d). The time constant of this relaxation is T_1ρ.

The amplitude of the spin-locking pulse, B_SL, and the duration of the spin-locking pulse or time of spin-lock length (TSL), can be arbitrarily chosen within the limits posed by the electronics of the MR system. In conventional T₁ relaxation, longitudinal magnetization relaxes by exchanging energy with the processes in the lattice that are occurring at the Larmor frequency. In the case of the spin-lock pulse, transverse magnetization can exchange energy with lattice processes occurring at the spin-locking frequency that is directly proportional to the amplitude of the spin-locking pulse (B_SL). Therefore, T_1ρ is spin-locking frequency-dependent similarly to the dependence of T₁ is dependent on the Larmor frequency (33). T₁ and T₂ that are intrinsic properties of tissue and are generally not affected by pulse sequence parameters. On the other hand, the value of T_1ρ is determined both by tissue properties and features of the applied spin-locking pulse. By changing the spin-locking pulse amplitude one can alter the spin-locking frequency and as a result influence the spin interaction with the lattice. Hence, T_1ρ can be sensitized to slow motion in the lattice, which is important for biomedical imaging, since slow motion in the lattice is associated with large molecules such as proteins. In fact, T_1ρ has been shown to be sensitive to the protein composition of tissue (32, 34), and may therefore offer information about properties of macromolecules in the tissue that is otherwise not possible when imaging using conventional T₁ and T₂ relaxation measurements. For a fixed spin-lock amplitude, T_1ρ values can be calculated by the acquisition of a series of T_1ρ -weighted images at various spin-locking pulse durations, termed time of spin-lock length (TSL). Although a wealth of literature concerning the T_1ρ applications in research circles (25, 27–30, 34–38) has been published, the T_1ρ relaxation mechanism in biological tissues, mostly in cartilage and neighboring meniscus, remains poorly understood. Further exploration to better understand this relaxation mechanism is warranted.

The History of T_1ρ Pulse Sequence development

Table. 1 summarizes the evolution of T_1ρ pulse sequnce development.

As shown in Figures 1 and 2, T_1ρ-weighted MR imaging sequences mainly consist of two steps: first a spin-locking pulse cluster is applied prior to imaging. After allowing adequate time for T_1ρ relaxation to occur and while maintaining this T_1ρ prepared magnetization along the Z-axis; a 2D or 3D data acquisition is performed. This means that the T_1ρ prepared magnetization can be imaged using any type of pulse sequences such as spin-echo (45), fast spin-echo (28, 46), single-shot fast spin-echo (33), 3D gradient-echo (32, 33, 39, 47), spiral imaging (41), and steady-state free precession (SSFP/True-FISP)-based sequences (43, 48, 49). Figure 3 shows the T_1ρ-weighted MR imaging pulse sequence diagram based on single-slice 2D fast spin-echo (FSE) sequence [Figure 3(a)] and 3D gradient echo sequence with a self-compensating spin-lock pulse cluster [Figure 3(b)], respectively. In the T_1ρ-weighted imaging sequence, a three-pulse cluster ((π/2)_x − (spin-lock)_y − (π/2)_−x − crusher gradient) prepares the T_1ρ-weighted magnetization. The initial (π/2)_x pulse flips the longitudinal magnetization into the transverse plane along the y-axis. A long low power pulse is then applied along the y-axis to spin-lock the magnetization. The second (π/2)_−x pulse flips this spin-locked magnetization back to the z-axis. Residual transverse magnetization is then dephased by a crusher gradient. The magnetization stored along the z-axis can then be read out by different pulse sequences such as fast spin-echo, single-shot fast spin-echo, and 3D gradient-echo (20, 26). These different pulse sequences were performed in the spin-lock pulse cluster in order to improve the spin-locking robustness to the B₀ and B₁ inhomogeneity. A composite spin-lock pulse showing the tolerance to shimming and radio frequency errors was designed to allow spin locking with relatively weak RF (low spin-lock frequency) and therefore low specific energy absorption rate (SAR) (50). A “self-compensated” spin-locking pulse (25, 26, 33, 51) with the phase of the second half of the spin-lock pulse being shifted 180° from the first half was proposed to reduce the artifacts due to the B₁ inhomogeneity. Application of spin-locking pulses in a system with species at a range of chemical shifts or in a system with significant B₀ inhomogeneity can result in more complex rotations arising from off-resonance spin-locking. In the off-resonance rotating frame, the “effective” spin-locking vector is not located in the transverse plane. Therefore, shifting the phase of the spin-locking RF by 180° does not result in an opposite orientation of the spin-locking vector. The resultant magnetization rotations and artifacts are not modeled by the theoretical discussion presented here, but the off-resonance effect becomes worse for larger resonance offsets (i.e., greater B₀ inhomogeneity or larger chemical shifts) or weaker spin-locking powers (i.e., lower B₁) [13]. T_1ρ quantification techniques can be performed on two-dimensional (2D) spin-echo, fast spin-echo, three-dimensional (3D) gradient echo (28, 32, 33, 39, 45–47), spiral imaging (35) or SSFP/True-FISP based sequences (43, 48, 49). The corresponding pulse sequences have been implemented on human subjects at both 1.5T and 3.0T on MR scanners by different manufacturers.

Fig. 2.

Schematic diagram for spin lock magnetization in the transverse plane (upper row). Initially, a π/2 pulse flips the longitudinal magnetization into the transverse plane. The horizontal rectangle pulse represents the spin lock (SL) pulse. And TSL and B₁ are the duration and amplitude, respectively. The dotted line represents the decay of the magnetization in the absence of spin-locking and is governed by time constant T₂*. If the magnetization is spin-locked, it decays according to T_1ρ (solid line) for the duration of time TSL. After implementation of the SL pulse the decay is again dictated by T₂*, during which time the signal may be acquired (right in lower row also); Radiofrequency pulse cluster for T_1ρ-encoding (left in lower row). TSL and B₁ are the duration and amplitude of the SL pulse, respectively.

Fig. 3.

(a) Schematic diagram of the pulse sequence for implementation of T_1ρ-weighted fast spin-echo (FSE) MR imaging. The magnetization undergoes T_1ρ relaxation during the time of spin locking, which generates T_1ρ weighting. A crusher gradient is used after the SL pulse cluster to dephase any residual transverse magnetization. The magnetization is then imaged with a standard two-dimensional fast spin-echo sequence. Only three echoes are shown for the sake of clarity. ¹H RF = proton radio frequency, SL = spin lock pulse. (b) The pulse sequence diagram for the GRE sequence with a self-compensating spin-lock pulse cluster. A strong crusher gradient is applied before the α pulse to destroy any residual magnetization in the transverse plane and prevent the information of unwanted coherences. The stacked lines in the gradient region indicate the phase-encoding gradients.

For a fixed spin-lock frequency, a series of T_1ρ-weighted images can be obtained with varying time of spin-lock length (TSL). The spin-lock frequency may be calibrated from the amplitude and length of the hard 90° pulse employed. The typical length of the hard 90° pulse is ~100–200 us. The spin-lock field varies from 0–1500 Hz (27, 32). 2D spin-echo or fast spin-echo (FSE) sequence-based T_1ρ-weighted imaging was first developed (30). The 2D T_1ρ maps can be constructed from a series of T_1ρ-weighted images acquired with various TSL values and with the image signal intensity data fit according to equation 1:

$$S(\mathit{TSL})=S_{0}exp\left(\frac{-\mathit{TSL}}{T_{1\rho }}\right)$$ (1)

Here, S(TSL) and S₀ are the signal intensity at a given TSL and at the shortest TSL, respectively. The image data are fitted on a pixel-by-pixel basis.

Wheaton and colleagues (40, 41) presented a multislice spin-lock (MS-SL) pulse sequence to acquire multiple images with spin-lock generated contrast that produces images with T_1ρ contrast with an additional factor of intrinsic T_2ρ weighting, which hinders direct measurement of T_1ρ. A T_2ρ compensation method was proposed to accurately calculate multislice T_1ρ maps in an acceptable experimental time in this work which demonstrated that the T_2ρ-compensated multislice T_1ρ maps produced errors in the measurement of T_1ρ in healthy patellar cartilage of roughly 5% compared to the so-called gold standard measurement of T_1ρ acquired with single-slice spin-lock pulse sequence (28). Li and co-workers (42) implemented a multi-slice T_1ρ imaging method using spin-lock techniques based on spiral imaging. The adverse effect of T₁ regrowth during the multi-slice acquisition was eliminated by RF phase cycling. Their research study demonstrated the feasibility of applying this imaging technique to in vivo knee cartilage T_1ρ mapping and quantification. In recent years, Du et al (44) reported a novel ultrashort echo time (UTE) T_1ρ sequence that combines a spin-lock preparation pulse with a two-dimensional ultrashort echo time sequence of a nominal echo time 8 μsec. The UTE-T_1ρ sequence was employed to quantify T_1ρ in short T₂ tissues including the Achilles tendon and the meniscus. Their preliminary results on the cadaveric ankle specimens and healthy controls show that this UTE T_1ρ sequence can provide high signal and contrast in imaging the Achilles tendon and the meniscus. Although several studies have demonstrated that 2D T_1ρ MR imaging is sensitive to changes in the proteoglycan content of cartilage, they are restricted to single-section imaging and are hence impractical as guides for T_1ρ imaging of a typical anatomic region (30). Applying single-section techniques is a consequence of the need to make the spin-locking pulse section selective. Compared to the 2D-T_1ρ-weighted imaging technique, the 3D-T_1ρ-weighted imaging method has several advantages. 3D imaging allows rapid acquisition of images of the entire anatomic region. 3D sequences generally have a thinner slice thickness and may provide more accurate evaluation of cartilage degeneration. High-resolution 3D MR imaging is particularly significant for the assessment of OA where cartilage becomes very thin, (i.e. less than 1 mm) (30, 32). In addition, by suppressing the fluid and fat signals, 3D-T_1ρ-weighted MR imaging can generate better demarcation in compared to 2D T_1ρ-weighted MR imaging between cartilage on the one hand and bone marrow (fat) and fluid on the other. 3D acquisition is desirable because of the non-slice-selective nature of the T_1ρ preparation pulses (spin-lock pulses) [32]. On the other hand, several studies (30, 48) have validated that there was very good agreement between T_1ρ values computed by 2D and 3D T_1ρ-weighted imaging methods.

It is impossible to reduce scan times below a certain point because of hardware and physiological limits stemming from the limits of the switching rates. Parallel imaging technique can provide solutions to this limitation. The basic principle of parallel imaging is to utilize the spatial sensitivity of each coil element of a phased-array (PA) coil for spatial encoding (28). Parallel imaging reduces the number of phase-coding lines that must be acquired in order to keep the same field of view (FOV). Therefore, the acquisition time can be reduced measurably (25, 26).

Biochemical T_1ρ Validation in Ex-vivo Model Systems

In vitro and ex-vivo studies have assessed the relationship between T_1ρ relaxation time and the biochemical composition of cartilage. As shown in Fig. 4, Akella and colleagues (4, 27) implemented T_1ρ mapping of healthy bovine patella as a function of PG depletion. Their work suggested that T_1ρ-mapping may be used to sensitively measure and spatially map the changes in the PG component of articular cartilage. Specifically, as the early stages of OA involve the loss of PG with only minor changes in collagen content, T_1ρ-mapping could potentially be beneficial in measuring and monitoring these early degenerative changes. Regatte and co-workers demonstrated that T_1ρ mapping is sensitive to sequential depletion of proteoglycan in bovine cartilage and may be exploited to sensitively measure and spatially map the changes in the PG component of articular cartilage (27, 37, 52–54) (Fig. 5). Their results suggest that T_1ρ has a higher dynamic range (>100%) for detecting early pathology compared to T₂ using human cartilage specimens for T₂- and T_1ρ-weighted imaging. They also implied that T_1ρ relaxation mapping is a sensitive noninvasive marker for quantitatively predicting and monitoring the status of macromolecules in early OA (53). Taylor et al (55) used human osteochondral specimens from OA patients and cadavers for T_1ρ, T₂, and dGEMRIC magntic resonance imaging and indicated that T_1ρ, T₂, and contrast enhanced techniques may provide complementary information about the molecular environment in cartilage during the evolution of OA. Koskinen and co-workers showed that T_1ρ relaxation time is longer in advanced cartilage degeneration than in intermediate degeneration via T_1ρ imaging of cadaveric patellae (56). Wheaton and colleagues (37, 57) investigated the correlation of changes in cartilage biomechanical and biochemical properties with T_1ρ relaxation rate in a cytokine-induced model of degeneration. By T_1ρ imaging of cultured bovine cartilage explants, they demonstrated that T_1ρ MRI can detect changes in proteoglycan content and biomechanical properties of cartilage in a physiologically relevant model of cartilage degeneration; the T_1ρ technique can potentially be used to noninvasively and quantitatively assess the biochemical and biomechanical characteristics of articular cartilage in humans during the progression of osteoarthritis. Duvvuri and co-workers demonstrated that T_1ρ relaxation changes are correlated with PG loss in vitro, and T_1ρ measurements alone can be used to indicate PG loss data (29, 58). Using T₂ and T_1ρ MR imaging of suspensions of cartilage macromolecules and bovine cartilage samples, Menezes and colleagues (59) illustrated that T₂ and T_1ρ are sensitive to biologically meaningful changes in cartilage, and these two relaxation mechanisms are not specific to any one inherent tissue parameter which is contrary to some published reports (60).

Fig. 4.

Comparison of representative T_1ρ maps of control and 40% PG depleted bovine patellae, control T_1ρ map (upper row of left column) and 40% PG depleted T_1ρ map (lower row of left column), respectively. The color bar scale shows the relaxation time values distribution ranging from 0 to 256 ms; The right column shows a plot of 1/T_1rho vs. PG loss from a group of bovine cartilage patellae subjected to serial depletion of PG content. The solid line indicates the linear fit to the experimental data. T_1ρ data demonstrated a strong correlation (R² = 0.90, p < 0.0001) between changes in PG content and 1/T_1rho. Figure reprinted from reference (4), with permission.

Fig. 5.

Representative T₂ (upper row of left column) and T_1ρ (lower row of left column) maps. T₂ map constructed from a series of T₂-weighted images obtained with parameters TR/TE = 4000 ms/15 ms, FOV = 7 cm × 7 cm, slice thickness = 3 mm, matrix = 512 × 128, NEX = 1. T_1ρ map constructed from a series of T_1ρ-weighted images of the same slice obtained with parameters TR/TE+TSL = 4000 ms/17 ms; FOV = 7 cm × 7 cm, slice thickness = 3 mm, matrix = 512 × 128, NEX = 1, and ω1 = 500 Hz. The color bar scale shows the relaxation time values distribution ranging from 0 ms to 100 ms; (right column) comparison of T₂ and T_1ρ relaxation times as a function of various clinical grades of OA cartilage (early OA, moderate OA, and advanced OA). Figure reprinted from reference (52), with permission.

In vivo MSK Applications

The interplay between motion-restricted water molecules and their local macromolecular environment can be monitored by measuring T_1ρ, making it a promising technique for assessing the composition of human MSK system like cartilage, since its ECM provides a motion-restricted environment for the water molecules. In the osteoarthritic knee, damaged hyaline cartilage demonstrates higher T_1ρ values than normal cartilage, and T_1ρ imaging has higher sensitivity than T₂-weighted imaging for differentiating between normal cartilage and early-stage osteoarthritis (36). There is evidence that several factors other than proteoglycan depletion may contribute to variations in T_1ρ values; these factors include collagen fiber orientation and concentration and the concentration of other macromolecules (60). T_1ρ imaging may provide valuable etiologic, diagnostic, or prognostic information regarding knee osteoarthritis (61).

T_1ρ Imaging of the Knee Joint

Lozano et al (62) described two cases in which cartilage injuries could not be detected with standard morphological magnetic resonance imaging but in which T_1ρ-weighted imaging was able to detect cartilage abnormalities in vivo that were confirmed at the time of arthroscopy. A recently published research study (63) suggested that there exists some degree of association between knee alignment and subregional T_1ρ values of femorotibial cartilage and menisci in patients with clinical OA. Wang et al (64) compared parallel changes of quantitative T₂, T_1ρ, and dGEMRIC mapping of human cartilage and suggested that T_1ρ and dGEMRIC mapping seem to be more sensitive in detecting early stage of cartilage degeneration. Goto and co-workers (65, 66) evaluated age- and weight-bearing-related changes of T_1ρ values in normal cartilage. Stahl et al (36) suggested that T_1ρ is well suited to differentiate healthy subjects and early OA patients and is more sensitive than T₂ relaxations times, yet T_1ρ is also dependent on age. Pakin et al (25, 26) demonstrated the feasibility of 3D T_1ρ-weighted imaging of the knee joint at 3.0T. Souza and co-workers (67) observed an association between cartilage defects and meniscal damage in advanced disease via T_1ρ MR imaging in knee OA subjects. Regatte et al (30, 34) and Duvvuri et al (28) demonstrated the feasibility of quantifying early biochemical changes in symptomatic OA subjects employing T_1ρ-weighted MRI on a clinical scanner. Nishioka and colleagues (9) suggested that using both T_1ρ and T₂ imaging methods could provide an evaluation of cartilage constituents, detect qualitative changes in the ECM, and quantitatively measure cartilage degeneration. Li et al (68–70) indicated that T_1ρ and T₂ values show different spatial distributions and may provide complementary information regarding cartilage degeneration in OA. Link and colleagues (71) suggested that dGEMRIC, T₂ and T_1ρ mapping techniques can detect cartilage damage at a stage when changes are potentially still reversible before cartilage tissue is lost. At the early stage of OA, an overall increase in cartilage parameters (T_1ρ, T₂) were observed in patients (72). Fig. 6 shows representative T_1ρ maps of femorotibial cartilage obtained from patients with mild (KL1), minimal (KL2), moderate (KL3), and advanced (KL4) OA, respectively.

Fig. 6.

Representative T_1ρ maps of femorotibial cartilage in the lateral (a, c, e, g) and medial (b, d, f, h) compartments obtained from patients with mild (KL1) (a, b), minimal (KL2) (c, d), moderate (KL3) (e, f), and advanced (g, h) (KL4) OA, respectively. The color bar scale on the right shows the T_1ρ values range of distribution. The imaging parameters for 3D-T_1ρ mapping were: duration of each 90° pulse = 200 μs, TR/TE = 175 ms/2.04 ms, spin-lock frequency = 300 Hz, number of slices = 30, time of spin-lock (TSL) = 2/10/20/30 ms, slice thickness = 3 mm, matrix = 256 × 128, FOV = 15 cm, flip angle = 25°, bandwidth = 260 Hz/pixel, respectively.

MR imaging of the menisci of the knee has proven useful for more than 10 years (73). Zarins et al (74) suggested that meniscal damage has been implicated in OA progression and is correlated with cartilage degeneration by way of MR T_1ρ and T₂ measurements. Rauscher and co-workers (75) demonstrated that meniscal T_1ρ and T₂ values correlate with clinical findings of OA and can be used to differentiate healthy subjects from patients with mild or severe OA. Son et al (76) indicated that both T_1ρ and T₂ relaxation times correlated strongly with water content and moderately with mechanical properties in osteoarthritic menisci, but not as strongly with glycosaminoglycans (GAGs) or collagen contents alone. They suggested that T_1ρ and T₂ relaxation times have limited ability to detect compositional variations in degenerate menisci. Bolbos and colleagues (77) demonstrated a strong injury-related association between meniscus and cartilage biochemical changes by quantitatively T_1ρ mapping of meniscus and cartilage matrix in patients with anterior cruciate ligament (ACL) injuries. Recent studies (38, 78) suggested that the compartment-specific damage of menisci may be associated with osteoarthritis and T_1ρ mapping could be sensitive to meniscus degeneration. Fig. 7 shows representative T_1ρ maps of menisci obtained from a healthy control, a minimal (KL2) OA patient, and a severe (KL4) OA patient, respectively.

Fig. 7.

Representative T_1ρ maps of menisci in the lateral (a, c, e, g, i, k) and medial (b, d, f, h, j, l) compartments, obtained from a healthy control (a, b, c, d), a minimal (KL2) OA patient (e, f, g, h), and a severe (KL4) OA patient (i, j, k, l), respectively. The color bar on the right shows the range of T_1ρ values. The imaging parameters for 3D-T_1ρ mapping were the same as those in Fig. 6.

ACL injuries occur mostly in individuals with early-onset of OA with associated pain, functional limitations, and decreased quality of life in the ages between 30 and 50 years (79–81). Clinical reports suggest that patients with ACL-deficient knees have an increased incidence of knee osteoarthritis (OA) compared to uninjured knees (82, 83). Nishioka et al demonstrated that T_1ρ values were significantly higher for cartilage pathology than for normal cartilage and that T_1ρ mapping technique was able to non–invasively detect cartilage pathology in the ACL–injured knee with a higher detectability than other sequences (84). Li and co-workers (77, 85) suggested that quantitative T_1ρ MR imaging can detect the changes of the cartilage matrix in ACL-reconstructed knees as early as 1 year after ACL reconstruction and can offer tools of quantitatively assessing meniscus and cartilage matrix in patients with ACL injuries. Fig. 8 shows representative T_1ρ maps of ACL-injured knee at baseline and 1-year follow-up, respectively.

Fig. 8.

T_1ρ maps of lateral (a, b) and medial (c, d) side of ACL-injured knee at (a, c) baseline and (b, d) 1-year follow-up. T_1ρ values in lateral posterior tibia (LT 3) were elevated significantly in ACL-injured knees at baseline and remained high at 1-year follow-up despite resolution of bone bruise in the LT. T_1ρ values in the contacting area of MFC and MT were significantly elevated in ACL-injured knees at 1-year follow-up. Figure reprinted from reference (85), with permission.

Aside from the applications in cartilage and meniscus, T_1ρ mapping offers tools to quantitatively evaluate bone marrow edema-like lesions (BMEL) in OA and knee injury (86). Du et al (44) hypothesized that ligament or tendon measurements may have useful pre- and post-operative clinical applications in the future. In their study, ultra-short TE T_1ρ imaging has been applied at Achilles tendon and showed increased values in one cadaver specimen with tendon degeneration. Quantitative T_1ρ relaxation time measurements are promising biomarkers for evaluation of preventive surgery and for knee repair surgery, which addresses traumatic or degenerative pathologies (87). T_1ρ mapping may also be used in postoperative imaging for outcome evaluation of different surgical procedures as well as individual outcome analysis and prediction, and early detection of degenerative changes before distinct morphological changes occur, although only a small number of studies has peen performed to date using T_1ρ imaging for evaluation of knee repair procedures (85, 88–97). Fig. 9 shows representative T_1ρ and T₂ maps 3–6 months and 1 year after mosaicplasty (MOS) surgery, respectively. Specifically, published literature (21, 64) has suggested that T_1ρ may provide complementary information in detecting cartilage degeneration when combined with both T₂ and (dGEMRIC)-T₁ mapping techniques, as is shown in Fig. 10. On the other hand, the reliability of cartilage T_1ρ and T₂ measurements has been limited to single-vendor studies, Mosher and colleagues (98) performed the reproducibility measurement of quantitative magnetic resonance (MR) imaging biomarkers of the morphology and composition (T_1ρ and T₂) of knee cartilage in a multi-center and multi-vendor trial involving patients with OA and asymptomatic control subjects and indicated that MR imaging measurements of cartilage morphology, T₂, and patellar T_1ρ demonstrated moderate to excellent reproducibility in a clinical trial network. One limitation exists that T_1ρ’s specificity to cartilage GAG introducing residual dipolar interactions leads to a contribution from the collagen matrix, which complicates the analysis and interpretation of T_1ρ.

Fig. 9.

Representative T_1ρ (left column) and T₂ (right column) maps 3–6 months and 1 year after mosaicplasty (MOS) surgery. Colour bar scale shows T_1ρ and T₂ values range of distribution. NC, normal cartilage; RT, regenerated tissue. Figure reprinted from reference (88), with permission.

Fig. 10.

Two representative sagittal T₂ (a, b) and T_1ρ (c, d) weighted images obtained from an OA subject along with overlaid maps respectively. The relaxation times for the T₂ and T_1ρ mapping are obvious in different range of values with a distinct higher dynamic range in T_1ρ for the OA subject. (e) The profile plots of the T₂ and T_1ρ relaxation times measured in the same subject as shown in (a) and (c). The vertical rectangular regions of interest (ROIs) were used for profile plotting (solid line for T₂ and dashed line for T_1ρ respectively). Each point on the profile is an average of 5×5 pixels and error bars show the respective standard deviation. Figure reprinted from reference (64), with permission.

Other Potential Applications

T_1ρ mapping offers the potential to identify biochemical damage to MSK system prior to the onset of irreversible tissue loss. As a promising biomarker, it may allow earlier detection of MSK pathology before the development of radiographic evidence of disease (99). Hip deformity such as acetabular dysplasia and cam as well as pincer deformities are thought to be a major cause of hip OA, and T_1ρ imaging has a clinically important role in diagnosing and staging chondral damage in the hip (100). Subburaj and others (101–103) analyzed regional variations of T_1ρ in hip joint cartilage and suggested that locally regional analysis was more sensitive than global measurements in subjects with and without femoral acetabular impingement. Fig. 11 shows a representative T_1ρ mapping example from a subject with mild hip OA (104).

Fig. 11.

T_1ρ map example of a subject with mild hip OA. Figure reprinted from reference (104), with permission.

The studies in T_1ρ imaging of the shoulder, wrist, and ankle are relatively sporadic (44, 51, 69, 105). La Rocca Vieira and co-workers (51) offered the baseline T_1ρ values of the humeral and glenoid cartilages of healthy subjects and suggested that this work may be useful for quantitative comparison with diseased shoulders as is shown in Fig. 12. Akella et al (105) demonstrated the feasibility of T_1ρ MR imaging of human wrist cartilage in vivo with standard clinical scanners.

Fig. 12.

Representative in vivo axial 3D T_1ρ-weighted images (a, b) and corresponding maps (c, d) of the shoulder (glenoid and humeral) cartilages in a 32-year-old male asymptomatic volunteer. Figure reprinted from reference (51), with permission.

Degenerative disc disease (DDD) of the intervertebral disc (IVD) is the most common cause of back-associated disability in adults across the world (106) as is shown in Fig. 13. Blumenkrantz et al (107) suggested that in vivo T_1ρ quantification of IVD is feasible and could become a clinical way of identifying early degenerative changes in the IVD. Auerbach and co-workers (108) suggested that T_1ρ in vivo has strong potential to be a non-invasive biomarker of proteoglycan loss and early disc degeneration. Results from Johannessen et al (109) suggested that T_1ρ may offer the tool of diagnosing early degenerative changes in the disc, and T_1ρ-weighted MR imaging may provide higher dynamic range than T₂. Likewise, Zuo and colleagues (110) indicated that MR T_1ρ relaxation time may potentially serve as biomarkers of symptomatic IVD disease.

Fig. 13.

The left image shows a color T_1ρ map overlaid on a T₁-weighted image (grayscale) of the lumbar region from a 22 year-old and a 59 year-old human cadaver spine. The right plot shows the average T_1ρ value of the nucleus pulposus (NP) region for each L2/L3 IVD against its specimen’s age. The standard deviation of all T_1ρ values within each NP region was graphed as the error bar. A linear fit of the data points yielded a correlation coefficient of 0.82, which suggests a strong relationship between T_1ρ of NP and the specimen’s age. Figure reprinted from reference (106), with permission.

Another application of T_1ρ imaging lies in the evaluation of muscle disease. Parkin et al (26) assessed T_1ρ mapping of periarticular muscles at the knee. Lamminen and co-workers (111) used T_1ρ dispersion for in vivo muscle tissue characterization and found definite statistically significant difference between the relative T_1ρ dispersion values of normal and diseased muscle tissue. Dixon and others (50, 112) concluded that spin locking is an effective way of increasing blood-to-myocardium contrast and suggested that T_1ρ can offer the same contrast to images as magnetization transfer and T₂. Virta et al (113) indicated that T_1ρ as well as T₂ is more sensitive to the composition of muscle in comparison with T₁ and that T_1ρ and T₂ offer a relatively similar tissue contrast.

Besides the extensive applications in human MSK system, T_1ρ MR spectroscopy and imaging also deal with the detection of tumors, liver and brain diseases, myocardium, blood flow, and Alzheimer’s disease (AD) (46, 50, 111, 114–130) as shown in Fig. 14 (liver), Fig. 15 (myocardium), and Fig. 16 (brain), respectively. Recent studies (119, 126) showed that T_1ρ MR imaging is able to detect liver fibrosis, and the degree of fibrosis is correlated with the degree of the T_1ρ measurements, suggesting liver T_1ρ quantification may play an important role for liver fibrosis early detection and grading. Compared to T₁ and T₂, T_1ρ apprars to have unique capability to distinguish tumor from normal fat and fibrous breast tissues (114). Markkola et al (116) evaluated the potential of T_1ρ dispersion and spin lock techniques to differentiate benign and malignant head and neck tumors and suggested that low T_1ρ dispersion effect values are characteristic of a benign tumor. T_1ρ-weighted imaging performed at low spin-lock strengths quanlitatively depicted tumor borders better than proton-density or T₂-weighted imaging and could be useful in treatment planning when combined with other imaging sequences (115, 117). Alzheimer’s disease is the most common form of dementia in the elderly. Borthakur and others (120–123) suggested that T_1ρ may potentially constitute an important biomarker of AD.

Fig. 14.

Segmented T_1ρ map of a 28-year old healthy control subject (a) and a 45-year old patient with liver cirrhosis Child-Pugh score A (b). T_1ρ value of the healthy subject was 49.1 ms and T_1ρ value of the patient with liver cirrhosis was 55.7 ms. Colour scale bar shows T_1ρ values range of distribution. Figure reprinted from reference (125), with permission.

Fig. 15.

T_1ρ relaxation time map from a single animal 8 weeks post-MI. The infarct ¹H nuclear relaxation time in vivo are measured T1ρ_infarct = 91.7 ms and T1ρ_myocardium = 47.2ms. Relaxation time are twice as long in the myocardial infarction than the myocardium, which means that T_1ρ-weighted images have high signal intensity at the infarct and low signal intensity in the myocardium. Figure reprinted from reference (132), Open Access article.

Fig. 16.

Images acquired at different TSL and TE values were used to calculate T_1ρ (a) and T₂ (b) maps, respectively. Note the greater dynamic range of the T_1ρ map than the T₂ map. The maps were windowed identically from 0–400 msec. Relaxation times greater than 400 msec are assigned the brightest color. Furthermore, pixels for which the linear fit’s correlation coefficient (R²) was less than 0.9 were set to zero, e.g., dark pixels inside the ventricles in the T₂ map. Figure reprinted from reference (119), with permission.

Future Prospects

T_1ρ-weighted imaging has been extensively used in the past decades to assess pathology of liver, muscle, breast, brain, and tumor (32, 33, 47, 115, 122, 129, 130, 132–134) besides those of human MSK system (9, 34, 36, 38, 44, 47, 53, 62–69, 72–103, 105–110, 124). These studies have demonstrated the potential values of T_1ρ imaging for evaluating various physiologic and pathologic states. Aside from the application in early detection of cartilage degeneration, the above-mentioned different T1ρ imaging applications provide evidence that T_1ρ is sensitive to slow spin motions in tissue, and with appropriately chosen pulse sequence parameters, T_1ρ may show properties of a tissue of interest that are not observable through T₁ and T₂ imaging. T_1ρ-weighted imaging has been applied to a number of disease states and organ systems and has shown promise as a tool for early detection of pathology.

The influencing factors for calculating T_1ρ values of MSK tissues like cartilage and meniscus are individual subject variation, regions of interest (ROIs) measured, water content, spin-lock power, pulse sequence parameters, and protocol settings etc. Up to now, despite the promising results in evaluating small cohorts (OA, ACL injury and cartilage repair), the value of individual T_1ρ measurements in a clinical setting requires further standardization of image acquisition and post-processing methods, improved spatial resolution, rapid mapping techniques, reproducibility, specificity and validation with reference standard, larger propsective multi-center trials on multi-vendor platforms. Longitudinal changes among healthy controls, OA subjects, and the natural effects of ageing on T_1ρ are not yet well evaluated, which is another limitation. Although there still exist other technical challenges for T_1ρ imaging such as high SAR (especially at high and ultra high magnetic field strength) and relatively long acquisition time, T_1ρ cartilage imaging has the obvious advantages of no contrast agent administration, no RF hardware modification, higher dynamic range compared T₂ and the ability to be implemented on any standard clinical scanners (1.5T or 3.0T). The prospect of clinical applications exploiting the information available through T1ρ-weighted imaging is sure to be promising and will be of greatly progressive.

Summary

This review mainly deals with T_1ρ MR imaging of human MSK system and summarizes the practical applications of T_1ρ MR imaging technique in MSK system. T_1ρ imaging in human MSK system can offer clinically significant information associated with biochemical variations in human tissues such as the cartilage of the knee joint. In vivo T_1ρ mapping technique has demonstrated its feasibility in a number of cross sectional OA studies and limited number of longitudinal and multi-center trials. In vivo T_1ρ mapping needs further cross validation with other imaging modalities such as dGEMRIC, sodium MRI, and gagCEST. Additional studies are warranted to address the issues of sensitivity, specificity, and reproducibility (multi-center and multi-vendor trials) to OA severity, progression as well as prediction of development of OA. T_1ρ imaging has extensive prospect and great potential to be a valuable clinically diagnostic measure for the early detection of human MSK pathology and is a promising quantitative biomarker for MSK applications.

Table 1.

History of T_1ρ Pulse Sequence Development

T1ρ Sequence	Publication
2D single-slice T1ρ-weighted fast spin-echo (FSE) sequence	Duvvuri et al. Radiology 2001; 220: 822–826. (Ref. 28)
3D fast gradient-echo (FGRE) T1ρ-weighted sequence	Regatte et al. Acad Radiol 2004; 11: 741–749. (Ref. 34); Borthakur et al. J Magn Reson Imaging 2003; 17: 730–736. (Ref. 39)
2D multi-slice spin-lock (MS-SL) pulse sequence	Wheaton et al. Magn Reson Med 2004; 52: 1453–1458 (Ref. 40); Wheaton et al. J Magn Reson Imaging 2004; 20: 519–525 (Ref. 41).
multi-slice spiral T1ρ-weighted imaging pulse sequence	Li et al. Magn Reson Med 2005; 54: 929–936 (Ref. 42).
3D T1ρ-prepared balanced gradient echo (b-GRE) pulse sequence	Witschey et al. J Magn Reson Imaging 2008; 28: 744–754 (Ref. 43).
Ultrashort TE T1ρ (UTE T1ρ) sequence	Du et al. Magn Reson Med 2010; 64: 834–842 (Ref. 44).