Ultrashort TE T1rho (UTE T1rho) Imaging of the AchillesTendon and Meniscus

Jiang Du,1* Michael Carl,2 Eric Diaz,1 Atsushi Takahashi,2 Eric Han,2

Nikolaus M. Szeverenyi,1 Christine B. Chung,1 and Graeme M. Bydder1

In this study, we report the use of a novel ultrashort echo timeT1rhoT1 sequence that combines a spin-lock preparation pulsewith a two-dimensional ultrashort echo time sequence of anominal echo time 8 msec. The ultrashort echo time-T1rhosequence was employed to quantify T1rho in short T2 tissuesincluding the Achilles tendon and the meniscus. T1rho disper-sion was investigated by varying the spin-lock field strength.Preliminary results on six cadaveric ankle specimens and fivehealthy volunteers show that the ultrashort echo time-T1rhosequence provides high signal and contrast for both the Achil-les tendon and the meniscus. The mean T1rho of the Achillestendon ranged from 3.06 6 0.51 msec for healthy volunteers to5.22 6 0.58 msec for cadaveric specimens. T1rho increased to8.99 6 0.24 msec in one specimen with tendon degeneration. Amean T1rho of 7.98 6 1.43 msec was observed in the meniscusof the healthy volunteers. There was significant T1rho disper-sion in both the Achilles tendon and the meniscus. Mean T1rhoincreased from 2.06 6 0.23 to 7.85 6 0.74 msec in normalAchilles tendon and from 7.08 6 0.64 to 13.42 6 0.93 msec innormal meniscus when the spin-lock field was increased from250 to 1,000 Hz. Magn Reson Med 64:834–842, 2010. VC 2010Wiley-Liss, Inc.

Key words: ultrashort TE; short T2; T1rho; Achilles tendon;meniscus

Musculoskeletal diseases, particularly osteoarthritis(OA), comprise a set of debilitating conditions that affectmillions of people and are a tremendous burden onhealth care systems worldwide. MRI provides excellentsoft-tissue contrast and high spatial resolution and hasbeen used to evaluate OA. Unfortunately, conventionaltechniques such as T1-weighted, T2-weighted, and protondensity–weighted MRI can only detect the late stages ofOA, which are associated with gross morphological alter-ation (1,2). Techniques such as the delayed gadolinium-enhanced MRI of cartilage and T1rho imaging have beenproposed to detect the early stage of OA associated withproteoglycan (PG) depletion (3,4). Recent studies havedemonstrated that T1rho has high sensitivity to PG lossin bovine cartilage samples, as well as OA patients (5–7).As a result, T1rho imaging has emerged as a potentiallyimportant noninvasive method for detecting the bio-chemical changes that may precede morphologicaldegeneration in cartilage.

OA is a heterogeneous and multifactorial diseaseinvolving many tissues, although emphasis is usually onhyaline articular cartilage (8–14). T1rho imaging of thesetissues may provide a more systematic evaluation of OA(12–14). However, many of the tissues of the musculo-skeletal system, including the deep radial and calcifiedlayers of cartilage, meniscus, tendon, ligament, and sub-chondral bone, have T2s that are short and range fromsubmilliseconds to several milliseconds (9–11). Thesetissues show little or no signal with conventional MRsequences, which typically have echo times (TEs) of sev-eral milliseconds or longer. Studies on the involvementof meniscus in OA using delayed gadolinium-enhancedMRI of cartilage and T1rho imaging techniques based onclinical MR sequences may be limited by the fact thatthe meniscus shows a relatively low signal with sequen-ces with TEs of around 4 ms (9–11). Tendon, ligament,and the calcified layer of cartilage have even shorter T2sthan meniscus and are therefore more difficult to evalu-ate with delayed gadolinium-enhanced MRI of cartilageor current T1rho imaging techniques.

By using half-pulse excitations and radial mapping ofk-space, TEs less than 100 msec have been achieved withtwo-dimensional (2D) ultrashort TE (UTE) sequences(15–18). These pulse sequences make it possible todirectly image short T2 species, using clinical systems.In this study, we have combined a spin-lock preparationpulse with 2D UTE sequence with a minimal TE of 8msec to image and quantify T1rho values in short T2 tis-sues (19). This article describes the development of the2D UTE-T1rho sequence, as well as its use on a whole-body 3-T MR system in cadaveric tissue specimens andhealthy volunteers for imaging and quantifying T1rhoand T1rho dispersion of the Achilles tendon and themeniscus.

MATERIALS AND METHODS

Technical Considerations

All scans were performed on a 3-T Signa TwinSpeedscanner (GE Healthcare Technologies, Milwaukee, WI)with a peak gradient amplitude of 40 mT/m and a maxi-mum slew rate of 150 mT/(m/sec). A single-channeltransmit/receive quadrature knee coil was used. TheUTE-T1rho sequence combined a 2D UTE sequence witha spin-lock preparation pulse, as shown in Fig. 1. The2D UTE pulse sequence employed paired half-pulses forsignal excitation, followed immediately by radial rampsampling, with a TE delay of 8 msec (17–19). The projec-tion data were mapped onto a square grid, typically 512� 512, using a Kaiser-Bessel kernel and reconstructed by2D inverse Fourier transformation. The preceding spin-

1Department of Radiology, University of California, San Diego, California,USA.2Global Applied Science Laboratory, GE Healthcare, Menlo Park, California,USA.

Grant sponsor: GE Healthcare; VA 1136264.

*Correspondence to: Jiang Du, Ph.D., Department of Radiology, Universityof California, 200 West Arbor Drive, San Diego, CA 92103-8756. E-mail:jiangdu@ucsd.edu

Received 5 May 2009; revised 26 February 2010; accepted 7 March 2010.

DOI 10.1002/mrm.22474Published online 9 June 2010 in Wiley Online Library (wileyonlinelibrary.com).

Magnetic Resonance in Medicine 64:834–842 (2010)

VC 2010 Wiley-Liss, Inc. 834

lock pulse cluster consisted of a hard 90� pulse followedby a spin-lock pulse and a hard � 90� pulse (20,21). Thefirst 90� pulse was applied along the x-axis to flip thelongitudinal magnetization into the transverse planealong the y-axis. A composite pulse was then appliedalong the y-axis to spin lock the magnetization. The sec-ond 90� pulse flipped this spin-locked magnetizationback along the z-axis. Residual transverse magnetizationwas dephased with a crusher gradient. Magnetizationstored along the z-axis was then read out by the 2D UTEsequence. The T1rho of short T2 tissues such as theAchilles tendon and meniscus was quantified throughfitting UTE images acquired with a series of spin-locktimes (TSL).

There are three particular technical issues associatedwith UTE-T1rho imaging and quantification of short T2

tissues. These include (i) short T2 excitation, (i) eddycurrents, and (iii) T1 contamination. The approachesused to address these technical issues in this study arediscussed below.

Short T2 Excitation

Conventional MRI is largely focused on long T2 specieswhere radiofrequency (RF) pulse duration can be consid-ered negligibly small compared with tissue T2s. Short T2

tissues can have T2s of the same order as RF pulse dura-tions. As a result, short T2 tissues may experience signif-icant transverse relaxation during the excitation process,resulting in a lower flip angle than for long T2 species(15,22). The spin-lock preparation pulse cluster requiresa hard 90� pulse followed by a spin-lock pulse and ahard -90� pulse. Errors may be introduced if accurate 90�

and -90� magnetization rotations cannot be achieved forall short T2 species. This error can be reduced by usingthe maximal peak RF power available on the clinicalwhole-body MR system to achieve the shortest possibleduration 90� pulse. We used a 90� pulse with a durationof 256 msec, which was limited by the peak amplitude ofRF field of 24 mT but was significantly shorter than theT2 of Achilles tendon and meniscus. Bloch equation sim-

ulations were performed to evaluate the impact of usinga 256-msec hard 90� pulse for spin-lock preparation. Weused a Bloch equation simulation that calculates the pre-cession, as well as T1 and T*2 decay matrices, using thehard RF waveforms. The Bloch simulation was writtenin Matlab (The MathWorks, Natick, MA).

Eddy Currents

The 2D UTE sequence is based on radial ramp sampling,which is sensitive to eddy currents, gradient anisotropy,and timing errors, which may result in out-of-slice exci-tation and a broadened slice profile (16). Timing forslice-selection gradients and readout gradients can bemanually tuned to within an error of 62 msec by moni-toring image quality. Artifacts originating from theseerrors can be further reduced by empirically shifting theradial k-space trajectories during on-line image recon-struction. In this study, we used the UTE-T1rhosequence in single-slice mode for T1rho quantificationbecause slice crosstalk in multislice UTE imaging mighthave affected the quantification accuracy.

UTE-T1rho imaging employs a conventional 2D UTEsequence preceded by a spin-lock preparation pulse. TheTSLs are varied to detect the T1rho decay of the short T2

species. However, the delay time between each half-pulse and the radial ramp sampling was kept constant at8 msec. As a result, the gradient distortion and eddy cur-rents were kept the same when obtaining a series ofT1rho images with different TSL values (19).

T1 Contamination

T1 contamination is due to incomplete recovery of thelongitudinal magnetization due to the use of relativelyshort pulse repetition times (TRs) (21). This is a particu-lar problem with UTE-T1rho imaging because radial sam-pling requires a large number of projections to controlstreak artifacts. The need for two half-pulses (number ofexcitation, NEX ¼ 2) for selective excitation furtherincreases the acquisition time. As a result, the total scantime can be excessively long when long TRs (TR > 3 T1)are used for UTE-T1rho imaging and quantification.

Using shorter TR values is one way of reducing thetotal scan time. However, this approach may result insignificant T1 dependence. This can be dealt with bycompensating for the T1 effect on T1rho measurementsby using the following equation (23):

SðTSLÞ / e�TSL=T1rhoð1�e�ðTR�TSLÞ=T1Þ1�e�TSL=T1rhoe�ðTR�TSLÞ=T1 cos a

sin aþ Constant ½1�

TR is the time between a imaging pulses, and T1rho isthe spin-lattice relaxation time of the tissue of interest. Aconstant term is introduced to account for backgroundnoise and artifacts associated with UTE data acquisitionand image reconstruction. For short T2 tissues, thisapproach requires knowledge of the T1, which can beobtained with a saturation recovery sequence using ashort hard 90� pulse followed by a UTE acquisition witha series of saturation recovery times (14,24). T1 can thenbe quantified through exponential fitting of the recoverycurve.

FIG. 1. The spin-lock prepared UTE sequence. This combines aregular 2D UTE sequence (minimum TE ¼ 8 msec) with a spin-

lock preparation pulse consisting of a hard 90� pulse followed bya composite spin-lock pulse and another -90� hard pulse. The

phase of the second half of the composite spin-lock pulse isshifted 180� from the first half to reduce artifacts caused by am-plitude of RF field inhomogeneity.

UTE-T1rho Imaging 835

Phantom Study

UTE-T1rho imaging was first performed on a ball phan-tom containing CuSO4. This phantom was also studiedwith a published T1rho imaging sequence, namely,three-dimensional (3D), magnetization-prepared, angle-modulated, partitioned k-space, spoiled gradient echosnapshots (3D MAPSS) (21), and T1rho values obtainedwith the two techniques were compared. The phantomwas imaged five times on five different days. The repro-ducibility was estimated with coefficients of variationcalculated as the ratio of the standard deviation to theaverage UTE T1rho value.

UTE-T1rho Imaging

After obtaining written informed consent and approvalfrom the investigational review board of our institution,UTE T1rho imaging of the Achilles tendon and meniscuswas performed on five healthy male volunteers (ages 24–54 years, median ¼ 38 years). The UTE-T1rho imageswere acquired in the axial plane for the Achilles tendonand in an oblique sagittal plane for the meniscus. Chemi-cal shift–based fat suppression was applied with theUTE-T1rho acquisitions. Typical acquisition parameterswere as follows: field of view ¼ 12 cm for the Achillestendon and 16 cm for the meniscus, TR ¼ 400 msec, TE¼ 8 msec, bandwidth ¼ 662.5 kHz, flip angle ¼ 45�,readout ¼ 512, slice thickness ¼ 3 mm, 355 half-projec-tions, NEX ¼ 2, spin-lock field strength ¼ 500 Hz, TSL¼ 0.2 msec, scan time ¼ 4.7 min. For comparison, the3D MAPSS sequence was employed to image the Achil-les tendon and meniscus with similar imaging parame-ters, except a shorter TR of 14.1 msec, a longer TE of 2.4msec, and a readout matrix of 320 � 320 in a total scantime of 4.1 min. Signal-to-noise ratio (SNR) was meas-ured for the Achilles tendon and meniscus to evaluatethe efficiency of the UTE-T1rho and 3D MAPSS sequen-ces, respectively.

T1rho and T1rho Dispersion Quantification

The 2D UTE-T1rho sequence was performed on theAchilles tendon of six cadaveric ankle specimens (fourfemales and two males, ranging in age from 67 to 89years, with an average age of 76 years), as well as theAchilles tendon and meniscus of five volunteers to quan-tify T1rho and T1rho dispersion. Similar acquisition pa-rameters used in volunteer studies were used for thecadaveric ankle specimens, except with a larger numberof projections of 511 and longer TR of 500 msec. SixTSLs (0.2, 0.8, 1.5, 3, 6, and 12 msec) were used for thefirst cadaveric ankle specimen, and four TSLs (0.2, 2, 5,and 10 msec) were used for the other five ankle speci-mens to save scan time in quantifying T1rho and T1rhodispersion at three spin-lock field strengths (250, 500,and 1000 Hz) (25–27). Four TSLs (0.2, 1, 4, and 12 msec)were used for the Achilles tendon, and four TSLs (0.2, 1,5, and 25 msec) were used for the meniscus to quantifyT1rho in vivo. Longer TSLs were used for the meniscusbecause of its longer T*2 and T1rho values. Only threeTSLs (0.2, 5, and 20 msec) were used in quantifyingT1rho dispersion in vivo. For comparison, T*2 was also

measured for each subject, with UTE acquisitionsobtained with a series of different TE values. T1 values,which were required for calculating T1rho using Eq. 1,were measured for each subject, using saturation recov-ery UTE acquisitions with saturation recovery time/pro-jections of 10 msec/355, 50 msec/355, 100 msec/311, 200msec/255, 400 msec/211, 800 msec/165, and 1600 msec/165. The projections were progressively undersampled toreduce the total T1 quantification time to 20 min. All thevolunteers were scanned twice on two different days toexamine in vivo reproducibility using the UTE-T1rhosequence.

Image Analysis

SNR was measured for each TSL and was calculated asthe ratio of the mean signal intensity inside a user-drawnregion of interest to the standard deviation of the signalin region of interest placed in the background. T1rho val-ues were obtained using a Levenberg-Marquardt fittingalgorithm developed in-house, based on Eq. 1. The anal-ysis algorithm was written in Matlab (The MathWorks)and was executed offline on axial and sagittal imagesobtained with the protocols described above. The pro-gram allowed placement of regions of interest on the firstimage of the series, which were then copied to the corre-sponding position on each of the subsequent images.The mean intensity within each of the regions of interestwas used for subsequent curve fitting. Five differentregions of interest were fitted to determine the averageT1rho relaxation times for the phantom, the Achilles ten-don, and meniscus, using a single-component curve-fit-ting model. T*2 was derived through exponential fittingof the equation S(TE) ! exp(-TE/T*2) þ constant. T1 wasderived through exponential fitting of the equation S(sat-uration recovery time) ! 1 � exp(�saturation recoverytime/T1).

RESULTS

Simulations

The Bloch equation simulations show that the hard 90�

pulse with a duration of 256 msec results in a compara-tive signal loss of 1–2% for meniscus, with a T*2 of 4–6msec and 4–8% for tendons with a T*2 of 1–2 msec,when compared with tissues with infinitely long T*2. Asapparent from this trend, more signal loss is expected fortissues with even shorter T*2. Additional simulationsshow an 18% signal loss for cortical bone, with a T*2 ofaround 400 msec and a 90% signal loss for protein-boundprotons with a T*2 of around 10 msec (14), suggesting thatspin-lock preparation would fail for tissues withextremely short T*2.

Phantom Study

Figure 2 shows curve fitting of phantom data obtainedwith the UTE-T1rho and 3D MAPSS sequences. Bothtechniques provide excellent single-component curve fit-ting, which accounts for more than 99.9% of the signalvariance and showed a T1rho value of 63.7 6 3.9 msecwith UTE-T1rho and 66.3 6 2.7 msec with 3D MAPSS.The average coefficient of variation for the UTE T1rho

836 Du et al.

value of the phantom on five repeated acquisitions was2.3%. These results show that the UTE-T1rho sequenceprovides reliable T1rho quantification for long T2 values.The excellent curve fit in the short TSL range shown inthe zoomed subfigure suggests that this method may alsobe valid for short T2 values.

UTE-T1rho Imaging

Figure 3 shows 3D MAPSS imaging and UTE-T1rhoimaging of the Achilles tendon and meniscus of a younghealthy volunteer. The 3D MAPSS sequence providedlittle signal for the Achilles tendon with a measuredSNR of 5.41 6 0.73 and voxel size of 0.38 � 0.38 �3.0 mm3 and limited signal for the meniscus with anSNR of 21.84 6 3.89 and voxel size of 0.5 � 0.5 � 3.0mm3. In contrast, the UTE-T1rho sequence providedhigher signal and spatial resolution for the Achilles ten-don with an SNR of 24.17 6 3.21 and voxel size of 0.23� 0.23 � 3.0 mm3 and for the meniscus with an SNR of28.59 6 4.27 and voxel size of 0.31 � 0.31 � 3.0 mm3.

T1rho and T1rho Dispersion Quantification

Figure 4 shows T1rho imaging of the Achilles tendon ofa cadaveric ankle specimen. The Achilles tendon wasdepicted with high SNR, contrast, and spatial resolution.SNR ranged from 38.3 with a TSL of 0.2 msec to 8.5with a TSL of 12 msec. Significant T1rho dispersion wasobserved in the Achilles tendon, as shown in the firstthree rows. Figure 5 shows zoomed T1rho mapping, aswell as T*2, T1rho fitting, and T1rho dispersion. A T*2 of0.78 6 0.07 msec was demonstrated for the Achilles ten-don, using a single exponential component curve fit,which accounted for 99% of the signal variance. The sat-uration recovery UTE technique demonstrated a T1 of675 msec for the Achilles tendon (14,24). This was usedfor T1rho fitting using Eq. 1. T1rho for the Achilles ten-don was 4.95 6 0.23 msec with a spin-lock field of 500Hz. There was marked T1rho dispersion for the Achillestendon with T1rho increased from 2.19 6 0.12 msec witha spin-lock field of 250 Hz to 7.43 6 0.56 msec with a

spin-lock field of 1000 Hz, corresponding to an increaseof 239%.

Figure 6 shows UTE-T1rho images of an ankle speci-men with Achilles tendon with degenerative changes.SNR ranged from 34.7 with a TSL of 0.2 msec to 11.6with a TSL of 10 msec. T1rho increased from 5.93 6 0.38msec for normal tendon to 8.99 6 0.24 msec for degener-ated tendon. Figure 7 shows UTE-T1rho images of theAchilles tendon of a 53-year-old healthy male volunteer.

FIG. 2. T1rho quantification of a CuSO4 ball phantom imaged with a 2D UTE-T1rho sequence (a) and a 3D MAPSS sequence (b). TheT1rho values are consistent with 63.7 6 3.9 msec for the 2D UTE-T1rho sequence and 66.3 6 2.7 msec for the 3D MAPSS sequence. Azoom-in figure for short TSLs is also shown for UTE-T1rho fitting.

FIG. 3. 3D MAPSS imaging shows little signal for the Achilles ten-don (a) and limited signal for the meniscus (c) of a healthy young

volunteer. 2D UTE-T1rho imaging of the same volunteer showshigh signal for both the Achilles tendon (b) and meniscus (d).

UTE-T1rho Imaging 837

SNR values of 16.8 and 4.1 were obtained with TSL of0.2 and 12 msec, respectively. A T1rho value of 3.45 60.21 msec at a spin locking field strength of 500 Hz was

obtained for the Achilles tendon of this volunteer, usingsingle-component curve fitting that accounted for morethan 98% of the signal variance.

FIG. 4. UTE-T1rho imaging of the Achilles tendon of an ankle specimen with TSL values of 0.2, 0.8, 1.5, 3.0, 6.0, and 12 msec, usingspin-lock fields of 250, 500, and 1000 Hz (first three rows). For comparison, T*2 imaging with TE delays from 8 msec to 6 msec are shown

in the fourth row. The T1rho images show less loss of signal with increasing TSL than the T*2 images do with increasing TEs.

FIG. 5. T1rho mapping (a–d) andsingle-component fitting (e)showing a short T*2 of 0.78 msec

and T1rho values of 2.19, 4.95,and 7.43 msec, with spin-lock

fields of 250, 500, and 1000 Hz,respectively. There is significantT1rho dispersion in the Achilles

tendon (f).

838 Du et al.

Figure 8 shows UTE-T1rho imaging of the knee joint ofa 31-year-old healthy male volunteer. SNR ranged from32.0 with a TSL of 0.2 msec to 5.8 with a TSL of 25 msec.The saturation recovery UTE technique showed a T1 of620 msec for the meniscus, which was used for T1rho fit-ting with Eq. 1. Single-component curve fitting showed aT1rho of 8.02 6 0.37 msec at a spin-locking field of 500Hz. Figure 9 shows the curve fit for both T1rho dispersionand T*2 of a healthy male volunteer. A short T*2 of 4.15 60.36 msec and relatively short T1rho of 8.24 6 0.49 msecwere obtained at 500 Hz. There was significant T1rho dis-persion, with T1rho increasing from 6.93 6 0.44 msecwith a spin-locking field of 250 Hz to 12.96 6 0.54 msecwith a spin-locking field of 1000 Hz, corresponding to anincrease of 87%. T1rho of the cartilage was measured tobe 34.36 6 3.47 msec, which is about 16% lower than thevalue of 39.85 6 3.18 derived from the MAPSS sequence,probably due to some short T2 components in the articu-lar cartilage, which can be detected by the UTE-T1rhosequence but not by the MAPSS sequence (21).

The mean T1rho of the Achilles tendon in healthy vol-unteers was 3.06 6 0.51 msec compared to 5.22 6 0.58msec in cadaveric specimens. The T1rho was increasedto 8.99 6 0.24 msec for one specimen with obvious ten-don degeneration. A mean T1rho of 7.98 6 1.43 msecwas observed in the meniscus of healthy volunteers.There was a significant T1rho dispersion in both theAchilles tendon and the meniscus. Mean T1rhoincreased from 2.06 6 0.23 to 7.85 6 0.74 msec in nor-mal Achilles tendon and from 7.08 6 0.64 to 13.42 60.93 msec in normal meniscus when the spin-lock fieldwas increased from 250 to 1000 Hz. The average coeffi-cient of variation for the T1rho values of the Achillestendon and the meniscus in five healthy male volunteerswas 4.3 and 5.8%, respectively.

DISCUSSION

Human joints are composed of several different tissues,including the superficial layers of cartilage, the deep

FIG. 6. Fat-suppressed axial UTE-T1rho images of cadaveric ankle specimen at TSLs of 0.02 msec (a), 2 msec (b), 5 msec (c), and 10

msec (d), and the corresponding single-component curve fits that show a T1rho value of 5.93 6 0.38 msec for normal tendon and 8.996 0.24 msec for degenerated tendon with a 500-Hz spin-lock field (e).

FIG. 7. Fat-suppressed axial UTE-T1rho images of a healthy volunteer with TSLs of 0.2 msec (a), 1 msec (b), 4 msec (c), and 12 msec(d), and the corresponding single-component curve fit that gives a T1rho value of 3.45 6 0.21 msec with a 500-Hz spin-lock field (e).

UTE-T1rho Imaging 839

radial and calcified cartilage, bone, synovium, meniscus,tendon, and ligament. These tissues are all important forthe health of the joint. When one tissue begins to deteri-orate, it is likely to affect others and contribute to failureof the joint as a whole (28). Therefore, the etiology andprogression of OA should not be thought of as beinginvariably attributable to a single tissue, such as articularcartilage, but as possibly due to disease in any tissue ofthe affected organ (the diarthrodial joint) (28–30). Recentfindings by Tan et al. (29) indicate that structuralchanges in the collateral ligaments of the interphalangealjoints were consistently present even when the articularcartilage appeared normal, suggesting that abnormalitiesof collateral ligaments may lead to interphalangeal jointOA. Hunter et al. (30) investigated the association ofmeniscal pathologic changes with cartilage loss in symp-tomatic knee OA and found a strong association betweenmeniscal position, meniscal damage, and cartilage loss.Each aspect of meniscal abnormality (whether change inposition or damage) had a major effect on the risk of car-tilage loss.

The UTE-T1rho sequence described in this article pro-vides consistent T1rho imaging and quantification of

both the long T2 and short T2 tissues of the Achilles ten-don and meniscus. In contrast, the 3D MAPSS sequence,which is based on a 3D gradient echo acquisition, pro-vides a near-zero signal for the Achilles tendon and lim-ited signal for the meniscus with a TE of 2.4 msec. TheUTE-T1rho sequence is based on a conventional 2D UTEexcitation and acquisition combined with a nonselectivespin-lock preparation pulse. A minimal TE of 8 msec wasused for all UTE-T1rho images with different TSLs, pro-viding a constant eddy current pattern for all imagesused for curve fitting. The phantom results shown inFig. 2 confirmed that UTE-T1rho provides reliable T1rhoquantification of a long T2 phantom when comparedwith 3D MAPSS, which is based on a Cartesian acquisi-tion (21). Consistency was demonstrated through the sin-gle-component curve fitting shown in Figs. 5–9, whichaccounted for more than 99% of the signal variance.

The limited cadaveric ankle specimen study showed asignificant T1rho increase for degenerated relative to nor-mal tendon. Our preliminary results also show that thereis a significant T1rho dispersion in both the Achilles ten-don and the meniscus. Mean T1rho was increased by281% for the Achilles tendon and 90% for the meniscus

FIG. 8. Fat-suppressed oblique sagittal UTE-T1rho images of the knee joint of a 31-year-old healthy volunteer with TSLs of 0.2 msec(a), 1 msec (b), 5 msec (c), and 25 msec (d). The femorotibial cartilage and meniscus are seen with high spatial resolution, signal, andcontrast. Single-component curve fitting shows a T1rho value of 8.02 6 0.37 for the meniscus (e).

FIG. 9. Single-component curve fits of T1rho and T*2 (a) for meniscus in a healthy male volunteer show a short T*2 of 4 msec, T1 of 584

msec, and T1rho of 6.93 6 0.44 msec, 8.24 6 0.49 msec, and 12.96 6 0.54 msec with spin-lock fields of 250, 500, and 1000 Hz,respectively (b). Marked T1rho dispersion is shown in (b).

840 Du et al.

when the spin-lock field was increased from 250 to 1000Hz. This increase is significantly higher than that for thearticular cartilage (26), probably due to the strongdipole–dipole interaction that contributes both to T2 andT1rho relaxation mechanisms. A study by Knispel et al.(25) suggested that at least three relaxation mechanisms,including proton exchange between water molecules,rotational motion of a fraction of water bound to pro-teins, and molecular rotational and translational diffu-sion, are required to qualitatively describe the dispersionof the spin relaxation in tissues. Akella et al. (31) investi-gated the influence of RF spin-lock pulse on the laminarappearance of articular cartilage and found that T1rhoimaging reduced effects due to residual dipolar interac-tion. Mlynarik et al. (26) investigated transverse relaxa-tion mechanisms in cartilage and concluded that thedominant T1rho relaxation mechanism at an amplitudeof static field � 3 T was a dipolar interaction due toslow anisotropic motion of water molecules in the colla-gen matrix. In addition, the effect of diffusion of watermolecules in inhomogeneous fields produced by micro-scopic variations in tissue susceptibility can be sup-pressed by the spin-lock pulses.

The T1rho dispersion may be particularly importantfor short T2 tissues with a majority of type I collagen.The collagen fibers in the Achilles tendon are orientatedin parallel bundles with a highly ordered structure andrestricted motion of water molecules bound to collage-nous tissue. As a result, there is a significant enhance-ment in dipole–dipole interaction that is orientation de-pendent and is usually described as a magic-angle effect(24,27). The collagen fibers in meniscus show a morecomplicated pattern with three distinct groups of fibers,including the lamellar group (lamella fibers), the centralmain circumferential fibers, and a radial group (32). Thecollagen fibril bundles intersect at various angles and aretherefore subject to different degrees of magic-angleeffect. The dipole–dipole interaction directly affects T2

and T*2 relaxation times of the Achilles tendon and themeniscus and may play an important role in T1rho relax-ation time of these collagen-rich tissues. Furthermore,T1rho dispersion may provide information on boundwater and bulk water, which may have differentresponses to the spin-lock field (25).

Recent studies demonstrated a good correlationbetween T1rho signal distribution and cartilage PG distri-bution (26–29). However, it is unclear whether T1rho iscorrelated with PG in the short T2 tissues, which mainlycontain type I collagen, water, and a small amount ofPG. Besides the influence of PG, there may be an addi-tional contribution from collagen and hydration water tothe T1rho characteristics of each short T2 tissue (9–11).A recent paper by Krishnan et al. (9), using the delayedgadolinium (Gd)-enhanced MRI of cartilage technique,reported significant correlations between the T1(Gd) ofthe meniscus and T1(Gd) of the articular cartilage of theknee in OA, indicating that the degeneration occurred inboth tissues. However, the authors suggested that thedistribution of Gd-DTPA2- might not be dominated bythe PG distribution in the meniscus. Other factors suchas the collagen matrix and vascular supply differencesmight also contribute to the distribution of Gd-DTPA2-.

Bolbos et al. (10) investigated T1rho of the meniscus andits relationship with T1rho of adjacent cartilage in kneeswith acute anterior cruciate ligament (ACL) injuries andfound a strong injury-related relationship between me-niscus and cartilage biochemical changes. Rauscher et al.(11) investigated T1rho of the meniscus and found a highcorrelation between meniscal T1rho and clinical findingsin OA. However, all these techniques utilized clinicalgradient echo sequences with a TE of around 4 msec.Therefore, UTE-T1rho may provide more information onT1rho relaxation of the short T2 tissues.

The 2D UTE-T1rho acquisition is sensitive to eddy

currents, field homogeneity, and gradient nonlinearity

(16,33). Half-pulse excitation requires the summation of

two acquisitions with reversed slice-selection gradient

polarities to form a conventional slice profile. Gradient

profile distortion may result in mismatch between the

two acquisitions with nonideal cancellation of the imag-

inary parts of the complex signals and out-of-slice sig-

nal contamination. Correction of the residual slice-

select gradients and time-varying main amplitude of

static field(t) caused by eddy currents are helpful in

reducing this out-of-slice signal contamination (16,33).

However, the 2D UTE-T1rho sequence used in this

study employed the same minimal TE value of 8 msecand avoided effects due to variation in eddy currents

with TE. The combination of UTE with preparation

pulses for quantification of T1, T2, and T*2 is expected

to provide robust quantification and will be investigated

further in the future.

Another limitation of the UTE-T1rho sequence is the

relatively long scan time associated with its radial sam-

pling and the need for two NEX acquisitions. Accurate

T1rho measurement requires at least 35 min, where 20

min are required for T1 mapping and 15 min for UTE-

T1rho acquisitions at three TSLs. Undersampled projec-

tion acquisitions and shorter TRs can be used to reduce

the total scan time at the cost of streak artifacts and

reduced SNR (34). The UTE T1rho quantification is sub-

ject to errors in T1 quantification. Based on Eq. 1, a 4%

error in T1 quantification roughly leads to 1% error in

T1rho quantification, with a typical TR of 400 msec and

T1 of 650 msec. This study was also limited by the fact

that only a single slice was imaged and quantified. The

slice crosstalk in multislice UTE imaging can be reduced

by increasing the slice gap and using gradient calibration

methods to improve the slice profile.In this study, only six cadaveric ankle specimens and

five healthy volunteers were studied, and there was onlyone ankle specimen with degeneration. It is therefore dif-ficult to tell how sensitive UTE-T1rho values are likelyto decrease. Future work will include a larger number ofspecimens and patients for a more systematic study. TheUTE-T1rho sequence will also be used to investigate thecontribution of dipolar interaction in T1rho relaxation byusing UTE-T1rho imaging of collagenous tissues (such asthe Achilles tendon) at a series of orientations relative tothe amplitude of static field. Potential applications ofT1rho imaging of other short T2 tissues or tissue compo-nents, including calcified cartilage, ligaments, tendons,and entheses, will also be investigated.

UTE-T1rho Imaging 841

CONCLUSION

High-quality T1rho imaging and quantification of theAchilles tendon and meniscus can be achieved with 2DUTE sequences combined with spin-lock preparationpulses. The high signal from the Achilles tendon andmeniscus may provide improved confidence in morpho-logical and quantitative evaluation of tissues with shortT2 components in the MSK system.

ACKNOWLEDGMENTS

The authors thank Sheronda Statum and Richard Zna-mirowski for sample preparation.

REFERENCES

1. Eckstein F, Burstein D, Link TM. Quantitative MRI of cartilage and

bone: degenerative changes in osteoarthritis. NMR Biomed 2006;19:

822–854.

2. Potter HG, Black BR, Chong LR. New techniques in articular cartilage

imaging. Clin Sports Med 2009;28:77–94.

3. Bashir A, Gray ML, Burstein D. Gd-DTPA2- as a measure of cartilage

degradation. Magn Reson Med 1996;36:665–673.

4. Duvvuri U, Reddy R, Patel SD, Kaufman JH, Kneeland JB, Leigh JS.

T1rho-relaxation in articular cartilage: effects of enzymatic degrada-

tion. Magn Reson Med 1997;38:863–867.

5. Regatte RR, Akella SVS, Wheaton AJ, Lech G, Borthakur A, Kneeland

JB, Reddy R. 3D-T1q-relaxation mapping of articular cartilage: in vivo

assessment of early degenerative changes in symptomatic osteoar-

thritic subjects. Acad Radiol 2004;11:741–749.

6. Duvvuri U, Kudchodkar SB, Reddy R, Leigh JS. T1q relaxation can

assess longitudinal proteoglycan loss from articular cartilage in vitro.

Osteoarthritis Cartilage 2002;10:838–844.

7. Li X, Ma B, Link TM, Castillo D, Blumenkrantz G, Lozano J, Carbal-

lido-Gamio J, Ries M, Majumdar S. In vivo T1q and T2 mapping of

articular cartilage in osteoarthritis of the knee using 3T MRI. Osteoar-

thritis Cartilage 2007;15:789–797.

8. Brandt KD, Doherty M, Lohmander LS. Osteoarthritis. New York:

Oxford University Press; 1998.

9. Krishnan N, Shetty SK, Williams A, Brandon Nikulis, McKenzie C,

Burstein D. Delayed gadolinium-enhanced magnetic resonance imag-

ing of the meniscus. Arthritis Rheum 2007;56:1507–1511.

10. Bolbos RI, Link TM, Ma B, Majumdar S, Li X. T1q relaxation time of

the meniscus and its relationship with T1q of adjacent cartilage in

knees with acute ACL injuries at 3T. Osteoarthritis Cartilage 2008;17:

12–18.

11. Rauscher I, Stahl R, Cheng J, Li X, Huber MB, Luke A, Majumdar S,

Link TM. Meniscal measurements of T1rho and T2 at MR imaging in

healthy subjects and patients with osteoarthritis. Radiology 2008;249:

591–600.

12. Gatehouse PD, Bydder GM. Magnetic resonance imaging of short T2

components in tissue. Clin Radiol 2003;58:1–19.

13. Gold GE, Thedens DR, Pauly JM, Fechner KP, Bergman G, Beaulieu

CF, Macovski A. MR imaging of articular cartilage of the knee: new

methods using ultrashort TEs. AJR Am J Roentgenol 1998;170:

1223–1226.

14. Robson MD, Gatehouse PD, Bydder M, Bydder GM. Magnetic reso-

nance: an introduction to ultrashort TE (UTE) imaging. J Comput

Assist Tomogr 2003;27:825–846.

15. Sussman MS, Pauly JM, Wright GA. Design of practical T2-selective

RF excitation (TELEX) pulses. Magn Reson Med 1998;40:890–899.

16. Wansapura JP, Daniel BL, Pauly J, Butts K. Temperature mapping of

frozen tissues using eddy current compensated half excitation RF

pulses. Magn Reson Med 2001;46:985–992.

17. Du J, Hamilton G, Takahashi A, Bydder M, Chung CB. Ultrashort TE

spectroscopic imaging (UTESI) of cortical bone. Magn Reson Med

2007;58:1001–1009.

18. Du J, Takahashi A, Bydder M, Chung CB. Two dimensional ultra-

short echo time imaging using a spiral trajectory. Magn Reson Imag-

ing 2008;26:304–312.

19. Du J, Carl M, Takahashi AM, Diaz E, Chung CB, Han E, Bydder GM.

T1rho Imaging and quantification of the meniscus using a T1rho-pre-

pared ultrashort TE (T1rho-UTE) sequence. In: Proceedings of the

17th Annual Meeting of ISMRM, Honolulu, Hawaii, 2009. p 288.

20. Witschey WRT, Borthakur A, Elliott MA, Mellon E, Niyogi S, Wang

C, Reddy R. Compensation for spin-lock artifacts using an off-reso-

nance rotary echo in T1qoff-weighted imaging. Magn Reson Med

2007;57:2–7.

21. Li X, Han ET, Ma B, Busse RF, Majumdar S. In vivo T1q mapping in

cartilage using 3D magnetization-prepared angle-modulated parti-

tioned k-space spoiled gradient echo snapshots (3D MAPSS). Magn

Reson Med 2008;59:298–307.

22. Carl M, Bydder M, Takahashi A, Han E, Bydder GM. Maximizing RF

signal in the presence of rapid T2 relaxation. In: Proceedings of the

17th Annual Meeting of ISMRM, Honolulu, Hawaii, 2009. p 1991.

23. Aronen HJ, Ramadan UA, Peltonen TK, Markkola AT, Tanttu JI, Jaas-

kelainen J, Hakkinen AM, Sepponen R. 3D spin-lock imaging of

human gliomas. Magn Reson Imaging 1999;17:1001–1010.

24. Du J, Pak BC, Znamirowski R, Statum S, Takahashi A, Chung CB,

Bydder GM. Magic angle effect in magnetic resonance imaging of the

Achilles tendon and enthesis. Magn Reson Imaging 2009;27:557–564.

25. Knispel RR, Thompson RT, Pintar MM. Dispersion of proton spin-lat-

tice relaxation in tissues. J Magn Reson 1974;14:44–51.

26. Mlynarik V, Szomolanyi P, Toffanin R, Vittur F, Trattnig S. Trans-

verse relaxation mechanisms in articular cartilage. J Magn Reson

2004;169:300–307.

27. Fullerton GD, Cameron IL, Ord VA. Orientation of tendons in the

magnetic field and its effect on T2 relaxation time. Radiology 1985;

155:433–435.

28. Brandt KD, Radin EL, Dieppe PA, Putte L. Yet more evidence that

osteoarthritis is not a cartilage disease [editorial]. Ann Rheum Dis

2006;65:1261–1264.

29. Tan AL, Toumi H, Benjamin M, Grainger AJ, Tanner SF, Emery P,

McGonagle D. Combined high-resolution magnetic resonance imaging

and histological examination to explore the role of ligaments and

tendons in the phenotypic expression of early hand osteoarthritis.

Ann Rheum Dis 2006;65:1267–1272.

30. Hunter DJ, Zhang YQ, Niu JB, Tu X, Amin S, Clancy M, Guermazi A,

Grigorian M, Gale D, Felson DT. The association of meniscal patho-

logic changes with cartilage loss in symptomatic knee osteoarthritis.

Arthritis Rheum 2006;54:795–801.

31. Akella SVS, Regatte RR, Wheaton AJ, Borthakur A, Reddy R. Reduc-

tion of residual dipolar interaction in cartilage by spin-lock tech-

nique. Magn Reson Med 2004;52:1103–1109.

32. Petersen W, Tillmann B. Collagenous fibril texture of the human

knee menisci. Anat Embryol 1998;197:317–324.

33. Lu A, Daniel BL, Pauly JM, Pauly KB. Improved slice selection for

R2* mapping during cryoablation with eddy current compensation. J

Magn Reson Imaging 2008;28:190–198.

34. Du J, Lu A, Block WF, Thornton FJ, Grist TM, Mistretta CA. Time-

resolved undersampled projection reconstruction MR imaging of the

peripheral vessels using multi-echo acquisition. Magn Reson Med

2005;53:730–734.

842 Du et al.