1

Validation and optimization of adiabatic T1ρ and T2ρ for

quantitative imaging of articular cartilage at 3T

Victor Casula 1,2, Joonas Autio 3, Mikko J. Nissi 1,4-6, Edward J. Auerbach 4, Jutta Ellermann

4, Eveliina Lammentausta 7, and Miika T. Nieminen 1,2,7

1 Research Unit of Medical Imaging, Physics and Technology, University of Oulu, Oulu,

Finland,

2 Medical Research Center, University of Oulu and Oulu University Hospital, Oulu, Finland,

3 Center for Life Science and Technologies, RIKEN, Kobe, Japan

4 Center for Magnetic Resonance Research, Department of Radiology, University of

Minnesota, Minneapolis, United States

5 Department of Applied Physics, University of Eastern Finland, Kuopio, Finland,

6 Diagnostic Imaging Center, Kuopio University Hospital

7 Department of Diagnostic Radiology, Oulu University Hospital, Oulu, Finland

Address correspondence and reprint requests to: V. Casula, Research Unit of Medical

Imaging, Physics and Technology, University of Oulu, POB 50, FI-90029 Oulu, Finland Tel:

+358-505904448. E-mail address: victor.casula@oulu.fi (V.Casula).

Word count: 4803

Running title: Validation and optimization of adiabatic T1ρ and T2ρ for quantitative imaging

of articular cartilage at 3T

Keywords: articular cartilage; T1ρ; T2ρ; spin-lock; adiabatic; relaxation; in vivo.

2

Abstract

Purpose: The aim of the present work was to validate and optimize adiabatic T1ρ and T2ρ

mapping for in vivo measurements of articular cartilage at 3T.

Methods: Phantom and in vivo experiments were systematically performed on a 3 T clinical

system to evaluate the sequences using hyperbolic secant HS1 and HS4 pulses. R1ρ and R2ρ

relaxation rates were studied as a function of agarose and chondroitin sulfate concentration and

pulse duration. Optimal in vivo protocol was determined by imaging the articular cartilage of

two volunteers and varying the sequence parameters, and successively applied in eight

additional subjects. Reproducibility was assessed in phantoms and in vivo.

Results: Relaxation rates depended on agarose and chondroitin sulfate concentration. The

sequences were able to generate relaxation time maps with pulse length of 8 ms and 6 ms for

HS1 and HS4, respectively. In vivo findings were in good agreement with the phantoms. The

implemented adiabatic T1ρ and T2ρ sequences demonstrated regional variation in relaxation

time maps of femorotibial cartilage. Reproducibility in phantoms and in vivo was good to

excellent for both adiabatic T1ρ and T2ρ.

Conclusions: The findings indicate that sequences are suitable for quantitative in vivo

assessment of articular cartilage at 3T.

Keywords: articular cartilage; T1ρ; T2ρ; spin-lock; adiabatic; relaxation; in vivo.

3

Introduction

The biological applications of rotating frame relaxation techniques extend the capabilities of

quantitative MRI to study the biophysical properties of tissues, due to the high sensitivity of

the techniques for slow molecular motion. Relaxation during a continuous wave spin-lock pulse

for evaluation of articular cartilage has so far been extensively investigated [1]. T1ρ has been

shown to be sensitive to cartilage degradation [2-5], particularly to glycosaminoglycan (GAG)

content [4-8]. However, dependence on the properties of the collagen network has also been

reported [9]. Recent studies proposed novel imaging techniques where trains of adiabatic fast

passage pulses [10-13] have been adopted for spin-locking [14-18]. The main advantage of

using adiabatic pulses for in vivo implementation of rotating frame experiments resides in their

ability to overcome some of the limitations intrinsically associated with continuous wave spin-

lock sequences. Most notably, adiabatic pulses are insensitive to tilt angle inaccuracies by

definition [13], and hence less prone to the undesirable effects of non-uniform magnetic fields

compared to continuous wave spin-lock pulses [19]. Moreover, the adiabatic pulse approach

allows more flexibility in the pulse design and thus benefits moderation of the RF power

deposition [13], which instead significantly hampers the application of continuous wave pulses

in the clinical realm [20]. Furthermore, through an extended range of correlation times

effectively involved in the relaxation during adiabatic pulses, more of the physicochemical

mechanisms underlying pathological changes in tissues may be monitored to elucidate

information potentially relevant for diagnostic purposes.

Adiabatic fast passage pulses have been successfully applied for assessing longitudinal and

transverse relaxation times in the rotating frame, namely adiabatic T1ρ (AdT1ρ) and adiabatic

T2ρ (AdT2ρ), and their ability to probe slow molecular motion has been proven at different field

strengths [14, 16-18, 21]. For articular cartilage, preclinical studies on AdT1ρ and AdT2ρ have

demonstrated evidence of improved sensitivity for detection of early degenerative changes in

both animal and human specimens over rotating frame relaxation times assessed with

conventional spin lock experiments or free precession relaxation times [22-25]. Furthermore

AdT1ρ has been reported less sensitive to the magic angle effect compared to T2 and continuous

wave T1ρ [26]. Such ability in characterizing degeneration of cartilage could be exploited to

provide valuable and practical diagnostic application for clinical use. The rationale of this study

was to validate and optimize adiabatic T1ρ and adiabatic T2ρ sequences on a 3 T clinical scanner

for in vivo quantitative measurements of cartilage.

4

5

Methods

MR pulse sequences

AdT1ρ and AdT2ρ sequences were implemented on a 3 T clinical system (Skyra, Siemens,

Erlangen, Germany) using the IDEA sequence programming environment (Version

VD13A/SP4, Siemens). Adiabatic T1ρ mapping was performed as described previously [18, 21,

22, 24, 25] using a preparation block which consisted, for both phantom and in vivo

experiments unless otherwise stated, of a train of 0, 4, 8, 12 and 16 adiabatic full passage (AFP)

hyperbolic secant pulses of the HSn family (n = 1,4) with R = 10 and phase cycled with MLEV4

[13, 27, 28] (Fig. 1a). For AdT2ρ, the AFP pulses were placed between two adiabatic half

passage pulses (AHP), and two spatial pre-saturation slabs were applied above and below the

FOV to correct for off-resonance effects. A 15-channel transmit/receive knee coil (single

channel transmit) was used. For both AdT1ρ and AdT2ρ, the peak RF amplitude was set to γB1max

= 800 Hz (highest achievable with the selected coil). The preparation block was followed by a

segmented gradient recalled echo (FLASH) readout.

Phantom measurements

Eleven gel tubes were prepared with different concentrations of agarose (0.50 - 4.00% w/v) to

yield transverse relaxation times within the range of biological tissue, and served as a test

phantom for evaluating the feasibility, performance, stability, reproducibility and operational

limits of the adiabatic T1ρ and T2ρ sequences. Imaging parameters were adjusted for the

phantom (TR / TE = 20 s / 3.75 ms, flip angle = 15°, matrix = 256 x 256, FOV = 120 x 120

mm2, slice thickness = 7 mm, 2 averages), and AdT1ρ and AdT2ρ were measured with five

different AFP pulse durations (Tp), i.e. 3, 6, 8, 10 and 12 ms. Additionally, T2 relaxation time

of the phantom was measured with a multi-echo spin echo sequence (TR = 10 s, 32 TE between

10.7 and 342.4 ms, matrix = 256 x 256, FOV = 120 x 120 mm2, slice thickness = 7 mm, 2

averages). To assess the homogeneity of B1, flip angle maps were measured using B1 mapping

sequence provided by the scanner manufacturer (Siemens). The relaxation times were

investigated as relaxation rates R2 (= 1/T2), R1ρ (= 1/T1ρ) and R2ρ (= 1/T2ρ) due to the more

straightforward analysis. Furthermore relaxivities r2, r1ρ and r2ρ as a function of agarose

concentration were determined.

6

Additionally, AdT1ρ and AdT2ρ experiments were performed for the three lowest agarose

concentration tubes and for a supplementary distilled water phantom using longer total spin

lock durations (trains of 0, 4, 8, 12, 16, 20, 24, 28, 32, 36 and 40 pulses, Tp = 8 ms for HS1 and

Tp = 6 ms for HS4) to better observe the relaxation behavior when approaching pure water

relaxation.

To specifically study the dependence of adiabatic relaxation time measurements on cartilage

proteoglycans, chondroitin sulfate (CS) phantoms with various concentrations (0, 2, 4, 6, 8 %

w/w) were prepared from chondroitin sulfate A (CS; Sigma-Aldrich, Schnelldorf, Germany) in

PBS containing 0.2 mM MnCl2 to reduce T1, with pH of the solutions titrated to 7.1, which has

been shown to be the pH of clinically normal human cartilage [29]. R1ρ and R2ρ were measured

(TR / TE = 5 s / 3.75 ms, flip angle = 15°, matrix = 256 x 256, FOV = 120 x 120 mm2, slice

thickness = 7 mm, 2 averages) and studied as functions of CS concentration with four different

Tps (6, 8, 10 and 12 ms). For reference, R2 was also measured (TR = 5 s, 32 TE between 10.7

and 342.4 ms, matrix = 256 x 256, FOV = 120 x 120 mm2, slice thickness = 7 mm, 2 averages).

Relaxivities r2, r1ρ and r2ρ were calculated as a function of the chondroitin sulfate concentration.

In vivo measurements

To demonstrate the feasibility of AdT1ρ and AdT2ρ measurements in the human knee joint, ten

asymptomatic adult volunteers (age range 25-35 y, mean age 27.4 y; six male, four female)

were included in this study. Additionally, three subjects with radiographic signs of

osteoarthritis (Kellgren-Lawrence (KL) grade ≥ 2) were enrolled: a 52 y male (KL = 2), a 62 y

female (KL = 3) and a 57 y male (KL = 4). Approval from the local institutional review board

on human research ethics and informed consents from the subjects were obtained prior to the

study.

To optimize the relevant sequence parameters, including flip angle, repetition time and pulse

duration, the first two subjects were scanned with the following protocols:

1. Single sagittal slices of right medial condyle were taken with no T1ρ-preparation; the

flip angle was varied (6, 10, 15, 20, 30, 45, 60 and 90 degrees) and the optimal value

was evaluated in terms of SNR and absence of artifacts.

7

2. At the same anatomical location, adiabatic T1ρ relaxation maps were obtained with four

different TRs (1, 2.5, 4, 6 s) and analyzed to find the shortest suitable TR. Flip angle

was taken from the previous step and the preparation pulse lengths were the shortest

fulfilling the adiabatic condition [13, 28] on the basis of the phantom findings, i.e. Tp,HS1

= 8 ms and Tp,HS4 = 6 ms.

3. AdT1ρ and AdT2ρ experiments were finally repeated with three different Tps (3, 6, 8, 10

and 12 ms), using the determined optimal TR and flip angle from steps 1 and 2.

In this adopted approach, the optimal parameters were determined by evaluating the relaxation

time maps and by minimizing the scanning time and RF power deposition while maximizing

the quality of the images in terms of signal to noise ratio, stability and absence of artifacts. The

other acquisition parameters for all the in vivo experiments were TE = 3.36 ms, FOV = 180 x

180 mm2, matrix = 256 x 256, and slice thickness = 3 mm. In addition, parallel imaging

acceleration factor = 2 (GRAPPA) and segmented k-space acquisition (23 acquisitions per

preparation) were used to decrease the scanning time to be clinically suitable with no

significant reduction of image quality. Similar to phantom experiments, the uniformity of B1

was evaluated by acquiring the flip angle maps using the B1 mapping sequence described

above.

The optimized protocols were then used to image the knees of all the ten volunteers. T1ρ and

T2ρ data were acquired and maps computed in slices covering the centers of medial and lateral

condyles. To determine the in vivo reproducibility, measurements with the optimized protocols

were repeated three times in four subjects, with knee repositioning between consecutive scans.

For comparison purposes, T2 measurements were performed scanning the same joint locations

with a multi-echo spin echo sequence (TR = 1.680 s, five TEs between 13.8 and 69 ms, matrix

= 384 x 384, FOV = 160 x 160 mm2, slice thickness = 3 mm).

To explore the potential of the optimized protocol in clinical setting, we also scanned three OA

patients. The pulse type with the shortest pulse duration Tp based on the phantom and in vivo

measurements (here HS4), less demanding in terms of power deposition and thus more suitable

for clinical examinations, was employed. The relaxation time constants of articular cartilage

were determined in two slices positioned at the centers of the lateral and medial condyles.

8

Data processing

Phantom and in vivo image analyses were executed with an in-house MATLAB-based

software. Longitudinal and transverse relaxation time constants in the rotating frame were

calculated by mono-exponential fitting of the signal intensity decays on a pixel-by-pixel basis.

In all the in vivo images, articular cartilage was manually segmented by a single operator (V.C.)

and 18 ROIs were considered: each condyle was divided into six superficial and six deep ROIs,

and each tibial plateau into three superficial and three deep ROIs (Fig. 1b).

Statistical analysis

Coefficient of determination (R2) and root-mean-square deviation (RMSD) values were

calculated between the relaxation rates and agarose or chondroitin sulfate concentrations. The

reproducibility of the techniques was assessed with the agarose tubes by calculating the average

root-mean-square coefficient of variation (CVRMS) of three measurements repeated within one

month using Tp = 8 ms and Tp = 6 ms for HS1 and HS4, respectively. To determine the in vivo

reproducibility of AdT1ρ and AdT2ρ, CVRMS of three consecutive scans in four subjects was

calculated for bulk cartilage and for each ROI.

9

Results

Agarose gel and chondroitin sulfate phantoms

R1ρ and R2ρ were strongly linearly correlated with agarose concentration (Fig. 2 and Sup. Fig.

S1, R2 = 0.98 – 0.99, RMSD = 0.24 – 1.25). Using Tp = 6 ms with HS1, linear correlation

between relaxation rates and agarose concentration was smaller (R2 = 0.93 - 0.94, RMSD =

0.69 – 1.39) and the maps were less homogeneous (Fig. 3). For shortest pulse length (Tp = 3

ms) relaxation rate values exhibited notable errors, the correlation with concentration was

reduced (R2 = 0.70 – 0.85, RMSD = 3.02 – 18.67) and the relaxation time maps were largely

inhomogeneous (Fig. 3). Table 1 reports the relaxivities as a function of agarose concentration.

Relaxivity r1ρ was smaller than r2ρ. Furthermore, r1ρ increased with pulse duration and was

greater with HS4. On the contrary, r2ρ decreased with pulse duration and was greater with HS1.

In the agarose phantom, relaxation rates showed a tendency to converge at lower concentrations

(Fig. 2), except for HS1 with Tp = 6 ms. This convergence was better shown for R1ρ,HS1 and

R2ρ,HS1 with Tp = 8 ms and R1ρ,HS4 and R2ρ,HS4 with Tp = 6 ms, when measured using a larger

number of and including longer duration spin-lock times in the low concentration samples

(Sup. Fig. S1). Note that all the relaxation times in the reference water phantom were very close

to each other (Sup. Fig. S1).

For the agarose phantom, the mean relative error of the flip angle in the imaged region was

3.6%. The reproducibility assessed with agarose phantom (the CVRMS) for the AdT1ρ sequence

was 2.2% with HS1 and 1.3% with HS4. For the AdT2ρ sequence the reproducibilities were

2.9% and 2.3% with HS1 and HS4, respectively.

Although relaxation rates for the CS phantom were generally smaller than those of the agarose

tubes, all of the aforementioned findings were confirmed with the CS phantom as well (Fig. 4).

R1ρ and R2ρ were strongly linearly correlated with CS concentration (R2 = 0.98 – 0.99, RMDS

= 0.08 – 0.22). Table 1 shows the relaxivities for CS phantoms. For CS, r2ρ was greater than

r1ρ.

In vivo measurements of articular cartilage

10

The SNR of the FLASH images in cartilage region showed a maximum between flip angles of

20 to 30 degrees (SNR = 68 - 98) and monotonically decreased to a minimum at 90 degrees

(SNR = 28 - 48). At larger flip angles the images were visibly affected by aliasing artifact,

which increased towards greater angles. At 15 degrees the SNR was almost doubled compared

to 6 degrees (SNR = 23 - 30) and the images were free of artifacts. For the FLASH readout

sequence utilized, flip angle of 15 degrees was found to be the best compromise between good

SNR (~ 54 - 60) and preventing the aliasing artifacts that occurred at larger flip angles.

Table 2 shows the T1ρ relaxation times of the entire femoral and tibial AC as obtained using

different TR values. Relaxation with the shortest TR (= 1 s) was considerably faster as

compared to those measured with longer TRs (mean difference = 27.2 %). Mean difference

between relaxation times obtained with TR of 4 and 6 s was minimal (2.2 %) and smaller than

the mean difference of T1ρ between TRs of 2.5 and 6 s (4.8 %). Therefore repetition time of 4

s was considered the best compromise between scan time and relaxation dependency on TR.

The variability in T1ρ and T2ρ relaxation times with HS1 and HS4 pulse duration is reported in

Table 3 and summarized below. Very short and incongruous relaxation values were measured

with Tp of 3 ms irrespective of the pulse type, in addition to atypical variances across T1ρ maps

caused by signal dropout and distortion affecting the images and increasing with spin-lock

time. Such features, indicating violation of the adiabaticity, were still observed, although

largely mitigated in T1ρ maps acquired with HS1 pulses of 6 ms. With longer pulse durations

for HS1, limited dependency on pulse duration was noted for relaxation time. Using HS4, T1ρ

revealed a maximum with pulse duration of 6 ms and then a small tendency to decrease with

increasing Tps. Conversely, T2ρ appeared to slightly increase with increasing pulse duration.

Based on these findings, Tp of 8 and 6 ms were the shortest pulse lengths satisfying the adiabatic

condition for HS1 and HS4, respectively. The scanning time for obtaining full relaxation data

necessary for fitting of the time constants with the final protocol was 2 min and 22 s per slice.

In the in vivo scans, the mean error of the flip angle was 3.2% in femoral cartilage and 3.5%

in tibial cartilage. Supporting Table S1 reports the in vivo reproducibilities. CVRMS was lower

than 5% in bulk ROIs of lateral and medial compartments for both T1ρ and T2ρ, except in medial

tibia for T1ρ,HS1.

Figure 5 shows representative T1ρ and T2ρ relaxation time maps of articular cartilage obtained

using the most appropriate settings found in this study (flip angle = 15 degrees, TR = 4 s, Tp =

11

8 ms for HS1 and Tp = 6 ms for HS4). Table 4 reports the mean and standard deviation of T2,

T1ρ and T2ρ for each ROI. Anterior trochlea could not be analyzed in the medial compartment

due to the slice orientation. In all the ROIs, T1ρ,HS1 was always longer than T1ρ,HS4. T2ρ,HS1 was

always shorter than T2ρ,HS4, except in deep central tibia. On average, relaxation times were

longer in femur than in tibia and longer in the superficial zone than in the deep cartilage. The

shortest relaxation values were found in deep lateral central tibia. The longest values of T2, T1ρ

and T2ρ were found among superficial trochlear and central femoral ROIs. Figure 6 shows a

comparison of relaxation time maps between a representative healthy subject and three patients

with radiographic OA. On average, T1ρ and T2ρ relaxation times of femorotibial cartilage with

OA were found respectively 13-52% and 18-68% longer compared to healthy control mean

values, with a trend for increasing values with increasing KL score.

12

Discussion

Relaxation in the rotating frame can probe the long time scale component of molecular

interactions; as a result the influence of processes such as chemical exchange becomes more

relevant as compared to free relaxation where the contribution of direct dipolar-dipolar

interactions is prevailing. This has been proposed as an explanation for the sensitivity of

continuous wave T1ρ to cartilage proteoglycans [30, 31] and reduced sensitivity to the magic

angle effect as compared to T2 [32]. To diminish the magic angle effect in continuous wave

T1ρ, it has been calculated that spin lock frequencies of 2 kHz or higher would be necessary

[33], far beyond the power accessible in the clinical realm. The significant advantages of AdT1ρ

are the simultaneous sensitivity to more than one time scale [16, 17] and the fact that it is

essentially unaffected by susceptibility and magic angle effect [26]. Previously, T1ρ dependence

on CS and proteoglycans of articular cartilage (of which CS is a major component) has been

shown [6, 7, 20, 34]. Nonetheless in literature there is no consensus regarding the T1ρ relaxation

mechanisms in cartilage. A recent study has showed that T1ρ does not correlate with

glycosaminoglycan [35] and another study suggested that the contribution of scalar relaxation

becomes relevant only at high fields, whilst at clinical fields dipolar interactions are the

dominant relaxation mechanism in T1ρ [36]. The relaxivities found in this study are greater than

those reported in previous studies at clinical fields and comparable to r1ρ at higher fields [34,

37]. This suggests that AdT1ρ has better sensitivity to CS as compared to continuous wave T1ρ.

AdT2ρ has been reported to be sensitive to early cartilage degeneration in an animal model of

OA and advanced OA in human tissue [22, 25, 38]. Although the biophysical mechanisms of

T2ρ contrast in cartilage remain unknown, it has been proposed that the contribution of proton

exchange is more effective as compared to T1ρ. Thus T2ρ relaxation process should be more

sensitive to cartilage proteoglycans than T1ρ. This is consistent with the r2ρ relaxivities found

in this study, which were greater than the r1ρ relaxivities (Table 1). Since continuous wave T2ρ

is particularly sensitive to B1 inhomogeneities and requires high spin-lock fields, its utilization

in clinical studies has been limited by SAR constrains. Therefore AdT2ρ, thanks to the

insensitivity of adiabatic pulses to B1 inhomogeneities, provides a unique means for measuring

transverse relaxation in the rotating frame in vivo.

In this study the feasibility of AdT1ρ and AdT2ρ on a clinical MRI system was documented via

systematic phantom and in vivo tests. Through phantom experiments it was verified that both

13

of the sequences functioned as expected and produced results consistent with previous studies.

Specifically, in agarose phantoms the relaxation rates increased with increasing gel density [39,

40] and converged to the pure water relaxation (R1ρ ≈ R2ρ ≈ R2) towards lower concentrations.

Furthermore, R1ρ,HS1 was smaller than R1ρ,HS4 [24] and conversely R2ρ,HS1 was greater than

R2ρ,HS4 in the agarose phantom [21], in agreement with previous studies and the in vivo

measurements. Relaxation was also found to be dependent on the pulse duration. Increasing

the length of an AFP pulse corresponds to decreasing of the frequency sweep amplitude [13],

which in turn yielded faster R1ρ and slower R2ρ, as expected [21]. The aforementioned results

were further confirmed by CS experiments, except for approximating the pure water relaxation,

effect not visible here since phantoms were paramagnetically doped.

Phantom and in vivo measurements were congruent indicating the shortest suitable pulse

duration Tp for HS1 and HS4. With 3 ms pulses, the images exhibited severe random artifacts

suggesting violation of the adiabatic condition [13], wherefrom the weak association of

relaxation rates with agarose and CS content. With 6 ms pulses using HS1 shape, mild errors

were observed in the relaxation maps. The same inhomogeneities were found in vivo and not

imputable to the physiological variation within the tissue, rather revealing a sub-adiabatic

regime. This explains the unexpected behavior of the relaxation rates obtained using HS1 with

6 ms pulse duration. Conversely, using HS4 pulses with Tp of 6 ms resulted in homogenous

maps and in shorter R1ρ and greater R2ρ as compared to longer Tps. Such dissimilarity is

attributed to be a direct consequence of the different “stretching” factors of the HSn pulses.

HS4 pulses are de facto less demanding compared to HS1 in terms of peak RF power

requirements. The peak power required to fully satisfy the adiabatic condition with 6 ms HS1

pulses exceeded the capabilities of the RF transmit chain.

The reproducibility of AdT1ρ and AdT2ρ in phantom, as indicated by the CVRMS of agarose

tubes, was excellent for the shortest suitable pulse durations, i.e. 8 ms for HS1 and 6 ms for

HS4. Previous studies have reported in vivo T1ρ reproducibility to range from 1.7-19% [41-

43]. In vivo AdT1ρ reproducibility ranged from 1.1% to 10.8%, was excellent in over 70% of

the ROIs and was greater than 10% in only one ROI (Sup. Table S1). In vivo AdT2ρ

reproducibility ranged from 1.4% to 15.3%, was excellent in over 44% of the ROIs and was

14

greater than 10% in only three ROIs (Sup. Table S1). The relatively small CVRMS, while

providing supplementary evidence that sequences are correctly working, demonstrated in

addition their appropriateness for accurate and longitudinally stable quantitative analyses.

The experiments with human subjects demonstrated the in vivo feasibility of the sequences and

indicated the most appropriate protocol for reproducible and reliable relaxation time

quantification. Concerning the optimal flip angle and TR, the choice was mostly led by

conservative arguments. While 15 degrees was not the absolute best flip angle in terms of

signal-to-noise ratio, it produced artifact free images while still retaining excellent SNR; a

smaller flip angle is also conducive to mitigating SAR and was thus considered as the optimal

choice. The need for minimizing the acquisition time with short TRs must meet the minimum

length required to avoid excessive formation of steady-state if one wishes to retain quantitative

relaxation coefficients. Using a TR of 1 second, the relaxation was too short as compared to

longer TRs, suggesting that the signal reduction was rather due to partial saturation. Although

no strong variations in relaxation times were observed using the other considered TRs, a closer

look revealed that the least difference in average was between values obtained with 4 and 6s

TR, hence the former appears to be a more cautious selection than 2.5 s, yet preferable to 6 s

for decreasing the acquisition time. Different delay times between two consecutive RF spin-

lock pulses have been used in previous T1ρ cartilage studies. Witschey et al., for example, used

a conservative value of 6 s [44]. A delay time, shorter than that empirically determined in the

present study has been employed in combination with RF cycling [42]. The TR used in this

study with FLASH readout appeared as the optimal compromise between acceptable clinical

acquisition times and avoiding formation of a complex steady-state.

Regarding the durations of the HS1 and HS4 pulses, the most important results were the

shortest possible Tps, which were discussed above. The scanning time with the optimal protocol

was appropriate for clinical applications, allowing reliable quantification of T1ρ and T2ρ of two

slices. The small error in the flip angle both in vivo and in phantoms indicated presence of only

a mild B1 inhomogeneity at 3 T in the setup used, which could be compensated by the use of

adiabatic pulses. In vivo experiments showed variation in relaxation times depending on

topographical location (medial or lateral compartment; femur or tibia; anterior, central or

posterior region), depth (superficial or deep zone), pulse used (HS1 or HS4), and contrast (T1ρ

or T2ρ). The findings were consistent with those reported in previous studies. Specifically

longer T1ρ values were obtained with HS1 as compared to HS4, with a mean difference of about

15

40% [17, 21, 24]; superficial ROIs had longer relaxation times as compared to deep ROIs [23];

T1ρ were longer than T2ρ [23]; T2 were shorter than T1ρ [3, 45]. T1ρ, T2ρ and T2 exhibited

differences in regional distribution, attributable to the different information regarding the tissue

provided by the three parameters. These results confirm the ability of the sequences to probe

differences in the structure and composition of AC also at clinically accessible fields.

The increased adiabatic T1ρ and T2ρ observed in cartilage of patients with OA is in agreement

with previous reports that the techniques are sensitive biomarkers for cartilage degeneration

[22, 23, 25]. Further studies are warranted to compare adiabatic T1ρ and T2ρ in asymptomatic

subjects with early OA patients.

Certain limitations restrain the generalizability of some results of this study. For instance,

accuracy of the sequences is likely to change with systems of different manufacturers, in which

also the sequence stability has to be verified. Optimal TR and flip angle may also vary, since

they both are, for example, not independent of B0. Finally the minimum pulse lengths required

for achieving adiabaticity for HS1 and HS4 may differ using different RF coils and amplifiers,

since that depends on the attainable RF power. Implementation-wise, the adiabatic T1ρ and T2ρ

sequences require an introduction of a train of adiabatic inversion pulses in front of a readout

sequence, which could be FLASH, EPI or other appropriate sequence, and should not be a

major limiting factor. However, a limitation of the present setup is the relatively long scan time,

i.e. data for only two slices could be acquired within a reasonable time frame. The TR could

be shortened significantly with RF cycling [46] and further studies are needed to test the

method with trains of HSn pulses. The acquisition could also be sped up by other choices, such

as increasing the portion of k-space acquired per preparation or using partial acquisition [47],

by increasing the parallel acquisition factor or utilizing multiband excitation [48] or by

applying other techniques for speeding up the readout. T1ρ sequences are generally prone to

SAR issues due to the RF energy required to manipulate the contrast before acquisition [20],

the adiabatic pulses utilized here provide flexibility overcoming the SAR issues and changes

in the TR and factors speeding up the acquisition may be utilized to further alleviate SAR.

Longitudinal magnetization recovery and transient signal evolution during acquisition may

degrade T1ρ and T2ρ contrasts. RF cycling [46] and variable flip angle techniques may be tested

in combination with AFP pulse trains to reduce inaccuracies [42, 44].

16

Another possible source of inaccuracies is inherent in the quantitation method of transverse

relaxation time. In this study, T2 values of some phantoms or ROIs in in vivo data were longer

than the corresponding T2ρ values, which in theory should not occur. Rautiainen et al. recently

reported discrepancies in T2 as measured with different sequences on a 9.4 T system and mean

values of T2 longer than T2ρ, depending on the sequence used for measuring T2 [22]. Variability

of T2 has also been observed in phantoms and in vivo measurements carried out on different

clinical scanners and with different pulse sequences [49, 50]. In this study, the higher than

expected T2s were likely due to the properties of the T2-mapping sequence provided by the

system manufacturer.

In conclusion, adiabatic T1ρ and adiabatic T2ρ appear feasible for in vivo quantitative analysis

of cartilage tissue at 3 T within acceptable scanning times for clinical application. Previous

investigations have shown the superior sensitivity of these contrasts for the assessment of

articular cartilage as opposed to other relaxation time measurements [22-24] in the ex vivo

setting and thus with the demonstrated feasibility, the parameters appear promising for in vivo

as well. The potential of these techniques in detecting early osteoarthritis will be investigated

in subsequent studies.

17

Acknowledgments

The Center for Magnetic Resonance Research, University of Minnesota is gratefully

acknowledged for providing the adiabatic T1ρ and T2ρ sequences. The authors would like to

thank Silvia Mangia and Shalom Michaeli for constructive discussions, Anna-Leena Manninen

and Abdul Wahed Kajabi for their advice and assistance with phantom preparation, and Ute

Goerke for the original Siemens implementation of the adiabatic T1ρ sequence. This work was

supported by Academy of Finland (grants 260321, 285909 and 293970).

18

References

1. Wang L, Regatte RR. T1rho MRI of human musculoskeletal system. Journal of Magnetic

Resonance Imaging 2015;41(3):586-600.

2. Regatte RR, Akella SV, Borthakur A, Kneeland JB, Reddy R. In vivo proton MR three-

dimensional T1rho mapping of human articular cartilage: Initial experience. Radiology

2003 Oct;229(1):269-74.

3. Li X, Benjamin Ma C, Link TM, Castillo DD, Blumenkrantz G, Lozano J, Carballido-

Gamio J, Ries M, Majumdar S. In vivo T(1rho) and T(2) mapping of articular cartilage

in osteoarthritis of the knee using 3 T MRI. Osteoarthritis Cartilage 2007 Jul;15(7):789-

97.

4. Keenan KE, Besier TF, Pauly JM, Han E, Rosenberg J, Smith RL, Delp SL, Beaupre GS,

Gold GE. Prediction of glycosaminoglycan content in human cartilage by age, T1rho and

T2 MRI. Osteoarthritis Cartilage 2011 Feb;19(2):171-9.

5. Wheaton AJ, Casey FL, Gougoutas AJ, Dodge GR, Borthakur A, Lonner JH, Schumacher

HR, Reddy R. Correlation of T1rho with fixed charge density in cartilage. J Magn Reson

Imaging 2004 Sep;20(3):519-25.

6. Akella SV, Regatte RR, Gougoutas AJ, Borthakur A, Shapiro EM, Kneeland JB, Leigh JS,

Reddy R. Proteoglycan-induced changes in T1rho-relaxation of articular cartilage at 4T.

Magn Reson Med 2001 Sep;46(3):419-23.

7. Duvvuri U, Kudchodkar S, Reddy R, Leigh JS. T1ρ relaxation can assess longitudinal

proteoglycan loss from articular cartilage in vitro. Osteoarthritis and Cartilage 2002

11;10(11):838-44.

8. Wang N, Xia Y. Depth and orientational dependencies of MRI T(2) and T(1rho)

sensitivities towards trypsin degradation and gd-DTPA(2-) presence in articular cartilage

at microscopic resolution. Magn Reson Imaging 2012 Apr;30(3):361-70.

9. Menezes NM, Gray ML, Hartke JR, Burstein D. T2 and T1rho MRI in articular cartilage

systems. Magn Reson Med 2004 Mar;51(3):503-9.

10. Tannús A, Garwood M. Adiabatic pulses. NMR Biomed 1997;10(8):423-34.

11. Norris DG. Adiabatic radiofrequency pulse forms in biomedical nuclear magnetic

resonance. Concepts Magn Reson 2002;14(2):89-101.

12. De Graaf RA, Nicolay K. Adiabatic rf pulses: Applications to in vivo NMR. Concepts

Magn Reson 1997;9(4):247-68.

13. Garwood M, DelaBarre L. The return of the frequency sweep: Designing adiabatic pulses

for contemporary NMR. J Magn Reson 2001 Dec;153(2):155-77.

19

14. Mangia S, Traaseth NJ, Veglia G, Garwood M, Michaeli S. Probing slow protein

dynamics by adiabatic R 1ρ and R 2ρ NMR experiments. J Am Chem Soc

2010;132(29):9979-81.

15. Mangia S, Carpenter AF, Tyan AE, Eberly LE, Garwood M, Michaeli S. Magnetization

transfer and adiabatic T1 MRI reveal abnormalities in normal-appearing white matter of

subjects with multiple sclerosis. Multiple Sclerosis 2014;20(8):1066-73.

16. Michaeli S, Grohn H, Grohn O, Sorce DJ, Kauppinen R, Springer CS,Jr, Ugurbil K,

Garwood M. Exchange-influenced T2rho contrast in human brain images measured with

adiabatic radio frequency pulses. Magn Reson Med 2005 Apr;53(4):823-9.

17. Michaeli S, Sorce DJ, Springer CS,Jr, Ugurbil K, Garwood M. T1rho MRI contrast in the

human brain: Modulation of the longitudinal rotating frame relaxation shutter-speed

during an adiabatic RF pulse. J Magn Reson 2006 Jul;181(1):135-47.

18. Michaeli S, Sorce DJ, Garwood M. T2ρ and T1ρ adiabatic relaxations and contrasts.

Current Analytical Chemistry 2008;4(1):8-25.

19. Witschey II WRT, Borthakur A, Elliott MA, Mellon E, Niyogi S, Wallman DJ, Wang C,

Reddy R. Artifacts in T1ρ-weighted imaging: Compensation for B1 and B0 field

imperfections. Journal of Magnetic Resonance 2007 5;186(1):75-85.

20. Borthakur A, Mellon E, Niyogi S, Witschey W, Kneeland JB, Reddy R. Sodium and

T1rho MRI for molecular and diagnostic imaging of articular cartilage. NMR Biomed

2006 Nov;19(7):781-821.

21. Mangia S, Liimatainen T, Garwood M, Michaeli S. Rotating frame relaxation during

adiabatic pulses vs. conventional spin lock: Simulations and experimental results at 4 T.

Magn Reson Imaging 2009 Oct;27(8):1074-87.

22. Rautiainen J, Nissi MJ, Liimatainen T, Herzog W, Korhonen RK, Nieminen MT.

Adiabatic rotating frame relaxation of MRI reveals early cartilage degeneration in a

rabbit model of anterior cruciate ligament transection. Osteoarthritis and Cartilage 2014

2014/09;22(10):1444-52.

23. Rautiainen J, Nissi MJ, Salo E, Tiitu V, Finnilä MAJ, Aho O, Saarakkala S, Lehenkari P,

Ellermann J, Nieminen MT. Multiparametric MRI assessment of human articular

cartilage degeneration: Correlation with quantitative histology and mechanical

properties. Magnetic Resonance in Medicine 2015;74(1):249-59.

24. Ellermann J, Ling W, Nissi MJ, Arendt E, Carlson CS, Garwood M, Michaeli S, Mangia

S. MRI rotating frame relaxation measurements for articular cartilage assessment. Magn

Reson Imaging 2013;31(9):1537-43.

25. Nissi MJ, Salo EN, Tiitu V, Liimatainen T, Michaeli S, Mangia S, Ellermann J, Nieminen

MT. Multi-parametric MRI characterization of enzymatically degraded articular

cartilage. J Orthop Res 2015. doi: 10.1002/jor.23127.

20

26. Nissi MJ, Mangia S, Michaeli S, Nieminen MT. Orientation anisotropy of rotating frame

and T2 relaxation parameters in articular cartilage. . Proc Intl Soc Mag Reson Med

2013;21:3552.

27. Warnking JM, Pike GB. Bandwidth-modulated adiabatic RF pulses for uniform selective

saturation and inversion. Magnetic Resonance in Medicine 2004;52(5):1190-9.

28. Silver MS, Joseph RI, Hoult DI. Highly selective and p pulse generation. Journal of

Magnetic Resonance (1969) 1984;59(2):347-51.

29. Konttinen YT, Mandelin J, Li T-, Salo J, Lassus J, Liljeström M, Hukkanen M, Takagi

M, Virtanen I, Santavirta S. Acidic cysteine endoproteinase cathepsin K in the

degeneration of the superficial articular hyaline cartilage in osteoarthritis. Arthritis

Rheum 2002;46(4):953-60.

30. Duvvuri U, Reddy R, Patel SD, Kaufman JH, Kneeland JB, Leigh JS. T1rho-relaxation in

articular cartilage: Effects of enzymatic degradation. Magn Reson Med 1997

Dec;38(6):863-7.

31. Duvvuri U, Goldberg AD, Kranz JK, Hoang L, Reddy R, Wehrli FW, Wand AJ,

Englander SW, Leigh JS. Water magnetic relaxation dispersion in biological systems:

The contribution of proton exchange and implications for the noninvasive detection of

cartilage degradation. Proc Natl Acad Sci U S A 2001 Oct 23;98(22):12479-84.

32. Li X, Cheng J, Lin K, Saadat E, Bolbos RI, Jobke B, Ries MD, Horvai A, Link TM,

Majumdar S. Quantitative MRI using T1rho and T2 in human osteoarthritic cartilage

specimens: Correlation with biochemical measurements and histology. Magn Reson

Imaging 2011 Apr;29(3):324-34.

33. Akella SVS, Regatte RR, Wheaton AJ, Borthakur A, Reddy R. Reduction of residual

dipolar interaction in cartilage by spin-lock technique. Magnetic Resonance in Medicine

2004;52(5):1103-9.

34. Akella SV, Regatte RR, Borthakur A, Reddy R. T1ρ relaxation rate and dispersion of

cartilage at 1.5 and 4T. Proc Intl Soc Mag Reson Med 2003;10:1503.

35. van Tiel J, Kotek G, Reijman M, Bos PK, Bron EE, Klein S, Nasserinejad K, van Osch,

Gerjo J. V. M., Verhaar JAN, Krestin GP, et al. Is T1ρ mapping an alternative to delayed

gadolinium-enhanced MR imaging of cartilage in the assessment of sulphated

glycosaminoglycan content in human osteoarthritic knees? an in vivo validation study.

Radiology 2015 11/20; 2015/12:150693.

36. Mlynarik V, Szomolanyi P, Toffanin R, Vittur F, Trattnig S. Transverse relaxation

mechanisms in articular cartilage. J Magn Reson 2004 Aug;169(2):300-7.

37. Regatte RR, Akella SV, Borthakur A, Reddy R. Proton spin-lock ratio imaging for

quantitation of glycosaminoglycans in articular cartilage. J Magn Reson Imaging 2003

Jan;17(1):114-21.

21

38. Jokivarsi KT, Niskanen J-, Michaeli S, Gröhn HI, Garwood M, Kauppinen RA, Gröhn

OH. Quantitative assessment of water pools by T1ρ and T 2ρ MRI in acute cerebral

ischemia of the rat. Journal of Cerebral Blood Flow and Metabolism 2009;29(1):206-16.

39. Blumenkrantz G, Li X, Han ET, Newitt DC, Crane JC, Link TM, Majumdar S. A

feasibility study of in vivo T1rho imaging of the intervertebral disc. Magn Reson

Imaging 2006 Oct;24(8):1001-7.

40. Li X, Han ET, Ma CB, Link TM, Newitt DC, Majumdar S. In vivo 3T spiral imaging

based multi-slice T(1rho) mapping of knee cartilage in osteoarthritis. Magn Reson Med

2005 Oct;54(4):929-36.

41. Li X, Pedoia V, Kumar D, Rivoire J, Wyatt C, Lansdown D, Amano K, Okazaki N, Savic

D, Koff MF, et al. Cartilage T1ρ and T2 relaxation times: Longitudinal reproducibility

and variations using different coils, MR systems and sites. Osteoarthritis and Cartilage

2015 12;23(12):2214-23.

42. Li X, Han ET, Busse RF, Majumdar S. In vivo T(1rho) mapping in cartilage using 3D

magnetization-prepared angle-modulated partitioned k-space spoiled gradient echo

snapshots (3D MAPSS). Magn Reson Med 2008 Feb;59(2):298-307.

43. Mosher TJ, Zhang Z, Reddy R, Boudhar S, Milestone BN, Morrison WB, Kwoh CK,

Eckstein F, Witschey WR, Borthakur A. Knee articular cartilage damage in

osteoarthritis: Analysis of MR image biomarker reproducibility in ACRIN-PA 4001

multicenter trial. Radiology 2011 Mar;258(3):832-42.

44. Witschey WR, Borthakur A, Elliott MA, Fenty M, Sochor MA, Wang C, Reddy R.

T1rho-prepared balanced gradient echo for rapid 3D T1rho MRI. J Magn Reson Imaging

2008 Sep;28(3):744-54.

45. Li X, Pai A, Blumenkrantz G, Carballido-Gamio J, Link T, Ma B, Ries M, Majumdar S.

Spatial distribution and relationship of T1rho and T2 relaxation times in knee cartilage

with osteoarthritis. Magn Reson Med 2009 Jun;61(6):1310-8.

46. Wright GA, Brittain JH, Stainsby JA. Preserving T1 or T2 contrast in magnetization

preparation sequences. Proc Intl Soc Mag Reson Med 1996;4:1474.

47. Wheaton AJ, Borthakur A, Reddy R. Application of the keyhole technique to T1rho

relaxation mapping. Journal of Magnetic Resonance Imaging 2003;18(6):745-9.

48. Feinberg DA, Setsompop K. Ultra-fast MRI of the human brain with simultaneous multi-

slice imaging. Journal of Magnetic Resonance (San Diego, Calif.: 1997) 2013

02/13;229:90-100.

49. Matzat SJ, McWalter EJ, Kogan F, Chen W, Gold GE. T2 relaxation time quantitation

differs between pulse sequences in articular cartilage. J Magn Reson Imaging 2015

Jul;42(1):105-13.

22

50. Lammentausta E, Multanen J, Nieminen MT. Differences in T2 values of knee cartilage

measured with different scanners. Proc Int Soc Mag Reson Med 2009;17:3986.

23

Figure 1. (a) Schematic diagram of adiabatic T1ρ and T2ρ sequences. The preparation block

consisted in a variable train of adiabatic full passage (AFP) pulses (HS1 or HS4), preceding a

FLASH readout. To measure T2ρ, an adiabatic half passage (AHP) was placed prior to the

AFP train and a reverse AHP after the AFP pulses, before the imaging readout. (b) Articular

cartilage ROI segmentation and nomenclature.

24

Figure 2. R1ρ and R2ρ dependence on agarose concentration at different Tp (6, 8, 10 and 12 ms)

for HS1 (a) and HS4 (b). R2 is also shown for reference (red dots).

25

Figure 3. T1ρ maps (a) and T2ρ maps (b) of three representative agarose gel tubes (2.0, 2.25 and

2.75 % w/v).

26

Figure 4. R1ρ and R2ρ versus chondroitin sulfate concentration at Tp 6, 8, 10 and 12 ms for HS1

(a) and HS4 (b). R2 (red circles) is shown for reference.

27

Figure 5. T1ρ,HS1 (a), T1ρ,HS4 (b), T2ρ,HS1 (c) and T2ρ,HS4 (d) relaxation time constant maps of

articular cartilage

28

Figure 6. T1ρ and T2ρ maps and mean relaxation time values in femoral and tibial cartilage of

a representative asymptomatic subject and three patients with radiographic osteoarthritis

(KL=2-4). In the representative subject, mean T1ρ and T2ρ of the asymptomatic volunteer group

are shown.

29

Table 1. Relaxivity (r) as a function of agarose concentration and chondroitin sulfate concentration with 95% confidence interval (C.I.), coefficient of determination (R2) and root mean square

deviation (RMSD). Values affected by artifacts resulting from violation of the adiabatic condition are marked with a gray background.

AGAROSE CHONDROITIN SULFATE

PULSE

TYPE

Tp

[ms]

r

[%-1 s-1] 95% C.I. R2 RMSD

r

[%-1 s-1] 95% C.I. R2 RMSD

r1ρ HS1 3 27.06 (13.86, 40.27) 0.705 18.666

6 2.29 (1.81, 2.78) 0.926 0.689 0.26 (0.09, 0.42) 0.889 0.333

8 1.99 (1.82, 2.16) 0.987 0.243 0.17 (0.12, 0.21) 0.980 0.089

10 2.19 (2.00, 2.37) 0.988 0.260 0.18 (0.14, 0.22) 0.984 0.083

12 2.35 (2.13, 2.58) 0.984 0.319 0.19 (0.15, 0.23) 0.986 0.082

HS4 3 6.66 (4.53, 8.80) 0.847 3.020

6 4.20 (3.83, 4.56) 0.987 0.516 0.25 (0.20, 0.30) 0.989 0.095

8 4.79 (4.37, 5.21) 0.987 0.593 0.28 (0.23, 0.34) 0.989 0.108

10 5.25 (4.73, 5.77) 0.983 0.731 0.30 (0.25, 0.36) 0.990 0.112

12 5.33 (4.85, 5.81) 0.986 0.683 0.31 (0.25, 0.37) 0.990 0.116

r2ρ HS1 3 28.05 (16.82, 39.27) 0.780 15.868

6 5.37 (4.39, 6.36) 0.944 1.393 0.48 (-0.04, 1.00) 0.738 1.042

8 7.81 (6.93, 8.70) 0.978 1.249 0.63 (0.52, 0.74) 0.991 0.214

10 7.08 (6.44, 7.73) 0.986 0.906 0.62 (0.51, 0.73) 0.991 0.219

12 6.61 (5.86, 7.35) 0.978 1.054 0.61 (0.51, 0.72) 0.992 0.207

HS4 3 12.61 (7.58, 17.64) 0.781 7.114

6 6.91 (6.21, 7.62) 0.982 1.002 0.55 (0.48, 0.63) 0.995 0.149

8 6.76 (6.14, 7.38) 0.985 0.876 0.54 (0.47, 0.60) 0.996 0.129

10 6.46 (5.84, 7.08) 0.984 0.881 0.52 (0.46, 0.59) 0.996 0.127

12 6.36 (5.77, 6.95) 0.985 0.830 0.51 (0.43, 0.59) 0.993 0.159

r2 6.71 (6.05, 7.37) 0.983 0.935 0.50 (0.47, 0.53) 0.999 0.053

30

Table 2. In vivo T1ρ measurements (ms) of articular cartilage using different TRs. Values that demonstrated significant TR dependence are marked with a gray background.

PULSE

TYPE ROI

TR = 1.0 S TR = 2.5 S TR = 4.0 S TR = 6.0 S

SBJ #1 SBJ #2 SBJ #1 SBJ #2 SBJ #1 SBJ #2 SBJ #1 SBJ #2

HS1 femur 73.0 75.1 88.3 95.9 90.0 99.0 92.1 104.4

tibia 64.2 75.0 78.2 91.7 79.7 97.1 80.1 94.9

HS4 femur 53.1 51.0 63.1 65.1 67.2 66.0 67.3 70.2

tibia 46.3 51.9 58.9 65.5 59.5 64.9 62.3 65.1

ROI = Region of interest. SBJ = subject.

31

Table 3. In vivo T1ρ and T2ρ measurements (ms) of articular cartilage with different pulse lengths. Values affected by artifacts resulting from violation of the adiabatic condition are marked with

a gray background.

PULSE

TYPE ROI

Tp = 3 ms Tp = 6 ms Tp = 8 ms Tp = 10 ms Tp = 12 ms

SBJ #1 SBJ #2 SBJ #1 SBJ #2 SBJ #1 SBJ #2 SBJ #1 SBJ #2 SBJ #1 SBJ #2

T1ρ HS1 femur 18.1 17.5 69.0 79.9 90.0 99.0 95.6 99.7 93.2 96.9

tibia 18.2 16.5 51.2 74.9 79.7 97.1 83.8 94.8 79.0 91.8

HS4 femur 42.0 43.8 67.2 66.0 64.6 63.4 62.1 61.9 62.8 63.8

tibia 34.6 37.9 59.5 64.9 58.5 60.3 56.7 56.7 54.7 56.6

T2ρ HS1 femur 21.8 17.1 44.2 37.4 55.3 37.9 54.0 40.9 58.4 41.5

tibia 27.2 15.1 49.0 30.4 64.0 38.1 60. 38.9 62.9 38.9

HS4 femur 30.3 26.9 48.4 42.3 50.3 45.2 60.5 46.4 64.1 48.0

tibia 34.4 34.4 54.8 38.9 53.3 35.1 64.5 40.8 67.5 41.1

ROI = Region of interest. SBJ = subject.

.

32

Table 4. Means and standard deviations (SD) of T1ρ, T2ρ and T2 relaxation time values (ms) in superficial and deep articular cartilage of human subjects (N = 10).

COMPARTMENT ROI

T1ρ T2ρ T2

HS1 HS4 HS1 HS4

SUPERFICIAL DEEP SUPERFICIAL DEEP SUPERFICIAL DEEP SUPERFICIAL DEEP SUPERFICIAL DEEP

LATERAL

a Tr 110.6 (11.0) 95.4 (13.0) 78.8 (7.4) 68.5 (10.5) 39.5 (7.0) 31.6 (6.2) 50.3 (4.5) 39.7 (4.4) 60.0 (6.3) 45.1 (6.9)

p Tr 110.8 (11.1) 85.4 (9.1) 76.9 (5.0) 61.6 (8.2) 40.0 (6.2) 33.6 (7.3) 48.0 (7.8) 40.8 (3.9) 53.4 (6.8) 34.8 (11.4)

a CF 116.3 (19.4) 83.3 (11.2) 80.5 (11.2) 59.8 (10.0) 44.9 (9.1) 36.4 (8.1) 51.5 (13.9) 44.1 (9.9) 54.5 (7.4) 35.5 (7.1)

p CF 105.3 (12.6) 84.5 (5.0) 74.7 (7.9) 60.7 (8.0) 40.3 (6.8) 40.7 (9.5) 50.1 (6.0) 47.8 (5.5) 56.5 (7.6) 43.7 (8.6)

a PF 104.1 (11.5) 76.7 (13.6) 71.1 (11.6) 52.9 (7.2) 35.5 (9.3) 38.8 (7.4) 41.0 (5.1) 41.5 (6.7) 53.6 (12.2) 40.7 (7.5)

p PF 87.6 (29.3) 70.3 (30.0) 60.3 (20.7) 45.2 (15.8) 30.2 (11.9) 34.8 (14.7) 35.5 (12.8) 37.3 (13.8) 44.9 (6.3) 35.7 (5.5)

AT 92.5 (12.2) 78.5 (20.8) 66.3 (10.2) 56.2 (8.2) 35.2 (6.5) 31.5 (9.0) 42.9 (8.4) 37.3 (4.7) 49.5 (11.8) 31.0 (4.1)

CT 92.3 (12.3) 64.6 (11.0) 66.4 (9.3) 42.2 (2.6) 31.4 (4.1) 24.7 (7.7) 40.5 (5.8) 26.8 (11.8) 42.0 (5.9) 26.0 (5.0)

PT 97.3 (11.7) 75.0 (13.8) 68.7 (11.8) 52.3 (7.7) 36.1 (8.1) 29.3 (5.9) 45.6 (6.6) 36.9 (4.7) 52.7 (5.2) 37.1 (5.1)

MEDIAL

p Tr 90.1 (32.7) 84.9 (32.5) 65.4 (32.7) 61.5 (29.8) 35.8 (17.7) 38.3 (17..8) 44.6 (21.0) 44.0 (20.4) 62.7 (33.3) 54.5 (29.0)

a CF 105.6 (13.8) 87.7 (11.8) 78.3 (11.0) 64.7 (9.0) 37.1 (6.7) 34.8 (6.4) 47.7 (4.9) 44.9 (7.0) 59.1 (10.3) 37.6 (9.6)

p CF 103.1 (7.9) 82.6 (5.5) 73.3 (5.8) 62.8 (7.6) 37.3 (6.2) 38.5 (8.2) 47.9 (4.6) 48.9 (10.5) 52.3 (6.3) 34.5 (7.5)

a PF 101.3 (11.8) 77.5 (7.1) 71.0 (10.0) 57.2 (7.9) 34.9 (5.8) 35.7 (5.3) 43.5 (4.3) 39.6 (9.4) 54.1 (4.9) 39.7 (9.0)

p PF 92.0 (30.3) 76.3 (26.9) 64.9 (22.8) 54.5 (19.0) 31.5 (11.2) 32.6 (11.3) 39.1 (12.8) 36.9 (12.8) 46.5 (4.7) 35.3 (8.7)

AT 79.6 (35.7) 84.7 (49.6) 55.6 (25.6) 61.7 (35.6) 32.5 (16.4) 34.9 (11.3) 32.2 (15.2) 38.8 (22.4) 44.3 (23.8) 34.4 (18.8)

CT 113.7 (12.3) 71.5 (9.7) 87.7 (9.9) 48.1 (5.5) 42.8 (7.7) 34.9 (20.3) 51.6 (5.0) 33.8 (6.6) 56.3 (6.6) 26.8 (4.5)

PT 105.5 (15.0) 78.0 (11.4) 77.0 (10.9) 54.0 (8.9) 37.1 (9.1) 31.0 (5.4) 46.2 (7.6) 37.5 (5.2) 52.4 (6.9) 34.0 (6.6)

ROI = Region of interest. a = anterior; p = posterior; Tr = troclea; CF = central femur; PF = posterior femur. AT = anterior tibia; CT = central tibia; PT = posterior tibia.