Repeatability and reproducibility of in vivo magnetic resonance T1rho relaxation time measurements of hip cartilage at 3T
Angéline Nemeth1, Lucy Di Marco2, Denis Grenier1, Michaël Sdika1, Olivier Beuf1, and Jean-Baptiste Pialat3

1CREATIS, Université de Lyon ; CNRS UMR5220 ; Inserm U1044 ; INSA-Lyon ; Université Claude Bernard Lyon 1, Lyon, France, 2Radiologie et Imagerie médicale diagnostique et thérapeutique, Hôpital François Mitterrand, Dijon, France, 3Radiologie Pavillon B, Hôpital Edouard Herriot, Hospices Civils de Lyon, INSERM U1033 and Université Lyon 1, Lyon, France

### Synopsis

To detect early changes in cartilage, quantification methods were developed with Magnetic Resonance Imaging. T1rho sequence is a valuable tool to quantitatively access the proteoglycans content, complementary to T2 mapping technique that is correlated with the collagen content. The thin cartilage thickness and deep location of the hip joint require a strong compromise between SNR, pixel size and acquisition time. The aim of this study is to test the repeatability and reproducibility of in vivo (magnetic resonance) T1rho and T2 relaxation times measures in the hip joint cartilage of thirty healthy asymptomatic volunteers.

### Introduction

Qualitative degradation of the collagen matrix and the proteoglycans content of the cartilage is precursory of morphologic cartilage loss. To detect early changes, quantification methods were developed. T1rho sequence is a valuable tool to quantitatively access the proteoglycans content, complementary to T2 maps measures that are correlated with the collagen content1,2. Studies on specimens and in vivo assessment of several articulations including the knee were reported3, but very few studies focus on hip cartilage. Yet, the hip is the second most common large joint affected by osteoarthritis leading to joint arthroplasty. Hip imaging is challenging because of technical limitations caused by the deepness and thinness of the hip cartilage. However, interesting data was found on the knee cartilage, and improvement of the performance of 3T-MR magnets encourages further research on the field of quantitative assessment of hip cartilage.

### Purpose

To test the repeatability and reproducibility of in vivo magnetic resonance T1rho and T2 relaxation time measurements in the hip joint cartilage.

### Subjects and Methods

The right hip joint of nineteen healthy asymptotic volunteers (nine women and ten men) was explored. Exams were performed at 3T (GE MR750 3T Discovery) using a 32 channels abdominal coil. Subjects were aged 31.1 ± 7.4 years with a BMI of 22.25 ± 2.4 kg/m². Subjects were placed in supine position with feet positioned in adduction. Subjects first underwent a 3D water excitation spoiled gradient-echo (SPGR) acquisition, then a 3D T1rho weighted sequence and a 3D multiple TE for T2 mapping (Table 1, Figure 1). T1rho and multiple TE sequences were repeated 30 min after the beginning of the exam and the entire exam was repeated for all subjects 14 ± 7 days later. The cartilage was automatically segmented into 8 regions (Figure 2.B), and 3 regions were preferentially studied (R3: posterior region, R4: superior region, R5: antero-superior). The automatic segmentation was implemented on GNU Octave (version 3.6.2). A circular Hough transform was used to find a sphere encompassing the femoral head. The cartilage was then segmented by thresholding voxels within this sphere. The cartilage subregions were delimited by dividing the cartilage into 8 uniform angular sectors (see Figure 2). Femoral and acetabular cartilages were segmented as a single unit on SPGR images and full cartilage thickness was estimated. For T1rho-maps (as for T2-maps), a mono-exponential fit was performed [$S(TSL)=S_{0}*exp(-TSL/T1_{rho})$]. A determination coefficient R² was calculated to evaluate the fit quality. Only pixels with R² > 0.8 were kept. A mask resulting from segmentation on SPGR was transformed in T1rho- or T2-space using fsl-flirt4 and merged with respective maps (Figure 3). Repeatability of T1rho- and T2-values was assessed comparing the two acquisitions performed with 30min delay and reproducibility comparing two acquisitions with 14 ± 7 days delay. A paired t-test was performed on mean value and the coefficient of variation CV was calculated for region R3, R4, R5 using [$CV=(\frac{\sqrt{\sum_j^m SD^{2}_{j}/m}}{\sum_j^m \overline{x_{j}}/m}*100)$] with m the number of subjects, $\overline{x}_{j}$ and $SD_j$ the mean and standard deviation of repeated measurements5.

### Results

Subjects had a mean cartilage thickness of 2.90 ± 0.51 mm, for the region R3, 3.13 ± 0.75 mm for R4 and 3.64 ± 0.73 mm for R5. Due to subject motion during the one hour acquisition time, only data from 14 subjects were exploitable. Results of T1rho and T2 are summarized in Table 2. Mean SNR of T1rho weighted images was 15 ± 4.3 for the shorter TSL. Only 40 ± 15% pixels respected the criterion R²>0.80 in segmented cartilage (mean region size equal to 1 cm3) and contribute to the calculation of T1rho values.

### Discussion

The thin cartilage thickness and deep location of the hip joint require a strong compromise between SNR, pixel size and acquisition time. Mean T1-rho values in this study were higher than in the literature5, 6 but coherent.This could be due to mono-exponential fit that could provide overestimated values at low SNR7. According to the paired t-test, values measured repeatedly within the same exam or at 14 days interval were not significantly different (except for R3). Carballido-Gamio and al. 6 have established on 5 subjects the reproducibility of T1rho measurement with CV of 2.03%. In this work, CV ranged between 5.63% and 9.75% for T1rho measurements (5.09% and 10.44% for T2). Re-positioning the patient had no effect on T1rho and T2 values. Eleven more subjects will be included in this study for final results.

### Acknowledgements

This work was performed within the framework of the LABEX PRIMES (ANR-11-LABX-0063) of Université de Lyon, within the program "Investissements d'Avenir" (ANR-11-IDEX-0007) operated by the French National Research Agency (ANR).

### References

1Nishioka H, Hirose J, Nakamura E, et al. T1ρ and T2 mapping reveal the in vivo extracellular matrix of articular cartilage. J Magn Reson Imaging 2012;35:147-155

2Tsushima H, Okazaki K, Takayama Y, et al. Evaluation of cartilage degradation in arthritis using T1ρ magnetic resonance imaging mapping. Rheumatol Int. sept 2012;32(9):2867-2875.

3Li X, Benjamin Ma C, Link TM, et al. In vivo T1ρ and T2 mapping of articular cartilage in osteoarthritis of the knee using 3T MRI. Osteoarthr Cartilage. 2007;15(7):789-797.

4Jenkinson MetSmithSM A global optimisation method for robust affine registration of brain images. Medical Image Analysis. 2001;5(2):143-156.

5Carballido-Gamio J, Link TM, Li X, et al. Feasibility and reproducibility of relaxometry, morphometric, and geometrical measurements of the hip joint with magnetic resonance imaging at 3T. J MagnReson Imaging 2008;28:227–235.

6Wyatt C, Kumar D, Subburaj K, Lee S, et al. Cartilage T1ρ and T2 Relaxation Times in Patients With Mild-to-Moderate Radiographic Hip Osteoarthritis. Arthritis Rheumatol. 2015;67(6):1548-1556.

7Raya JG, Dietrich O, Horng A, et al. T2 measurements in articular cartilage : impact of the fitting method on accuracy and precision at low SNR. Magn Reson in Med. 2010;63 :181-193

### Figures

Table 1: Summary of parameters of 3D water excitation spoiled gradient-echo (SPGR) acquisition, 3D T1rho weighted acquisition (T1rho) and 3D multiple TE acquisition (T2).

Figure 1: Hip cartilage MRI images from : 3D water excitation spoiled gradient-echo (SPGR) acquisition (A), 3D T1rho weighted acquisition (B) and 3D T2 weighted acquisition (C).

Figure 2: Different steps of the automatic segmentation on 3D water excitation spoiled gradient-echo (SPGR). A, result of segmentation after edge detection and Hough’s transform to detect the center of femoral head. B, definition of 8 regions. C, final result of segmentation

Figure 3: Steps to assess T1rho-values in different regions of cartilage

Table 2: Summary of mean T1rho and T2 values, paired t-test p-values and CV with respect to the region at the different acquisition times.

Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
4490