Abstract
Background:
Three-dimensional MR fingerprinting (3D-MRF) techniques have been recently described for simultaneous multiparametric mapping of knee cartilage. However, investigation of repeatability remains limited.
Purpose:
To assess the intra-day and inter-day repeatabilities of knee cartilage T1, T2, and T1ρ maps using a 3D-MRF sequence for simultaneous measurement.
Study Type:
Prospective.
Subjects:
14 healthy subjects (35.4±9.3 years, eight males), scanned on day one and day seven.
Field Strength/Sequence:
3T / 3D-MRF, T1, T2, and T1ρ maps
Assessment:
The acquisition of 3D-MRF cartilage (simultaneous acquisition of T1, T2, and T1ρ maps) were acquired using a dictionary pattern-matching approach. Conventional cartilage T1, T2 and T1ρ maps were acquired using variable flip angles and a modified 3D-Turbo-Flash sequence with different echo and spin-lock times, respectively, and were fitted using mono-exponential models. Each sequence was acquired on day one and day seven with two scans on each day.
Statistical Tests:
The mean and standard deviation for cartilage T1, T2, and T1ρ were calculated in five manually segmented regions of interest (ROIs), including lateral femur, lateral tibia, medial femur, medial tibia, and patella cartilages. Intra-subject and inter-subject repeatabilities were assessed using coefficient of variation (CV) and intra-class correlation coefficient (ICC), respectively, on the same day and among different days. Regression and Bland-Altman analysis were performed to compare maps between the conventional and 3D-MRF sequences.
Results:
The CV in all ROIs was lower than 7.4%, 8.4%, and 7.5% and the ICC was higher than 0.56, 0.51and 0.52 for cartilage T1, T2, and T1ρ, respectively. The MRF results had a good agreement with the conventional methods with a linear regression slope >0.61 and R2>0.59.
Conclusion:
The 3D-MRF sequence had high intra-subject and inter-subject repeatabilities for simultaneously measuring knee cartilage T1, T2, and T1ρ with good agreement with conventional sequences.
Keywords: Quantitative mapping, repeatability, MRF, knee cartilage
INTRODUCTION
Osteoarthritis (OA) is one of the most common and debilitating chronic joint diseases that results in a socio-economic burden worldwide (1). OA is primarily characterized by the progressive degeneration of articular cartilage and other joint structures (2). Conventional MRI techniques have been used to quantify cartilage volume and thickness loss during the advanced stages of OA (3). These include spin–lattice relaxation time (T1), spin–spin relaxation time (T2), and, more recently, spin–lattice relaxation time in the rotating frame (T1ρ), which have also been used to detect changes in cartilage composition and structure during the early stages of the disease prior to the onset of morphologic cartilage loss (4). Cartilage T2 and T1ρ relaxation time measurements, which have been shown to reliably distinguish between subjects with and without knee OA (5,6), are commonly performed together in OA research studies for evaluation of cartilage composition and structure (7). However, commonly used techniques to obtain these conventional quantitative MRI values acquire a single parameter at a time, thus requiring multiple acquisitions that result in long MRI examination times, which could be range from 10 to 30 minutes of scanning to obtain all T1, T2, and T1ρ maps.
Multiparametric mapping acquired from one sequence has been gaining popularity recently (8, 11–13, 19–20), since it intrinsically provides co-registration of parameter maps, with desirable less acquisition time and reduced cost in clinical and research settings. Moreover, simultaneous cartilage T1, T2, and T1ρ mapping of the knee has shown the feasibility to detect age-related changes in cartilage composition and structure, and to distinguish between subjects with and without early OA (12,13). MR fingerprinting (MRF) is a versatile, MR imaging technique that allows for simultaneous measurement of multiple quantitative MRI parameters using dynamic signal patterns (8), which can reduce the overall MRI examination duration. In MRF, pulse sequence parameters such as flip-angles (FAs) and repetition times (TRs) vary along the sequence, systematically adjusted to results in a unique signal evolution (SE) for each tissue type, depending on the tissue T1, T2, and T1ρ relaxation times. A pattern matching algorithm is then used to match the measured SE, also called MR fingerprints, with a pre-computed dictionary containing the predicted SEs for different combinations of T1, T2, and T1ρ.
A two-dimensional (2D) MRF framework has been proposed for simultaneous T1, T2, and T1ρ mapping of the articular cartilage of the knee joint (9). However, the 2D technique is sensitive to through-plane motion (10), and its scan time duration is still too long to be used in a routine clinical setting where complete anatomic coverage of the knee joint is desired. A three-dimensional (3D) MRF sequence has recently been developed for simultaneous T1, T2, and T1ρ mapping (11). However, only the intra-day repeatability of cartilage T1, T2, and T1ρ mapping has been documented (11). The repeatability of the 3D-MRF sequence on different days has not been previously investigated.
The purpose of this study was to investigate the repeatability of the simultaneously acquired T1, T2, and T1ρ measurements of the articular cartilage of the knee joint using the 3D-MRF technique in healthy volunteers and to compare them with conventional relaxation mapping methods.
METHODS AND MATERIALS
Healthy Volunteers
The study was performed in compliance with the Health Insurance Portability and Accountability Act (HIPAA) regulations and with approval from our Institutional Review Board. All subjects provided written informed consent prior to their participation in the study. The study group consisted of 14 healthy volunteers (mean 35.4±9.3 years, age range 19–62 years; eight males). For the recruitment of the healthy volunteers, the inclusion criteria include: (1) Skeletally mature subjects with age 18–65 years; (2) Subject willing and able to comply with the MRF-MRI protocol; (3) No pain/inflammation. The exclusion criteria include: (1) Having history of diagnosed osteoarthritis, rheumatoid arthritis and knee surgery; (2) Previous knee injury; (3) Contraindications to 3T whole body MRI scanners.
Study Design
Each subject was scanned for two repeated sessions, with one session acquired on day one (Day1) and one session on day seven (Day7). In each session, two repeated scans were performed on each day, resulting in a total of four knee scans acquired. Each scan consists of one 3D-MRF, one conventional T1 map, one conventional T2 map and one conventional T1ρ map acquisition. All scans were conducted on a 3.0 T scanner (MAGNETOM Prisma, Siemens Healthcare, Erlangen, Germany) using a 1Tx/15Rx-channel transmit-receive coil (Quality Electrodynamics, Mayville, OH). The acquisitions were performed in the sagittal plane, focusing on either the left or right knee of each participant, depending on their individual comfort without balancing the number of left or right knee. Foam padding was used to secure the knee within the coil to minimize subject motion. All subjects were required to be still while scanning with free breathing, and there was no time gap or rest between the two repeated sessions acquired on the same day.
3D-MRF Sequence Design and Dictionary Construction
The slab-selective 3D-MRF sequence quantified four parameters: T1, T2, and T1ρ, and transmit RF magnetic field (B1+) inhomogeneity (11). The sequence consisted of three components: 1) an adiabatic inversion pulse followed by two fast imaging with steady-state precession (FISP) segments, with 250 excitations each, varying from 0° to 20°, and from 0° to 60° for the first and second segments, that was used to encode T1 and T2 relaxation times; 2) two fast low-angle shot (FLASH) segments, similar to the FISP segments, that were used to encode T1 and B1+; and 3) a series of four FLASH segments, with 125 excitations, with variable flip angles, preceded by a T1ρ preparation module with different spin-lock time (TSL) durations (11), that was used to encode T1ρ relaxation time. The TSL durations were optimized using Cramer-Rao lower bounds (15). A self-compensated T1ρ preparation module was used to compensate for the effects of B0 and B1 inhomogeneity (26). A 3D golden-angle radial stack-of-stars was used for the MRF acquisition (11). The 3D-MRF sequence was acquired with the following imaging parameters: repetition time (TR) = 7.5 msec; echo time (TE) = 15 ms; inversion time (TI) = 15 ms; TSL = 0.5, 3.8, 9.8, 21.8 ms; flip angle (FA) = 0° to 20°.; slice thickness = 3 mm; voxel size = 0.7 × 0.7 × 3.0 mm; field of view = 140 mm × 140 mm; matrix size = 192 × 192; slice bandwidth = 490 Hz/px; number of slices = 18; frequency of spin lock =500Hz, and total scan time = 9 minutes. The number of 18 slices was chosen as a compromise between acquisition time and volume coverage, as 18 slices can cover the knee cartilage region for majority of people and keep the scan time less than 10 minutes.
A dictionary of simulated MR fingerprints based on extended phase graphs (EPG) was created with a T1 range of 50–3000 ms, T2 range of 2–200 ms, and T1ρ range of 2–200 ms in steps of 5% increments, and flip angle ranges of 30° to 90° in steps of 15° (16). The measured MR fingerprints and the simulated dictionary were compressed using singular value decomposition (SVD) to speed up the offline dictionary matching (17). Cartilage T1, T2, and T1ρ maps were created pixel-wisely using an iterative dictionary pattern-matching algorithm utilizing maximum correlation. All algorithms were implemented in MATLAB (R2022b, MathWorks, Natick, MA, USA).
Conventional Mapping Sequences and Deep-learning-based Fitting
Conventional cartilage T1, T2, and T1ρ maps of the knee joint were also acquired on all subjects for comparison with the 3D-MRF. A 3D-VIBE sequence was acquired with 10 different FAs, ranging from 1° to 19° in 2° steps to estimate T1 using Siemens inline mapping software. A T1ρ-prepared Turbo-FLASH (TFL) sequence was used to acquire multiple scans with variable TSLs including 0.5, 4.2, 9.8, and 21.8 ms to estimate T1ρ with a spin-lock frequency of 500Hz. A T2-prepared TFL sequence was used to acquire multiple scans to estimate T2 (18). This was done using the same parameters as in the TFL sequence but without a spin-lock pulse applied, i.e. 0Hz. All the imaging parameters are shown in Table 1. Cartilage T1, T2, and T1ρ maps were fitted pixel-wisely using mono-exponential models with a deep learning approach implemented in MATLAB (R2022b, MathWorks, Natick, MA, USA) (15).
Table 1:
Imaging parameters of the 3D-MRF sequence and the conventional T1, T2, and T1ρ mapping sequences.
| MRF | T1 | T2 | T1ρ | |
|---|---|---|---|---|
| Sequence | 3D/ T1ρMRF | 3D/VIBE-VFA | 3D/T2TFL | 3D/T1ρTFL |
| Scan plane | SAG | SAG | SAG | SAG |
| FOV(mm2) | 140 | 140 | 140 | 140 |
| TR(ms) | 7.5 | 7.3 | 1500 | 1500 |
| TE(ms) | 3.4 | 2.9 | 3.9 | 3.9 |
| TI(ms) | 15 | - | - | |
| TSL/T2 preparation (ms) | 0.5,3.8,9.8,21.8 | - | 0.5,3.8,9.8,21.8 | 0.5,3.8,9.8,21.8 |
| FA(°) | 0–60 | 1:2:19 | 8 | 8 |
| Slice Thickness(mm) | 3 | 3 | 3 | 3 |
| Number of slices | 18 | 36 | 36 | 36 |
| Matrix size | 192*192 | 192*192 | 192*192 | 192*192 |
| Receiver bandwidth(Hz/px) | 490 | 275 | 490 | 490 |
| TA(min:s) | 4:30/shot | 6:18 | 1:28/TE | 1:28/TSL |
Abbreviations: FOV, field of view; TR, repetition time; TE, echo time; TI, inversion time; TSL, spin-lock pulse duration; FA, flip angel; TA, acquisition time.
Cartilage Quantitative MRI Analysis
Regions of interest (ROIs) were manually drawn on each sagittal proton-density (PD)-weighted image of the knee around five regions (Figure 1): the articular cartilage of the lateral femur, lateral tibia, medial femur, medial tibia, and patella. All ROIs segmentation and quantitative mapping was performed by two independent readers (reader 1: X.Z., 5 years of experience in medical imaging; and reader 2: H.D.M., 3 years of experience in medical imaging); both under the supervision of a musculoskeletal radiologist with 20 years of experience (R.K) using the built-in Volume Segmenter in MATLAB (R2022b, MathWorks, Natick, MA, USA). Extra attention was used in segmenting the cartilage-subchondral bone interface to exclude any fat voxels in the ROIs. This process was performed for each subject to create a 3D cartilage contour of each articular surface. The 3D cartilage contours were superimposed over the cartilage T1, T2, and T1ρ maps to measure the mean relaxation times (T1, T2, and T1ρ) on each articular surface. Global cartilage T1, T2, and T1ρ relaxation times for all articular surfaces of the knee were also calculated.
Figure 1:
Representative manual segmented ROIs. LFC: lateral femur cartilage; LTC: lateral tibia cartilage; MFC: medial femur cartilage; MTC: medial tibia cartilage; PC: patella cartilage.
Statistical Analysis
Statistical analysis was performed using SAS software. For each subject’s scan, the mean and standard deviation (SD) for T1, T2, and T1ρ were measured on each ROI of the knee.
The statistical analysis was stratified by parameter (T1, T2, T1ρ) and the ROI (LFC, LTC, MFC, MTC, PC). The analysis was estimated using data from all four scans of each subject. We used restricted maximum likelihood (REML) estimation of variance components in a random effects regression model to estimate the inter-subject variance and the intra-day and inter-day components of the intra-subject variance. The model included an anonymized subject ID and the day of measurement (Day1 versus Day7) as random classification factors, with days nested within subjects to account for the fact that Day1 was not necessarily the same day for all subjects. For each component of the intra-subject variance (overall, intra-day, inter-day), the CV was estimated as the square root of the estimated variance component expressed as a percentage of the overall mean (27). The ICC was estimated using the inter-subject variance over the sum of the inter-subject variance and the intra-subject variance (28). Regression and Bland-Altman analysis were performed to compare T1, T2, and T1ρ measurements acquired using the MRF sequence and the conventional techniques. The Bland-Altman plots were also used to assess the intra-day and inter-day limits of agreement in repeated T1, T2, and T1ρ measurements from the 3D-MRF. The Bland-Altman plots demonstrate agreement by plotting the difference between paired measurements-with 95% confidence limits-against their mean. Smaller limits indicated better agreement among repeated measurements or different methods.
Results
Table 2 shows the inter-rater reliability agreement of the manual segmentation of the five ROIs between the two independent researchers. The overall ICC reaches 0.90 between two independent readers.
Table 2:
The inter-rater reliability agreement (ICC and 95% CIs) of the manual segmentation of the five ROIs between the two independent researchers.
| ROI | ICC | 95% CIs |
|---|---|---|
| MFC | 0.91 | [0.91, 0.91] |
| MTC | 0.91 | [0.91, 0.91] |
| LFC | 0.89 | [0.89, 0.89] |
| LTC | 0.91 | [0.91, 0.91] |
| PC | 0.90 | [0.90, 0.90] |
| Over All | 0.90 | [0.90, 0.90] |
Abbreviations: MFC, medial femur cartilage; MTC, medial tibia cartilage; LFC, lateral femur cartilage; LTC, lateral tibia cartilage; PC, patella cartilage; ROI, region of interest; ICC, intraclass correlation coefficient; CIs, confident interval.
Figure 2 shows one representative slice of the cartilage T1, T2, T1ρ and B1+ maps of the lateral femur and lateral tibia from the four scans performed on a representative subject. The multi-parameter maps showed high agreement for cartilage T1, T2, T1ρ and B1+ measurements across the four scans acquired one week apart. Figures 3 and 4 show comparisons of cartilage T1, T2, T1ρ and B1+ maps of the lateral femur and lateral tibia. These comparisons are presented for both a male and a female subject, both aged 35, as well as a young male subject and an older male subject.
Figure 2:
Representative cartilage PD, T1, T2, T1ρ, and B1+ maps for a subject with four repeated scans. All four scans were acquired on day one and day seven in the lateral knee. As shown in the figure, the maps display high agreement of the T1, T2, T1ρ, and B1+ maps in the ROIs across the four scans acquired one week apart.
Figure 3:
Representative cartilage PD, T1, T2, T1ρ and B1+ maps for a male subject (left) and female subject (right), both of whom are 35 years old, showcasing the knee joint.
Figure 4:
Representative cartilage PD, T1, T2, T1ρ, and B1+ maps for a 30-year-old subject (left) and a 62-year-old subject (right), both males in the knee joint.
Table 3 summarizes the overall mean and REML estimates of the inter-subject variance and the intra-day and inter-day components of the overall intra-subject variance. The variance of the inter-subject is higher than the intra-subject in most ROIs for T1 and for all for T1ρ. For T2, the trending is not consistent across ROIs between the inter-subject and intra-subject variance.
Table 3.
The overall mean and the REML estimates of the inter-subject variance and the intra-day and inter-day components of the overall intra-subject variance of each measure within each region of interest.
| Overall Mean±STD | Intra-Subject Variance | Inter-Subject Variance | ||||
|---|---|---|---|---|---|---|
| Measure | Region | Overall | Intra-Day | Inter-Day | ||
| T1 | LFC | 984.36 ± 69.13 | 3333.12 | 2412.4 | 534.2 | 4168.78 |
| LTC | 950.26 ± 91.05 | 3644.79 | 2263.42 | 1003.23 | 8642.89 | |
| MFC | 962.04 ± 59.30 | 1509.17 | 1091.25 | 301.3 | 4625 | |
| MTC | 916.25 ± 59.70 | 1959.87 | 1129.85 | 780.02 | 3754.7 | |
| PC | 947.26 ± 95.03 | 2109.21 | 1288.3 | 646.96 | 7887.52 | |
| T2 | LFC | 25.15 ± 4.39 | 4.29 | 4.29 | 0.00 | 4.82 |
| LTC | 21.29 ± 4.36 | 3.99 | 1.67 | 2.32 | 13.91 | |
| MFC | 25.88 ± 4.24 | 11.67 | 6.87 | 4.80 | 12.75 | |
| MTC | 21.50 ± 4.21 | 3.77 | 2.65 | 1.11 | 6.15 | |
| PC | 26.98 ± 5.93 | 6.23 | 5.18 | 1.05 | 6.52 | |
| T1ρ | LFC | 31.90 ± 2.29 | 6.33 | 4.55 | 1.79 | 7.72 |
| LTC | 27.71 ± 5.16 | 6.44 | 2.38 | 4.07 | 17.17 | |
| MFC | 39.92 ± 3.75 | 9.08 | 5.79 | 3.29 | 12.71 | |
| MTC | 26.69 ± 3.20 | 6.59 | 3.18 | 3.41 | 7.19 | |
| PC | 30.52 ± 2.63 | 8.78 | 7.60 | 1.19 | 10.04 | |
Abbreviations: MFC, medial femur cartilage; MTC, medial tibia cartilage; LFC, lateral femur cartilage; LTC, lateral tibia cartilage; PC, patella cartilage; STD, standard deviation; MAD, median absolute deviation.
Table 4 summarizes results of the coefficient of variation (CV) associated with each component of the intra-subject variance, an estimate of intra-class correlation (ICC) across the four scans performed on day one and day seven. The intra-day CV were 2.9%−6.1%, 4.8%−7.1%, and 4.3%−6.5% for cartilage T1, T2, and T1ρ measurements, respectively. The inter-day CV were 1.8%−5.2%,1.2–5.7%, and 2.7%−5.7% for cartilage T1, T2, and T1ρ measurements, respectively. The ICC was higher than 0.56, 0.51 and 0.52 for cartilage T1, T2, and T1ρ measurements, respectively.
Table 4.
The coefficient of variation (CV) associated with each component of the intra-subject variance, an estimate of intra-class correlation (ICC).
| Intra-Subject CV | ICC | ||||
|---|---|---|---|---|---|
| Measure | Region | Overall | Intra-Day | Inter-Day | |
| T1 | LFC | 5.2% | 4.5% | 1.8% | 0.56 |
| LTC | 7.4% | 6.1% | 5.2% | 0.72 | |
| MFC | 4.7% | 3.6% | 2.7% | 0.67 | |
| MTC | 4.5% | 2.9% | 2.6% | 0.69 | |
| PC | 4.7% | 3.8% | 3.1% | 0.79 | |
| T2 | LFC | 6.5% | 6.5% | 1.2% | 0.53 |
| LTC | 7.4% | 4.8% | 5.6% | 0.78 | |
| MFC | 8.4% | 7.0% | 5.7% | 0.52 | |
| MTC | 7.4% | 6.2% | 4.0% | 0.62 | |
| PC | 7.7% | 7.1% | 3.2% | 0.51 | |
| T1ρ | LFC | 5.5% | 5.1% | 3.2% | 0.55 |
| LTC | 7.1% | 4.3% | 5.7% | 0.73 | |
| MFC | 7.4% | 5.9% | 4.5% | 0.58 | |
| MTC | 7.5% | 5.2% | 5.4% | 0.52 | |
| PC | 7.2% | 6.5% | 2.7% | 0.53 | |
Abbreviations : MFC, medial femur cartilage; MTC, medial tibia cartilage; LFC, lateral femur cartilage; LTC, lateral tibia cartilage; PC, patella cartilage.
Figure 5 shows the Bland-Altman plots for the intra-day and inter-day repeatability of cartilage T1, T2, and T1ρ measurements acquired with the 3D-MRF sequence. For display purpose, the inter-day Bland-Altman plots were computed by the mean values of the two scans of each day. The mean difference between the two scans acquired on the same day was 7.66, −0.57, and −0.41 for cartilage T1, T2, and T1ρ measurements, respectively. The mean difference between the mean values of the two scans acquired on day one and day seven was −8.16, −0.52, and −0.96 for cartilage T1, T2, and T1ρ measurements, respectively. Only one subject exceeded the SD limits of 1.96 for cartilage T1 and T1ρ measurements, and two subjects for T2 measurement.
Figure 5:
Bland–Altman plots for intra-day comparison (top) and inter-day comparison (bottom) for the cartilage T1 (left), T2 (middle), and T1ρ (right) measurements estimated from the 3D-MRF. The blue dashed line indicates the average of the difference values; the red dashed line indicates the upper and lower limit of agreement of the 95% confidence; the black solid line indicates the difference value at 0. In the intra-day comparison (top), each point represents difference between the two scans acquired on the same day. In the inter-day comparison (bottom), each point represents the difference between the averages on day one and day seven, and the average of each day was calculated from the two scans on the same day.
Figure 6 shows the regression plots and Bland-Altman plots for the repeatability of cartilage T1, T2, and T1ρ measurements between the conventional and the 3D-MRF sequence. For the regression analysis, correlation coefficients (R2) between the conventional and 3D-MRF sequences were 0.59, 0.68, and 0.62, regression slops were 0.61, 0.66 and 0.66 for cartilage T1, T2, and T1ρ measurements, respectively. For the Bland-Altman analysis, the mean difference between the conventional and 3D-MRF sequences was 5.84, −0.80 and −2.03 for cartilage T1, T2, and T1ρ measurements, respectively. Two subjects exceeded the SD limits of 1.96 for cartilage T2 and T1ρ measurements and one subject for T1 measurement. There was a bias toward higher cartilage T1ρ measurements for the 3D-MRF sequence compared to the conventional sequence for all subjects.
Figure 6:
Regression plots (upper row) and Bland–Altman plots (lower row) for the comparison between the cartilage T1 (left), T2 (middle), and T1ρ (right) measurements estimated from the conventional sequences and 3D-MRF. In the Bland-Altman plot: the blue dashed line indicates the average of the difference values; the red dashed line indicates the upper and lower limit of agreement of the 95% confidence; the black solid line indicates the difference value at 0. Each data point on the plots represents an average of the measurements of the two scans on the same day.
Discussion
This study has documented the assessment of both intra-day and inter-day repeatabilities of knee cartilage over a seven-day period using a 3D-MRF sequence for simultaneous multi-parameter mapping. This sequence enabled the quantification of T1, T2, and T1ρ relaxation times in the knee cartilage within a clinically feasible scan time duration. The intra-subject repeatability showed high agreement between cartilage T1, T2, and T1ρ measurements acquired on the same day and different days in healthy subjects.
Inter-subject variance was slightly higher than that of intra-subject, which indicates that the inter-subject repeatability is lower than that of intra-subject in cartilage T1, T2, and T1ρ measurements, which is likely due to the large span of ages of subjects in our study. Previous studies have documented higher regional and global cartilage T2 and T1ρ values of the knee in the older subjects compared to younger subjects (13,20).
The intra-subject and inter-subject repeatability of T1ρ and T2 measurements in knee cartilage have been extensively studied (21–25). In many of these studies, a 3D-MAPSS sequence was typically used for acquiring T1ρ and T2 relaxation times (21–23). In the current work, the CVs for intra-subject repeatability of T1ρ measurements were found to be in the range of 4.3%−6.5% for measurements obtained on the same day. These values are slightly higher than the reported range of 1.60%−3.93% in a previous study (21). However, it is important to note that the assessment of intra-subject repeatability in that study was only conducted on five knees (22).
For inter-day repeatability, the CVs for T1ρ measurements obtained in this study were between 2.7%−5.7%, compared to the CVs reported in literature that ranged from 2.6% to 4.86% (21). The ICC of T1ρ measurements were in the range 0.52–0.73 in this study contrasting with the reported range of 0.20–0.93 observed in a study involving 10 subjects aged 18 to 35 years (23). The CVs for intra-subject repeatability of T2 measurements were 4.8%−7.1% for intra-day in this study, compared to the previously reported CVs of 4.00%−5.00% (24) and 1.40–4.08% (22), which included only five knees.
For inter-day, the CVs for T2 measurements were 0.0%−5.7% in this study, compared to the reported CVs of 3.33%−7.25% (21). The ICC for inter-subject repeatability for T2 measurements were 0.51–0.78 in our study, compared to CVs of 0.61–0.98 observed in the study with 10 subjects aged 18–35 years (23).
A previous study explored the SNR and parameter values when using a different number of shots for this 3D-MRF sequence (11). In that study, the result indicates that an SNR of at least 100 is required to have reproducible measurements of T1ρ and it also shows that T1ρ is the most sensitive parameter to SNR, which explains why the maps shown in our study appear noisier than the maps for T1 and T2.
Regarding the higher bias observed for T1ρ compared to the conventional sequence, we speculated that this is due to the way the T1ρ is encoded in the 3D-MRF sequence. By including its preparation at the end of the sequence and using smaller flip angles, the signal intensity will be lower.
Overall, the findings of this work concerning both intra-subject and inter-subject repeatability of T2 and T1ρ measurements were comparable to previously reported results. The observed differences are likely due to the different sample sizes and the age ranges of the subjects under investigation.
Limitations
The relatively small subject cohort, which may limit the generalizability of the conclusion. In addition, the subject’s cohort of this study did not include individuals with knee OA, who may have lower repeatability of T2 and T1ρ measurements than healthy subjects due to the greater difficulty in segmenting the thinner articular cartilage of the knee joint. Another limitation was that this was a single-center and single-vendor study. Different MRI scanner sites and vendors have been shown to have different measurements of knee cartilage T2 and T1ρ relaxation times in phantoms and the same human subjects (21–24). Thus, future studies are needed to investigate the variations and effects caused by the use of different types of MRI scanners or the use of the same MRI scanner at different locations when acquiring multi-parameter cartilage maps of the knee joint simultaneously using the 3D-MRF sequence.
CONCLUSION
This study has demonstrated high inter-subject and intra-subject repeatabilities in the measurement of cartilage T1, T2, and T1ρ relaxation times within the knee joint of healthy subjects using the 3D-MRF sequence. The acquired measurements using the 3D-MRF sequence exhibited good agreement with those obtained from conventional T1, T2, and T1ρ mapping sequences. The 3D-MRF sequence has the potential to provide a more time-efficient comprehensive evaluation of cartilage composition and structure in subjects with knee OA, allowing to better characterize disease severity and monitor treatment response.
Grant Support:
This study was supported by National Institutes of Health (NIH) grants R01 AR076328, R01 AR076985, and R01-AR078308-01A1, performed under the Center of Advanced Imaging Innovation and Research (CAI2R) at NYU Grossman School of Medicine and NIBIB Biomedical Technology Resource Center (NIH P41 EB017183).
References
- 1.Vos T, Barber RM, Bell B, Bertozzi-Villa A, Biryukov S, Bolliger I, Charlson F, Davis A, Degenhardt L, Dicker D, Duan L. Global, regional, and national incidence, prevalence, and years lived with disability for 301 acute and chronic diseases and injuries in 188 countries, 1990–2013: a systematic analysis for the Global Burden of Disease Study 2013. The lancet. 2015. Aug 22;386(9995):743–800. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Li X, Benjamin Ma C, Link TM, et al. In vivo T1ρ and T2 mapping of articular cartilage in osteoarthritis of the knee using 3 T MRI. Osteoarthr Cartil. 2007;15(7):789–797. doi: 10.1016/j.joca.2007.01.011 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Gray ML, Eckstein F, Peterfy C, Dahlberg L, Kim YJ, Sorensen AG. Toward imaging biomarkers for osteoarthritis. Clin Orthop Relat Res. 2004;427(427): S175–S181. doi: 10.1097/01.blo.0000144972.50849.d9 [DOI] [PubMed] [Google Scholar]
- 4.Guermazi A, Alizai H, Crema MD, Trattnig S, Regatte RR, Roemer FW. Compositional MRI techniques for evaluation of cartilage degeneration in osteoarthritis. Osteoarthr Cartil 2015;23(10):1639–1653 [DOI] [PubMed] [Google Scholar]
- 5.Atkinson HF, Birmingham TB, Moyer RF, et al. MRI T2 and T1ρ relaxa- tion in patients at risk for knee osteoarthritis: A systematic review and meta-analysis. BMC Musculoskelet Disord 2019;20(1):182. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.MacKay JW, Low SBL, Smith TO, Toms AP, McCaskie AW, Gilbert FJ. Systematic review and meta-analysis of the reliability and discriminative validity of cartilage compositional MRI in knee osteoarthritis. Osteoarthr Cartil 2018;26(9):1140–1152. [DOI] [PubMed] [Google Scholar]
- 7.Li X, Kuo D, Theologis A, et al. Cartilage in anterior cruciate ligament– reconstructed knees: MR imaging T1ρ and T2—initial experience with 1-year follow-up. Radiology 2011;258(2):505–514. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Ma D, Gulani V, Seiberlich N, et al. Magnetic resonance fingerprinting. Nature. 2013;495(7440):187–192. doi: 10.1038/nature11971 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Sharafi A, Zibetti MVW, Chang G, Cloos M, Regatte RR. MR fingerprinting for rapid simultaneous T1, T2, and T1ρrelaxation mapping of the human articular cartilage at 3T. Magn Reson Med.2020;84(5):2636–2644. doi: 10.1002/mrm.28308 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Yu Z, Zhao T, Assländer J, Lattanzi R, Sodickson DK, Cloos MA. Exploring the sensitivity of magnetic resonance fingerprinting to motion. Magn. Reson. Imaging 2018;54:241–248 doi: 10.1016/j.mri.2018.09.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Sharafi A, Zibetti MVW, Chang G, Cloos M, Regatte RR. 3D MR- fingerprinting for rapid simultaneous T1, T2, and T1ρ volumetric map- ping of the human articular cartilage at 3T. NMR Biomed 2022;34(7): e4800. 10.1002/nbm.4531. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Sharafi A, Zibetti MVW, Chang G, Cloos MA, Regatte RR. Simultaneous bilateral T(1), T(2), and T(1ρ) relaxation mapping of the hip joint with magnetic resonance fingerprinting. NMR Biomed 2022;35:35. 10.1002/nbm.4651. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Kijowski R, Sharafi A, Zibetti MV, Chang G, Cloos MA, Regatte RR. Age-Dependent Changes in Knee Cartilage T1, T2, and T1 p Simultaneously Measured Using MRI Fingerprinting. Journal of Magnetic Resonance Imaging. 2023. Jun;57(6):1805–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Menon RG, Sharafi A, Muccio M, Smith T, Kister I, Ge Y, Regatte RR. Three-dimensional multi-parameter brain mapping using MR fingerprinting. [Google Scholar]
- 15.Zibetti MVW, Sharafi A, Regatte RR. Optimization of spin-lock times in T1ρ mapping of knee cartilage: Cramer-Rao bounds versus matched sampling-fitting. Magn Reson Med 2022;87(3):1418–1434. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Weigel M. Extended phase graphs: dephasing, RF pulses, and echoes - pure and simple. J Magn Reson Imaging 41, 266–295 (2015). 10.1002/jmri.2461952. [DOI] [PubMed] [Google Scholar]
- 17.McGivney DF et al. SVD Compression for Magnetic Resonance Fingerprinting in the Time Domain. Ieee Transactions on Medical Imaging 33, 2311–2322 (2014). 10.1109/Tmi.2014.2337321 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Sharafi A, Chang G, Regatte RR. Biexponential T2 relaxation estimation of human knee cartilage in vivo at 3T.J Magn Reson Imaging.2018;47(3):809–819. doi: 10.1002/jmri.25778 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Sharafi Azadeh, et al. “Simultaneous T1, T2, and T1ρ relaxation mapping of the lower leg muscle with MR fingerprinting.” Magnetic resonance in medicine 86.1 (2021): 372–381. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Zibetti Marcelo VW, et al. “Updates on Compositional MRI Mapping of the Cartilage: Emerging Techniques and Applications.” Journal of Magnetic Resonance Imaging (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Lartey R, Nanavati A, Kim J, Li M, Xu K, Nakamura K, Shin W, Winalski CS, Obuchowski N, Bahroos E, Link TM. Reproducibility of T1ρ and T2 quantification in a multi-vendor multi-site study. Osteoarthritis and Cartilage. 2023. Feb 1;31(2):249–57. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Kim J, Mamoto K, Lartey R, Xu K, Nakamura K, Shin W, Winalski CS, Obuchowski N, Tanaka M, Bahroos E, Link TM. Multi-vendor multi-site T1ρ and T2 quantification of knee cartilage. Osteoarthritis and cartilage. 2020. Dec 1;28(12):1539–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.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. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.MacKay JW, Low SB, Smith TO, Toms AP, McCaskie AW, Gilbert FJ. Systematic review and meta-analysis of the reliability and discriminative validity of cartilage compositional MRI in knee osteoarthritis. Osteoarthritis and cartilage. 2018. Sep 1;26(9):1140–52. T1p and T2, CV 4%−5% intra-subject on the same day, same vendor [DOI] [PubMed] [Google Scholar]
- 25.Chalian M, Li X, Guermazi A, Obuchowski NA, Carrino JA, Oei EH, Link TM, RSNA QIBA MSK Biomarker Committee. The QIBA profile for MRI-based compositional imaging of knee cartilage. Radiology. 2021. Nov;301(2):423–32. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Pang Y. A self-compensated spin-locking scheme for quantitative R1ρ dispersion MR imaging in ordered tissues. Magnetic Resonance Imaging. 2022. Dec 1;94:112–8. [DOI] [PubMed] [Google Scholar]
- 27.Abdi H. Coefficient of variation . Encyclopedia of research design. 2010. Jun;1(5). [Google Scholar]
- 28.Rodriguez GG, Yu Z, O′ Donnell LF, Calderon L, Cloos MA, Madelin G. Repeatability of simultaneous 3D 1H MRF/23Na MRI in brain at 7 T. Scientific Reports. 2022. Aug 19;12(1):14156. [DOI] [PMC free article] [PubMed] [Google Scholar]