Original ResearchFree Access

Magnetization-prepared 2 Rapid Gradient-Echo MRI for B1 Insensitive 3D T1 Mapping of Hip Cartilage: An Experimental and Clinical Validation

Published Online:https://doi.org/10.1148/radiol.2021200085

Abstract

Background

Often used for T1 mapping of hip cartilage, three-dimensional (3D) dual-flip-angle (DFA) techniques are highly sensitive to flip angle variations related to B1 inhomogeneities. The authors hypothesized that 3D magnetization-prepared 2 rapid gradient-echo (MP2RAGE) MRI would help provide more accurate T1 mapping of hip cartilage at 3.0 T than would 3D DFA techniques.

Purpose

To compare 3D MP2RAGE MRI with 3D DFA techniques using two-dimensional (2D) inversion recovery T1 mapping as a standard of reference for hip cartilage T1 mapping in phantoms, healthy volunteers, and participants with hip pain.

Materials and Methods

T1 mapping at 3.0 T was performed in phantoms and in healthy volunteers using 3D MP2RAGE MRI and 3D DFA techniques with B1 field mapping for flip angle correction. Participants with hip pain prospectively (July 2019–January 2020) underwent indirect MR arthrography (with intravenous administration of 0.2 mmol/kg of gadoterate meglumine), including 3D MP2RAGE MRI. A 2D inversion recovery–based sequence served as a T1 reference in phantoms and in participants with hip pain. In healthy volunteers, cartilage T1 was compared between 3D MP2RAGE MRI and 3D DFA techniques. Paired t tests and Bland-Altman analysis were performed.

Results

Eleven phantoms, 10 healthy volunteers (median age, 27 years; range, 26–30 years; five men), and 20 participants with hip pain (mean age, 34 years ± 10 [standard deviation]; 17 women) were evaluated. In phantoms, T1 bias from 2D inversion recovery was lower for 3D MP2RAGE MRI than for 3D DFA techniques (mean, 3 msec ± 11 vs 253 msec ± 85; P < .001), and, unlike 3D DFA techniques, the deviation found with MP2RAGE MRI did not correlate with increasing B1 deviation. In healthy volunteers, regional cartilage T1 difference (109 msec ± 163; P = .008) was observed only for the 3D DFA technique. In participants with hip pain, the mean T1 bias of 3D MP2RAGE MRI from 2D inversion recovery was −23 msec ± 31 (P < .001).

Conclusion

Compared with three-dimensional (3D) dual-flip-angle techniques, 3D magnetization-prepared 2 rapid gradient-echo MRI enabled more accurate T1 mapping of hip cartilage, was less affected by B1 inhomogeneities, and showed high accuracy against a T1 reference in participants with hip pain.

© RSNA, 2021

Summary

Compared with three-dimensional (3D) dual-flip-angle acquisition, 3D magnetization-prepared 2 rapid acquisition gradient-echo MRI enabled more accurate T1 mapping of hip cartilage, was less affected by B1 inhomogeneities, and showed high accuracy against a T1 reference in participants with hip pain.

Key Results

  • ■ In phantoms, hip cartilage maps obtained with B1 insensitive magnetization-prepared 2 rapid gradient-echo (MP2RAGE) MRI showed less deviation (mean, 3 msec ± 11 [standard deviation] vs 253 msec ± 85; P < .001) in T1 from the two-dimensional (2D) inversion recovery reference compared with dual flip angle (DFA) acquisition.

  • ■ In healthy volunteers, no variation in cartilage T1 relaxation was observed over the field of view for three-dimensional (3D) MP2RAGE MRI (mean difference, 8 msec ± 46; P = .45) compared with 3D DFA acquisition (mean difference, 109 msec ± 163; P = .008).

  • ■ In participants with hip pain, T1 relaxation assessed with 3D MP2RAGE MRI showed a small and unlikely to be clinically important mean underestimation of −23 msec ± 31 (P < .001) from 2D inversion recovery reference.

Introduction

Because of the rise in the number of hip joint–preserving procedures performed, there has been an increasing demand to improve surgical decision making based on the status of cartilage degeneration (1). Although standard morphologic MRI is accurate in diagnosing labrum tears (2,3), the evaluation of early cartilage damage remains elusive (4,5). By contrast, quantitative MRI techniques, such as delayed gadolinium-based contrast-enhanced MRI of cartilage (dGEMRIC) can be used as a marker for cartilage quality (6). The dGEMRIC is validated against histologic findings and is used most frequently in the hip (79), with promising results in predicting hip preservation after periacetabular osteotomy (10), outcome after correction of femoroacetabular impingement (11), and the natural course in femoroacetabular impingement (12). The most common technique for three-dimensional (3D) T1 mapping of the hip for dGEMRIC is based on a fast dual-flip-angle (DFA) gradient-echo method (10,11,1315). This method was introduced at 1.5 T (16) but is susceptible to B1-related flip angle variations that increase at 3.0 T (17). To limit these flip angle variations, B1 mapping methods were introduced (17,18). Despite applying a B1 correction for dGEMRIC at 3.0 T using the DFA technique, we observed considerable T1 variations in our clinical practice. Magnetization-prepared 2 rapid gradient-echo (MP2RAGE) MRI was modified to generate two different images at different inversion times (TIs) for neuroimaging. The images were combined to obtain high-spatial-resolution 3D T1 maps with minimized B1 inhomogeneity, even at 7.0 T (19). We hypothesized that 3D MP2RAGE MRI would enable a more accurate T1 mapping of hip cartilage than the 3D DFA technique at 3.0 T.

The aim of this study was threefold. First, we compared the 3D MP2RAGE technique and 3D DFA acquisition using the two-dimensional (2D) inversion recovery technique as the standard of reference in phantoms with varying T1 values. Second, we assessed whether the 3D MP2RAGE technique would yield T1 maps with less variability in precontrast T1 relaxation of cartilage and periarticular muscles than the DFA method in healthy volunteers. Third, we compared T1 measurements of articular cartilage with the 3D MP2RAGE technique against the standard of reference 2D inversion recovery technique in participants with hip pain undergoing indirect MR arthrography for dGEMRIC.

Materials and Methods

This prospective study was performed under institutional review board approval, and written informed consent was obtained. Two authors (M.K. and T.K.) are full employees of Siemens Healthcare in Switzerland. The remaining authors certify that neither he or she, nor any member of his or her immediate family, have funding or commercial associations (consultancies, stock ownership, equity interest, patent or licensing arrangements, etc) that might pose a conflict of interest in connection with the article. Each author certifies that his or her institution approved the reporting of this investigation and that all investigations were conducted in conformity with ethical principles of research.

3D MP2RAGE Sequence for T1 Mapping

We developed a hip protocol for the MP2RAGE sequence, as it was previously only optimized for gray matter and white matter differentiation in the brain (19). Because the number of phase-encoding steps is higher in phase direction (113 steps, including 40% phase oversampling, 6/8 partial Fourier transform in phase and section encoding direction, parallel imaging [generalized autocalibrating partial parallel acquisition {GRAPPA; Siemens Healthineers}, ×2]) compared with section direction (90 steps), all phase-encoding steps in phase direction were acquired per TI. This means that one partition (phase-encoding step in section direction) was acquired per repetition time to reduce acquisition time. To optimize the accuracy of T1 measurements within a range of expected cartilage T1 values between 300 msec and 1100 msec (11,13), the TIs and flip angle values for both MP2RAGE images (Table 1) were determined according to the simulations as described by Marques et al (19). This included 3D MP2RAGE MRI with TIs optimized for high cartilage T1 (TI msec, 700 [first TI]/2200 [second TI]) in unenhanced imaging and for low cartilage T1 (400 [first TI]/2500 [second TI]) in dGEMRIC (Table 1).

Table 1: Sequence Protocol Used for Phantoms, Healthy Volunteers, and Participants with Hip Pain

Table 1:

For the 3D DFA technique, sequence parameters were adapted as previously described for 3.0 T (11,13). The B1 field strength was mapped using an additional sequence in which a spin echo and a stimulated echo were generated using a 90°-90°-90° radiofrequency pulsing scheme (20). All images were acquired with a 3.0-T scanner (Magnetom Skyra; Siemens Healthcare) with sequence parameters summarized in Table 1.

Phantom Study

For the phantom study, a 15-channel transmit-receive knee coil was used (Siemens Healthcare). Eleven phantoms were filled with solutions of water-diluted contrast material (gadoterate meglumine) to obtain a range of T1 values, as follows: 1, 0.6, 0.5, 0.4, 0.34, 0.30, 0.26, 0.22, 0.18, 0.16, and 0.14 mmol/L (Fig 1). As a T1 reference, a 2D single-section fast spin-echo inversion recovery sequence based on an adiabatic inversion pulse was performed (21,22). Fitting was performed with an in-house–developed software according to the method described by Barral et al (20). An additional B1 mapping sequence with high spatial resolution was performed to assess a potential correlation between B1 deviations and T1 variations of the 3D DFA sequence with B1 correction (18) and the 3D MP2RAGE sequence (TI, 700 [first TI]/2200 [second TI]) from the reference 2D inversion recovery T1. Regions of interest were manually placed for each of the 11 vials to measure T1 for the 2D inversion recovery, 3D MP2RAGE (TI, 700 [first TI]/2200 [second TI]), 3D MP2RAGE (TI, 400 [first TI]/2500 [second TI]), 3D DFA, and B1 deviations.

T1 maps of 11-compartment phantom with different gadolinium                        concentrations. T1 map of two-dimensional (2D) inversion recovery (IR)                        technique, which served as T1 reference, and T1 maps of three-dimensional                        (3D) dual-flip-angle (DFA) method and 3D magnetization-prepared 2 rapid                        gradient-echo (MP2RAGE) technique are shown.

Figure 1: T1 maps of 11-compartment phantom with different gadolinium concentrations. T1 map of two-dimensional (2D) inversion recovery (IR) technique, which served as T1 reference, and T1 maps of three-dimensional (3D) dual-flip-angle (DFA) method and 3D magnetization-prepared 2 rapid gradient-echo (MP2RAGE) technique are shown.

Volunteer Population

Ten healthy volunteers without a history of hip disease were recruited. Volunteers underwent imaging of both hips (3D MP2RAGE [TI, 700 {first TI}/2200 {second TI}] and 3D DFA sequence) at 3.0 T using an array 18-channel flexible-body surface coil (Table 1). For the 3D DFA technique, a previously validated B1 correction based on a preconditioning radiofrequency pulse with turbo fast low-angle shot readout was applied with parameters as shown in Table 1 and in the study by Chung et al (18). Before the T1 measurements, all scans were reviewed by a musculoskeletal radiologist (S.D.B., with 19 years of experience) for cartilage damage and degenerative changes of bone and surrounding muscles, which did not reveal any abnormalities.

For T1 measurements, 12 radial images were reformatted perpendicular to the femoral head-neck junction, and regions of interest, including both articular cartilage layers, were manually placed at each “full hour” position using the greater trochanter as a landmark for the 12 o’clock position according to a previously validated technique (11,2325). For comparison, clockface positions of 1 o’clock to 5 o’clock and 12 o’clock to 7 o’clock were clustered as mean anterior and posterior cartilage regions, respectively.

T1 of periarticular muscles was measured to estimate anteroposterior (sartorius muscle and gluteus maximus muscle) and mediolateral (internal obturator muscle and vastus lateralis muscle) B1 inhomogeneity over the field of view (Fig 2a) on a radial image at the 3 o’clock and 9 o’clock positions. Calculating B1 deviation was based on two steps. First, the region of interest for measuring mean relative flip angle accuracy was placed over the central section of the hip on the B1 field map. This was performed using the lower flip angle images (Fig 2b, solid circle) as an anatomic reference to locate the hip joint region and to define the region of interest on the B1 map (Fig 2b, dashed circle). Selection of a large region of interest covering the entire joint was necessary because the used B1 correction requires a large field of view and low image matrix, yielding an in-plane resolution too low to exclusively measure flip angle accuracy of the thin hip cartilage layers. Second, B1 deviation was calculated by subtracting the measured mean relative flip angle accuracy from the optimal desired flip angle, which was set at 100%, as follows: flip angle accuracy of 100%−measured mean flip angle accuracy. For example, assuming that a measured mean flip angle accuracy of 100% would imply a full accordance with the desired optimal flip angle of 100%, the resulting B1 deviation would be 0%. In the example shown in Figure 2b, the measured mean flip angle accuracy was 73.9%, corresponding to a mean B1 deviation of 26.1% (100%−73.9% = 26.1%) (Fig 2b).

Images in 25-year-old healthy volunteer undergoing unenhanced T1                        mapping of hip. (a) Reformation of radial images from three-dimensional                        magnetization-prepared 2 rapid gradient-echo MRI (inversion time msec,                        700/2200) and region of interest placement. Radial images are aligned with                        femoral neck to obtain perpendicular sections through hip joint. Greater                        trochanter serves as landmark for 12 o’clock position. Regions of                        interest were placed for T1 measurements of anterior cartilage (Ac),                        posterior cartilage (Pc), and for periarticular muscles, as follows:                        sartorius (S), gluteus maximus (Gm), vastus lateralis (Vl), and obturator                        internus (Oi). (b) Calculation of B1 deviation was based on two steps.                        First, region of interest for measuring mean relative flip angle (FA)                        accuracy (FAaccuracy) was placed over central section of hip on B1 field map                        (right image). This was performed using lower FA image (left image) as                        anatomic reference to locate hip joint region (solid circle) and to place                        region of interest on B1 map (dashed circle). Second, B1 deviation was                        calculated by subtracting measured mean relative FA accuracy from optimal                        desired FA, which is set at 100% (FA100%). In example shown, measured mean                        FA accuracy was 73.9%, corresponding to mean B1 deviation of                        26.1%.

Figure 2a: Images in 25-year-old healthy volunteer undergoing unenhanced T1 mapping of hip. (a) Reformation of radial images from three-dimensional magnetization-prepared 2 rapid gradient-echo MRI (inversion time msec, 700/2200) and region of interest placement. Radial images are aligned with femoral neck to obtain perpendicular sections through hip joint. Greater trochanter serves as landmark for 12 o’clock position. Regions of interest were placed for T1 measurements of anterior cartilage (Ac), posterior cartilage (Pc), and for periarticular muscles, as follows: sartorius (S), gluteus maximus (Gm), vastus lateralis (Vl), and obturator internus (Oi). (b) Calculation of B1 deviation was based on two steps. First, region of interest for measuring mean relative flip angle (FA) accuracy (FAaccuracy) was placed over central section of hip on B1 field map (right image). This was performed using lower FA image (left image) as anatomic reference to locate hip joint region (solid circle) and to place region of interest on B1 map (dashed circle). Second, B1 deviation was calculated by subtracting measured mean relative FA accuracy from optimal desired FA, which is set at 100% (FA100%). In example shown, measured mean FA accuracy was 73.9%, corresponding to mean B1 deviation of 26.1%.

Images in 25-year-old healthy volunteer undergoing unenhanced T1                        mapping of hip. (a) Reformation of radial images from three-dimensional                        magnetization-prepared 2 rapid gradient-echo MRI (inversion time msec,                        700/2200) and region of interest placement. Radial images are aligned with                        femoral neck to obtain perpendicular sections through hip joint. Greater                        trochanter serves as landmark for 12 o’clock position. Regions of                        interest were placed for T1 measurements of anterior cartilage (Ac),                        posterior cartilage (Pc), and for periarticular muscles, as follows:                        sartorius (S), gluteus maximus (Gm), vastus lateralis (Vl), and obturator                        internus (Oi). (b) Calculation of B1 deviation was based on two steps.                        First, region of interest for measuring mean relative flip angle (FA)                        accuracy (FAaccuracy) was placed over central section of hip on B1 field map                        (right image). This was performed using lower FA image (left image) as                        anatomic reference to locate hip joint region (solid circle) and to place                        region of interest on B1 map (dashed circle). Second, B1 deviation was                        calculated by subtracting measured mean relative FA accuracy from optimal                        desired FA, which is set at 100% (FA100%). In example shown, measured mean                        FA accuracy was 73.9%, corresponding to mean B1 deviation of                        26.1%.

Figure 2b: Images in 25-year-old healthy volunteer undergoing unenhanced T1 mapping of hip. (a) Reformation of radial images from three-dimensional magnetization-prepared 2 rapid gradient-echo MRI (inversion time msec, 700/2200) and region of interest placement. Radial images are aligned with femoral neck to obtain perpendicular sections through hip joint. Greater trochanter serves as landmark for 12 o’clock position. Regions of interest were placed for T1 measurements of anterior cartilage (Ac), posterior cartilage (Pc), and for periarticular muscles, as follows: sartorius (S), gluteus maximus (Gm), vastus lateralis (Vl), and obturator internus (Oi). (b) Calculation of B1 deviation was based on two steps. First, region of interest for measuring mean relative flip angle (FA) accuracy (FAaccuracy) was placed over central section of hip on B1 field map (right image). This was performed using lower FA image (left image) as anatomic reference to locate hip joint region (solid circle) and to place region of interest on B1 map (dashed circle). Second, B1 deviation was calculated by subtracting measured mean relative FA accuracy from optimal desired FA, which is set at 100% (FA100%). In example shown, measured mean FA accuracy was 73.9%, corresponding to mean B1 deviation of 26.1%.

Participants with Hip Pain

This cohort reflects a nonconsecutive series of patients eligible for joint-preserving hip surgery from the greater Boston area who attended a tertiary academic center between July 2019 and January 2020. Participants were diagnosed with hip pain by senior orthopedic hip surgeons according to the presence of structural hip deformities, most commonly developmental dysplasia of the hip and femoroacetabular impingement deformities, and a history of and clinical findings for hip pain. Inclusion criteria were participants undergoing subsequent indirect MR arthrography for dGEMRIC. Among 21 participants, one needed to be excluded because of poor image quality.

Imaging was performed at 3.0 T using an array 18-channel flexible-body surface coil after intravenous injection of a double-dose solution of 0.2 mmol/kg gadoterate meglumine and subsequent ambulation for 30 minutes. In addition to morphologic standard sequences (2D proton-density-weighted turbo spin-echo and 3D double-echo steady-state images [26]), we acquired a midaxial oblique 2D inversion recovery image as a cartilage T1 reference and an axial-oblique 3D MP2RAGE image (TI, 400 [first TI]/2500 [second TI]) for dGEMRIC (Table 1). Radiographic evaluation was performed by a senior orthopedic hip surgeon (E.N.N., with 17 years of experience) who was blinded to the MRI results according to the 2020 Lisbon agreement on femoroacetabular impingement imaging (25).

For T1 measurements, the 2D inversion recovery image and the corresponding 3D MP2RAGE section were assessed independently from each other. For T1 measurements of 3D MP2RAGE MRI and 3D DFA techniques, cartilage borders were identified on 3D double-echo steady-state images and divided into equally sized peripheral and central regions of interest of the anterior and posterior opposing cartilage layers, leading to four regions of interest per participant (Fig 3).

Images in 37-year-old woman with hip pain undergoing indirect MR                        arthrography of hip. Midaxial oblique two-dimensional (2D) inversion                        recovery (IR) image was acquired, and corresponding section from                        three-dimensional (3D) magnetization-prepared 2 rapid gradient-echo                        (MP2RAGE) MRI (inversion time msec, 400/2500) was identified. For T1                        measurements, opposing cartilage layers were equally divided into anterior                        and posterior using line (solid line) passing through acetabulum and center                        of femoral head (black-outlined dot), which was determined with a perfectly                        fitting circle (circle). Regions of interest were further divided into two                        equally sized parts (dashed lines) to differentiate between peripheral (per)                        and central (cen) regions of interest, which were placed using morphologic                        3D double-echo steady-state (DESS) images as reference to identify cartilage                        borders, anteriorly and posteriorly. Accordingly, four regions of interest                        were measured in each participant for 3D MP2RAGE MRI and 2D inversion                        recovery.

Figure 3: Images in 37-year-old woman with hip pain undergoing indirect MR arthrography of hip. Midaxial oblique two-dimensional (2D) inversion recovery (IR) image was acquired, and corresponding section from three-dimensional (3D) magnetization-prepared 2 rapid gradient-echo (MP2RAGE) MRI (inversion time msec, 400/2500) was identified. For T1 measurements, opposing cartilage layers were equally divided into anterior and posterior using line (solid line) passing through acetabulum and center of femoral head (black-outlined dot), which was determined with a perfectly fitting circle (circle). Regions of interest were further divided into two equally sized parts (dashed lines) to differentiate between peripheral (per) and central (cen) regions of interest, which were placed using morphologic 3D double-echo steady-state (DESS) images as reference to identify cartilage borders, anteriorly and posteriorly. Accordingly, four regions of interest were measured in each participant for 3D MP2RAGE MRI and 2D inversion recovery.

Image Analysis

All quantitative 3D DFA, 3D MP2RAGE, and 2D inversion recovery MRI scans were analyzed by a radiology resident experienced in quantitative imaging of the hip joint (F.S., with 6 years of experience). This was followed by a repeat analysis after 3 months to evaluate intrareader reliability.

Statistical Analysis

Parametric data were reported as means ± standard deviations, and nonparametric data were reported as medians with interquartile ranges (first quartile–third quartile). The Spearman rank correlation coefficient (rs) was used for correlation among the T1 mapping techniques, B1 maps, gadolinium concentrations, and morphologic MRI score. Bland-Altman analysis was performed between the T1 mapping methods and the 2D inversion recovery T1 reference. The mean T1 of cartilage and periarticular muscles was compared (3D MP2RAGE MRI [TI, 700 {first TI}/2200 {second TI}] vs 3D DFA techniques) using paired t tests after testing for normal distribution using the Kolmogorov-Smirnov test. To assess variation in T1 between participants, we calculated variances of the sample mean, including the ratio of variances (ratio of variance at DFA to variance at MP2RAGE) and compared them using the Pitmann-Morgan test (27,28). Intrareader reliability was assessed with intraclass correlation coefficients. P < .05 indicated a statistically significance difference. SPSS software was used (version 25).

Results

Volunteer and Participant Characteristics

Ten healthy participants (median age, 27 years; range, 26–30 years; five men) were included. After exclusion of one participant because of poor image quality, 20 participants with hip pain (mean age, 34 years ± 10 [standard deviation]; 17 women) were included for the clinical validation of the 3D MP2RAGE MRI. Fifteen participants with hip pain (75%) presented with developmental dysplasia of the hip, and six participants underwent periacetabular osteotomy (Table 2).

Table 2: Demographic and Radiographic Characteristics of Participants with Hip Pain

Table 2:

Mapping in Phantoms

Almost perfect (3D DFA technique: rs = 0.99) to perfect (3D MP2RAGE MRI, 2D inversion recovery: rs = 1) correlations (all P < .001) were observed between T1 mapping techniques and gadolinium concentration. Almost perfect (rs = 0.99) to perfect (rs = 1) correlation (all P < .001) was observed between the 3D DFA, 3D MP2RAGE (TI, 700 [first TI]/2200 [second TI]), and 2D inversion recovery techniques, respectively. In contrast to the 3D MP2RAGE MRI (mean difference, 3 msec ± 11; P = .24), we found a significant systematic bias for the 3D DFA technique (mean, 253 msec ± 85; P = .003) compared with the 2D inversion recovery technique (Fig 4). This bias was lower for MP2RAGE MRI than for 3D DFA techniques (mean, 3 msec ± 11 vs 253 msec ± 85; P < .001).

(a) Bland-Altman plot shows systematic bias of three-dimensional (3D)                        dual-flip-angle (DFA) technique compared with the two-dimensional (2D)                        inversion recovery (IR) technique (mean difference, 253 msec ± 85                        [standard deviation]; 95% CI: 195, 310; P = .003) but not for                        magnetization-prepared 2 rapid gradient-echo (MP2RAGE) technique (inversion                        time msec, 700/2200) (mean difference, 3 msec ± 11; 95% CI: 4, 10; P                        = .24). Solid black and gray lines represent mean difference for 3D DFA                        and 3D MP2RAGE techniques compared with 2D IR technique, respectively. Plot                        shows a strong linear correlation for greater T1 deviation from 2D IR                        technique for 3D DFA technique with increasing T1. (b, c) B1 deviation                        plotted on x-axis and relative differences in T1 of 3D DFA (b) and 3D                        MP2RAGE (c) techniques compared with 2D IR technique are shown on y-axis.                        Relative difference in T1 using 3D DFA technique compared with 2D IR                        technique increases strongly with increasing B1 deviation. This was not the                        case for 3D MP2RAGE technique.

Figure 4a: (a) Bland-Altman plot shows systematic bias of three-dimensional (3D) dual-flip-angle (DFA) technique compared with the two-dimensional (2D) inversion recovery (IR) technique (mean difference, 253 msec ± 85 [standard deviation]; 95% CI: 195, 310; P = .003) but not for magnetization-prepared 2 rapid gradient-echo (MP2RAGE) technique (inversion time msec, 700/2200) (mean difference, 3 msec ± 11; 95% CI: 4, 10; P = .24). Solid black and gray lines represent mean difference for 3D DFA and 3D MP2RAGE techniques compared with 2D IR technique, respectively. Plot shows a strong linear correlation for greater T1 deviation from 2D IR technique for 3D DFA technique with increasing T1. (b, c) B1 deviation plotted on x-axis and relative differences in T1 of 3D DFA (b) and 3D MP2RAGE (c) techniques compared with 2D IR technique are shown on y-axis. Relative difference in T1 using 3D DFA technique compared with 2D IR technique increases strongly with increasing B1 deviation. This was not the case for 3D MP2RAGE technique.

(a) Bland-Altman plot shows systematic bias of three-dimensional (3D)                        dual-flip-angle (DFA) technique compared with the two-dimensional (2D)                        inversion recovery (IR) technique (mean difference, 253 msec ± 85                        [standard deviation]; 95% CI: 195, 310; P = .003) but not for                        magnetization-prepared 2 rapid gradient-echo (MP2RAGE) technique (inversion                        time msec, 700/2200) (mean difference, 3 msec ± 11; 95% CI: 4, 10; P                        = .24). Solid black and gray lines represent mean difference for 3D DFA                        and 3D MP2RAGE techniques compared with 2D IR technique, respectively. Plot                        shows a strong linear correlation for greater T1 deviation from 2D IR                        technique for 3D DFA technique with increasing T1. (b, c) B1 deviation                        plotted on x-axis and relative differences in T1 of 3D DFA (b) and 3D                        MP2RAGE (c) techniques compared with 2D IR technique are shown on y-axis.                        Relative difference in T1 using 3D DFA technique compared with 2D IR                        technique increases strongly with increasing B1 deviation. This was not the                        case for 3D MP2RAGE technique.

Figure 4b: (a) Bland-Altman plot shows systematic bias of three-dimensional (3D) dual-flip-angle (DFA) technique compared with the two-dimensional (2D) inversion recovery (IR) technique (mean difference, 253 msec ± 85 [standard deviation]; 95% CI: 195, 310; P = .003) but not for magnetization-prepared 2 rapid gradient-echo (MP2RAGE) technique (inversion time msec, 700/2200) (mean difference, 3 msec ± 11; 95% CI: 4, 10; P = .24). Solid black and gray lines represent mean difference for 3D DFA and 3D MP2RAGE techniques compared with 2D IR technique, respectively. Plot shows a strong linear correlation for greater T1 deviation from 2D IR technique for 3D DFA technique with increasing T1. (b, c) B1 deviation plotted on x-axis and relative differences in T1 of 3D DFA (b) and 3D MP2RAGE (c) techniques compared with 2D IR technique are shown on y-axis. Relative difference in T1 using 3D DFA technique compared with 2D IR technique increases strongly with increasing B1 deviation. This was not the case for 3D MP2RAGE technique.

(a) Bland-Altman plot shows systematic bias of three-dimensional (3D)                        dual-flip-angle (DFA) technique compared with the two-dimensional (2D)                        inversion recovery (IR) technique (mean difference, 253 msec ± 85                        [standard deviation]; 95% CI: 195, 310; P = .003) but not for                        magnetization-prepared 2 rapid gradient-echo (MP2RAGE) technique (inversion                        time msec, 700/2200) (mean difference, 3 msec ± 11; 95% CI: 4, 10; P                        = .24). Solid black and gray lines represent mean difference for 3D DFA                        and 3D MP2RAGE techniques compared with 2D IR technique, respectively. Plot                        shows a strong linear correlation for greater T1 deviation from 2D IR                        technique for 3D DFA technique with increasing T1. (b, c) B1 deviation                        plotted on x-axis and relative differences in T1 of 3D DFA (b) and 3D                        MP2RAGE (c) techniques compared with 2D IR technique are shown on y-axis.                        Relative difference in T1 using 3D DFA technique compared with 2D IR                        technique increases strongly with increasing B1 deviation. This was not the                        case for 3D MP2RAGE technique.

Figure 4c: (a) Bland-Altman plot shows systematic bias of three-dimensional (3D) dual-flip-angle (DFA) technique compared with the two-dimensional (2D) inversion recovery (IR) technique (mean difference, 253 msec ± 85 [standard deviation]; 95% CI: 195, 310; P = .003) but not for magnetization-prepared 2 rapid gradient-echo (MP2RAGE) technique (inversion time msec, 700/2200) (mean difference, 3 msec ± 11; 95% CI: 4, 10; P = .24). Solid black and gray lines represent mean difference for 3D DFA and 3D MP2RAGE techniques compared with 2D IR technique, respectively. Plot shows a strong linear correlation for greater T1 deviation from 2D IR technique for 3D DFA technique with increasing T1. (b, c) B1 deviation plotted on x-axis and relative differences in T1 of 3D DFA (b) and 3D MP2RAGE (c) techniques compared with 2D IR technique are shown on y-axis. Relative difference in T1 using 3D DFA technique compared with 2D IR technique increases strongly with increasing B1 deviation. This was not the case for 3D MP2RAGE technique.

The mean relative bias in T1 between the 3D DFA and 3D MP2RAGE techniques compared with the 2D inversion recovery T1 reference was 36.5% ± 11.6 and 0.2% ± 1.3, respectively. The mean B1 deviation was 21.1% ± 4.2. The T1 discrepancy between the 3D DFA and the 2D inversion recovery techniques increased strongly (rs = 0.69, P = .009), with increasing B1 deviation (Fig 4).

Mapping in Healthy Volunteers

The mean cartilage T1 values (range, 1434–1543 msec) were consistently higher (all P < .001) for the 3D DFA technique compared with the 3D MP2RAGE technique (TI, 700 [first TI]/2200 [second TI]; range, 1028–1036 msec) (Table 3). The regional difference between anterior and posterior cartilage T1 was observed for the DFA technique (mean difference, 109 msec ± 163; P = .008) but not for the 3D MP2RAGE technique (mean difference, 8 msec ± 46; P = .45). Variances were consistently higher (all P < .001) for the 3D DFA technique compared with the 3D MP2RAGE technique (eg, overall mean cartilage T1, 1489 msec ± 178 vs 1032 msec ± 40; ratio of variances, 20; P < .001), with the ratio of variances ranging from nine to 29 (Table 3). The mean regional difference in periarticular muscle T1 (internal obturator muscle vs vastus lateralis muscle) was higher (P < .001) for the 3D DFA technique (406 msec ± 192) compared with the 3D MP2RAGE technique (25 msec ± 25) (Table 3). Variances for all four muscles were higher (all P < .001) for the 3D DFA technique compared with the 3D MP2RAGE technique (Table 3). The mean B1 deviation was 18.4% ± 3.8.

Table 3: Comparison between 3D DFA and 3D MP2RAGE Techniques in Healthy Volunteers

Table 3:

Mapping in Participants with Hip Pain

A systematic underestimation in T1 between the 3D MP2RAGE technique (TI, 400 [first TI]/2500 [second TI]) and the 2D inversion recovery technique of −23 msec ± 31 (P < .001) was observed. The mean relative difference was −4% ± 5.3 (Fig 5).

Bland-Altman plots compare two-dimensional inversion recovery with                        three-dimensional (3D) magnetization-prepared 2 rapid gradient-echo                        (MP2RAGE) (inversion time msec, 400/2500) in (a) phantoms and (b)                        participants with hip pain. In phantoms, no systematic bias between the two                        techniques was observed (mean difference, 3 msec ± 7 [standard                        deviation]; P = .11). In participants with hip pain undergoing indirect                        MR arthrography, a systematic bias was observed (mean difference, −23                        msec ± 31; P < .001). Solid lines represent mean difference (MD),                        and dashed lines represent 95% limits of agreement.

Figure 5a: Bland-Altman plots compare two-dimensional inversion recovery with three-dimensional (3D) magnetization-prepared 2 rapid gradient-echo (MP2RAGE) (inversion time msec, 400/2500) in (a) phantoms and (b) participants with hip pain. In phantoms, no systematic bias between the two techniques was observed (mean difference, 3 msec ± 7 [standard deviation]; P = .11). In participants with hip pain undergoing indirect MR arthrography, a systematic bias was observed (mean difference, −23 msec ± 31; P < .001). Solid lines represent mean difference (MD), and dashed lines represent 95% limits of agreement.

Bland-Altman plots compare two-dimensional inversion recovery with                        three-dimensional (3D) magnetization-prepared 2 rapid gradient-echo                        (MP2RAGE) (inversion time msec, 400/2500) in (a) phantoms and (b)                        participants with hip pain. In phantoms, no systematic bias between the two                        techniques was observed (mean difference, 3 msec ± 7 [standard                        deviation]; P = .11). In participants with hip pain undergoing indirect                        MR arthrography, a systematic bias was observed (mean difference, −23                        msec ± 31; P < .001). Solid lines represent mean difference (MD),                        and dashed lines represent 95% limits of agreement.

Figure 5b: Bland-Altman plots compare two-dimensional inversion recovery with three-dimensional (3D) magnetization-prepared 2 rapid gradient-echo (MP2RAGE) (inversion time msec, 400/2500) in (a) phantoms and (b) participants with hip pain. In phantoms, no systematic bias between the two techniques was observed (mean difference, 3 msec ± 7 [standard deviation]; P = .11). In participants with hip pain undergoing indirect MR arthrography, a systematic bias was observed (mean difference, −23 msec ± 31; P < .001). Solid lines represent mean difference (MD), and dashed lines represent 95% limits of agreement.

Intrareader Reliability

Intrareader reliability was excellent for phantom measurements, and intraclass correlation coefficients values ranged from 0.9 (95% CI: 0.67, 0.97) for measurement of B1 deviation to 1 (95% CI: 1, 1) for the T1 (3D MP2RAGE MRI). Intrareader reliability was strong to excellent for measurements in volunteers and in participants with hip pain. Intraclass correlation values ranged from 0.86 (95% CI: 0.8, 0.91) for muscle T1 (3D MP2RAGE MRI) in volunteers to 0.92 (95% CI: 0.88, 0.95) for T1 cartilage relaxation in participants with hip pain (Table 4).

Table 4: Intraobserver Reliability for Each Measurement

Table 4:

Discussion

Albeit often used for T1 mapping of hip cartilage, the three-dimensional (3D) dual-flip-angle technique is highly sensitive to flip angle variations related to B1 inhomogeneities that increase at 3.0 T (17). Inversion recovery–based 3D magnetization-prepared 2 rapid gradient-echo MRI, previously applied for B1 insensitive T1 mapping at 3.0 T and 7.0 T in neuroimaging (19), was used for T1 mapping of hip cartilage at 3.0 T in this study.

In phantoms, 3D MP2RAGE MRI showed less T1 deviation (mean, 3 msec ± 11 vs 253 msec ± 85; P < .001) from the 2D inversion recovery reference compared with 3D DFA techniques, and this bias, unlike for 3D DFA techniques, did not correlate with increasing B1 deviation.

In healthy volunteers, no variation in cartilage T1 was observed over the field of view for 3D MP2RAGE MRI (mean difference, 8 msec ± 46; P = .45) compared with 3D DFA techniques (109 msec ± 163, P = .008), and less interindividual variation in cartilage T1 was observed between participants (3D MP2RAGE MRI: mean, 1032 msec ± 40; 3D DFA technique: mean, 1489 msec ± 178; ratio of variances, 20; P < .001).

In participants with hip pain, T1 assessed with 3D MP2RAGE MRI showed an underestimation of −23 msec ± 31 (P < .001) from 2D inversion recovery.

In phantoms, we showed that 3D MP2RAGE MRI was more accurate than 3D DFA techniques (3 msec vs 253 msec; P < .001) compared with a T1 reference at 3.0 T. The T1 bias of 3D DFA techniques (mean relative bias, 36.5%) strongly increased with increasing B1 deviation (rs = 0.69; P = .009), despite applying a preconditioning radiofrequency pulse with turbo fast low-angle shot readout for fast B1 correction (18). By contrast, a previous study reported a T1 bias of 5%–10% from 2D inversion recovery in phantoms at 3.0 T after B1 correction using a different spin-echo sequence with a section-selective excitation followed by two refocusing pulses to generate a spin echo and a stimulated echo (17).

In healthy volunteers, we observed a high interindividual mean variation of 1489 msec ± 178 for femoroacetabular cartilage T1 using 3D DFA techniques, similar to Hesper et al (13), who reported lower mean values but a comparable standard deviation for femoral cartilage T1 (860 msec ± 182) in volunteers, using 3D DFA techniques. In addition, we observed regional differences in cartilage T1 for 3D DFA techniques despite performing unenhanced imaging in healthy volunteers (109 msec ± 163; P = .008). To limit these variations, B1 insensitive alternatives such as a saturation-recovery technique for T1- (29) or an MRI fingerprinting technique for rapid 2D radial T1- and/or T2 mapping (30) have been developed. However, given the potential of automatic hip cartilage segmentation, compositional 3D cartilage MRI sequences (31,32), such as 3D MP2RAGE MRI, or improved methods, such as MPnRAGE MRI, are desirable (33). Indeed, when using 3D MP2RAGE MRI in volunteers, we found less interindividual variation in cartilage T1 compared with the 3D DFA techniques (ratio of variances, 20) and did not observe a significant regional variation in cartilage T1 (mean difference, 8 msec ± 46; P = .45).

In participants with hip pain, the systematic underestimation of the 3D MP2RAGE MRI was −23 msec (P < .001; relative bias, −4%) compared with the 2D inversion recovery T1 reference. This is comparable to the 5% variation of 3D MP2RAGE MRI reported for gray matter in the brain (19). This bias is relatively small and unlikely to be clinically important as it is less than the reported differences in T1 of 50–100 msec in different stages of intraoperative cartilage damage (14).

The first limitation of this pilot study was the absence of intraoperative correlation. Second, because of scanning time restrictions, we did not include a 2D inversion recovery sequence in the volunteer study. Third, k-space data were acquired in a linear reordering (ie, all ky lines for one kz per repetition time). However, contrast efficiency may be increased with a different sampling scheme that distributes the center of k-space more efficiently over the k-space lines that are acquired per repetition time (34), which could be the subject of future studies.

To conclude, compared with three-dimensional (3D) dual-flip-angle techniques, 3D magnetization-prepared 2 rapid gradient-echo MRI enabled more accurate T1 mapping of hip cartilage, was less affected by B1 inhomogeneities, and showed high accuracy against a T1 reference in participants with hip pain.

Disclosures of Conflicts of Interest: F.S. disclosed no relevant relationships. O.A. disclosed no relevant relationships. T.D.L. disclosed no relevant relationships. Y.J.K. disclosed no relevant relationships. K.A.S. disclosed no relevant relationships. M.I. disclosed no relevant relationships. J.L.C. Activities related to the present article: disclosed no relevant relationships. Activities not related to the present article: receives payment for development of educational presentations. Other relationships: disclosed no relevant relationships. T.K. Activities related to the present article: disclosed no relevant relationships. Activities not related to the present article: is employed with Siemens Healthcare AG Switzerland. Other relationships: disclosed no relevant relationships. M.K. Activities related to the present article: disclosed no relevant relationships. Activities not related to the present article: is employed with Siemens Healthcare AG Switzerland. Other relationships: disclosed no relevant relationships. M.T. disclosed no relevant relationships. S.D.B. disclosed no relevant relationships. E.N.N. disclosed no relevant relationships. B.J. disclosed no relevant relationships.

Author Contributions

Author contributions: Guarantors of integrity of entire study, F.S., K.A.S., B.J.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; approval of final version of submitted manuscript, all authors; agrees to ensure any questions related to the work are appropriately resolved, all authors; literature research, F.S., O.A., T.D.L., K.A.S., M.K., M.T.; clinical studies, K.A.S., M.T., S.D.B., E.N.N.; experimental studies, F.S., O.A., Y.J.K., K.A.S., M.I., T.K., M.K., S.D.B., B.J.; statistical analysis, F.S., T.D.L., K.A.S., M.I.; and manuscript editing, F.S., O.A., T.D.L., Y.J.K., K.A.S., M.I., T.K., M.K., M.T., E.N.N., B.J.

F.S. supported by the Swiss National Science Foundation (grant no. P1BEP3_181643).

References

  • 1. Sutter R, Pfirrmann CWA. Update on Femoroacetabular Impingement: What Is New, and How Should We Assess It? Semin Musculoskelet Radiol 2017;21(5):518–528.
  • 2. Sutter R, Zubler V, Hoffmann A, et al. Hip MRI: how useful is intraarticular contrast material for evaluating surgically proven lesions of the labrum and articular cartilage? AJR Am J Roentgenol 2014;202(1):160–169.
  • 3. Schmaranzer F, Cerezal L, Llopis E. Conventional and Arthrographic Magnetic Resonance Techniques for Hip Evaluation: What the Radiologist Should Know. Semin Musculoskelet Radiol 2019;23(3):227–251.
  • 4. Blankenbaker DG, Ullrick SR, Kijowski R, et al. MR arthrography of the hip: comparison of IDEAL-SPGR volume sequence to standard MR sequences in the detection and grading of cartilage lesions. Radiology 2011;261(3):863–871.
  • 5. Pfirrmann CWA, Duc SR, Zanetti M, Dora C, Hodler J. MR arthrography of acetabular cartilage delamination in femoroacetabular cam impingement. Radiology 2008;249(1):236–241.
  • 6. Sutter R, Zanetti M, Pfirrmann CWA. New developments in hip imaging. Radiology 2012;264(3):651–667.
  • 7. Schmaranzer F, Arendt L, Liechti EF, et al. Do dGEMRIC and T2 Imaging Correlate With Histologic Cartilage Degeneration in an Experimental Ovine FAI Model? Clin Orthop Relat Res 2019;477(5):990–1003.
  • 8. Kang Y, Choi JY, Yoo HJ, Hong SH, Kang HS. Delayed gadolinium-enhanced MR imaging of cartilage: A comparative analysis of different gadolinium-based contrast agents in an ex vivo porcine model. Radiology 2017;282(3):734–742.
  • 9. Zilkens C, Miese F, Herten M, et al. Validity of gradient-echo three-dimensional delayed gadolinium-enhanced magnetic resonance imaging of hip joint cartilage: a histologically controlled study. Eur J Radiol 2013;82(2):e81–e86.
  • 10. Kim SD, Jessel R, Zurakowski D, Millis MB, Kim YJ. Anterior delayed gadolinium-enhanced MRI of cartilage values predict joint failure after periacetabular osteotomy. Clin Orthop Relat Res 2012;470(12):3332–3341.
  • 11. Schmaranzer F, Haefeli PC, Hanke MS, et al. How does the dGEMRIC index change after surgical treatment for FAI? A prospective controlled study: Preliminary results. Clin Orthop Relat Res 2017;475(4):1080–1099.
  • 12. Palmer A, Fernquest S, Rombach I, et al. Diagnostic and prognostic value of delayed Gadolinium Enhanced Magnetic Resonance Imaging of Cartilage (dGEMRIC) in early osteoarthritis of the hip. Osteoarthritis Cartilage 2017;25(9):1468–1477.
  • 13. Hesper T, Bulat E, Bixby S, et al. Both 3-T dGEMRIC and acetabular-femoral T2 difference may detect cartilage damage at the chondrolabral junction. Clin Orthop Relat Res 2017;475(4):1058–1065.
  • 14. Bulat E, Bixby SD, Siversson C, Kalish LA, Warfield SK, Kim YJ. Planar dGEMRIC Maps May Aid Imaging Assessment of Cartilage Damage in Femoroacetabular Impingement. Clin Orthop Relat Res 2016;474(2):467–478.
  • 15. Schmaranzer F, Haefeli PC, Liechti EF, Hanke MS, Tannast M, Büchler L. Improved Cartilage Quality on Delayed Gadolinium-Enhanced MRI of Hip Cartilage after Subchondral Drilling of Acetabular Cartilage Flaps in Femoroacetabular Impingement Surgery at Minimum 5-Year Follow-Up. Cartilage 2020. 10.1177/1947603520941241. Published online July 19, 2020.
  • 16. Mamisch TC, Dudda M, Hughes T, Burstein D, Kim YJ. Comparison of delayed gadolinium enhanced MRI of cartilage (dGEMRIC) using inversion recovery and fast T1 mapping sequences. Magn Reson Med 2008;60(4):768–773.
  • 17. Siversson C, Chan J, Tiderius CJ, et al. Effects of B1 inhomogeneity correction for three-dimensional variable flip angle T1 measurements in hip dGEMRIC at 3 T and 1.5 T. Magn Reson Med 2012;67(6):1776–1781.
  • 18. Chung S, Kim D, Breton E, Axel L. Rapid B1+ mapping using a preconditioning RF pulse with TurboFLASH readout. Magn Reson Med 2010;64(2):439–446.
  • 19. Marques JP, Kober T, Krueger G, van der Zwaag W, Van de Moortele PF, Gruetter R. MP2RAGE, a self bias-field corrected sequence for improved segmentation and T1-mapping at high field. Neuroimage 2010;49(2):1271–1281.
  • 20. Barral JK, Gudmundson E, Stikov N, Etezadi-Amoli M, Stoica P, Nishimura DG. A robust methodology for in vivo T1 mapping. Magn Reson Med 2010;64(4):1057–1067.
  • 21. Bashir A, Gray ML, Hartke J, Burstein D. Nondestructive imaging of human cartilage glycosaminoglycan concentration by MRI. Magn Reson Med 1999;41(5):857–865.
  • 22. Kingsley PB. Methods of measuring spin-lattice (T1) relaxation times: An annotated bibliography. Concepts Magn Reson 1999;11(4):243–276.
  • 23. Klenke FM, Hoffmann DB, Cross BJ, Siebenrock KA. Validation of a standardized mapping system of the hip joint for radial MRA sequencing. Skeletal Radiol 2015;44(3):339–343.
  • 24. Hanke MS, Steppacher SD, Anwander H, Werlen S, Siebenrock KA, Tannast M. What MRI Findings Predict Failure 10 Years After Surgery for Femoroacetabular Impingement? Clin Orthop Relat Res 2017;475(4):1192–1207 [Published correction appears in Clin Orthop Relat Res 2017;475(4):1278.].
  • 25. Mascarenhas VV, Castro MO, Rego PA, et al. The Lisbon Agreement on Femoroacetabular Impingement Imaging-part 1: overview. Eur Radiol 2020;30(10):5281–5297 [Published correction appears in Eur Radiol 2020;30(12):6966–6967.].
  • 26. Schmaranzer F, Kallini JR, Miller PE, Kim YJ, Bixby SD, Novais EN. The Effect of Modality and Landmark Selection on MRI and CT Femoral Torsion Angles. Radiology 2020;296(2):381–390.
  • 27. Pitman EJG. A Note on Normal Correlation. Biometrika 1939;31(1–2):9–12.
  • 28. Morgan WA. Test for the Significance of the Difference between the Two Variances in a Sample from a Normal Bivariate Population. Biometrika 1939;31(1–2):13–19.
  • 29. Lattanzi R, Glaser C, Mikheev AV, et al. A B1-insensitive high resolution 2D T1 mapping pulse sequence for dGEMRIC of the HIP at 3 Tesla. Magn Reson Med 2011;66(2):348–355.
  • 30. Cloos MA, Assländer J, Abbas B, et al. Rapid Radial T1 and T2 Mapping of the Hip Articular Cartilage With Magnetic Resonance Fingerprinting. J Magn Reson Imaging 2019;50(3):810–815.
  • 31. Hesper T, Bittersohl B, Schleich C, et al. Automatic Cartilage Segmentation for Delayed Gadolinium-Enhanced Magnetic Resonance Imaging of Hip Joint Cartilage: A Feasibility Study. Cartilage 2020;11(1):32–37.
  • 32. Schmaranzer F, Helfenstein R, Zeng G, et al. Automatic MRI-based Three-dimensional Models of Hip Cartilage Provide Improved Morphologic and Biochemical Analysis. Clin Orthop Relat Res 2019;477(5):1036–1052.
  • 33. Kecskemeti S, Samsonov A, Hurley SA, Dean DC, Field A, Alexander AL. MPnRAGE: A technique to simultaneously acquire hundreds of differently contrasted MPRAGE images with applications to quantitative T1 mapping. Magn Reson Med 2016;75(3):1040–1053.
  • 34. Mussard E, Hilbert T, Forman C, Meuli R, Thiran JP, Kober T. Accelerated MP2RAGE imaging using Cartesian phyllotaxis readout and compressed sensing reconstruction. Magn Reson Med 2020;84(4):1881–1894.

Article History

Received: Jan 14 2020
Revision requested: Feb 17 2020
Revision received: Dec 2 2020
Accepted: Jan 5 2021
Published online: Feb 23 2021
Published in print: Apr 2021