May 2012, VOLUME 198

Recommend & Share

May 2012, Volume 198, Number 5

Musculoskeletal Imaging

Original Research

Fat-Corrected T2 Measurement as a Marker of Active Muscle Disease in Inflammatory Myopathy

+ Affiliation:
1 Both authors: Radiology and Imaging Sciences, NIH Clinical Center, 10 Center Dr, Bethesda, MD 20892.

Citation: American Journal of Roentgenology. 2012;198: W475-W481. 10.2214/AJR.11.7113

Next section

OBJECTIVE. We sought to improve the utility of T2 measurement as a marker of active muscle disease in patients with idiopathic inflammatory myopathy by correcting for T2 prolongations caused by fatty replacement of muscle that accompnaies chronic muscle damage.

SUBJECTS AND METHODS. Twenty-one patients with idiopathic inflammatory myopathy underwent a standardized MRI evaluation of the thighs. Fat fraction maps were calculated from dual-echo gradient-echo images. Fat-corrected T2 maps were generated from multiecho spin-echo images on the basis of a biexponential model that incorporated voxelwise fat fraction estimates. Semiautomated summaries of conventional and fat-corrected muscle T2 values were compared with one another and with standardized visual scores of muscle disease based on T1-weighted spin-echo and STIR images.

RESULTS. Fat-corrected muscle T2 maps showed lower mean values and greater histogram entropy than conventional T2 maps, as analyzed over a standardized portion of the thigh muscles. Conventional and fat-corrected T2 values correlated with visual scores of active muscle disease on STIR images and with the varying intensity of disease depicted with STIR in focal muscle regions.

CONCLUSION. MRI T2 maps of muscle can be corrected for varying fat content by combining the information from chemical shift–sensitive gradient-echo and multiecho spin-echo images. Use of this strategy may prove useful in the study of idiopathic inflammatory myopathy and other diseases characterized by both muscle inflammation and atrophy.

Keywords: MRI, muscle, myopathy, myositis, T2

Idiopathic inflammatory myopathies are systemic autoimmune disorders of uncertain pathogenesis characterized by chronic muscle weakness and inflammation. The most common idiopathic inflammatory myopathies are polymyositis, dermatomyositis, and sporadic inclusion body myositis [1]. The patchy muscle inflammation that characterizes active disease in idiopathic inflammatory myopathy is well depicted with MRI, and for this purpose, STIR MRI is particularly useful. Assessment of disease on STIR images is difficult to quantify, however, because of the diffuse extent and variable signal intensity of disease-related changes in muscle. Disease severity and extent can be estimated with semiquantitative scoring systems, but the process is subjective and can be influenced by the MRI parameters used.

Measurement of muscle T2 relaxation has been advocated to more objectively gauge active muscle disease in idiopathic inflammatory myopathy. A previous study [2] showed elevated T2 values in areas of active muscle disease in patients with juvenile dermatomyositis. With time, however, atrophy and fatty involution of muscle can occur as a manifestation of muscle damage in patients with idiopathic inflammatory myopathy. Published values for the T2 of fat vary widely but are substantially greater than the T2 of normal or even actively diseased muscle [35]. Fatty replacement of muscle in advanced idiopathic inflammatory myopathy confounds the interpretation of muscle T2 measurements.

T2 measurements are commonly derived from a multiecho spin-echo acquisition, but quantitative fat fraction is estimated from a separate multiecho chemical shift–sensitive acquisition. Innovations in pulse sequence design have enabled simultaneous separation of fat and water signal intensity and separate measurements of T2 relaxation for water and fat protons [6, 7]. These techniques, however, are not widely available on commercial MRI platforms. In this study, we tested a postprocessing strategy that integrates the information from a multiecho spin-echo acquisition and a separate chemical shift–sensitive acquisition to generate fat-corrected T2 maps. The intention behind this strategy is to establish a practical and robust MRI indicator of active muscle disease in patients with idiopathic inflammatory myopathy.

Subjects and Methods
Previous sectionNext section

The study protocol was approved by the internal review board at our institution and complied with HIPAA. All subjects gave written consent to participate in the protocol. The study included 21 patients with idiopathic inflammatory myopathy (six male, 15 female; average age, 33.5 [SD, 18.5] years; range, 6–61 years; polymyositis, 11; dermatomyositis, 10). Ten subjects without idiopathic inflammatory myopathy (seven men, three women; average age, 46.6 [SD, 13.8] years; range, 31–78 years) constituted a reference group. These subjects had normal findings at an MRI evaluation of the thigh muscles, as interpreted by an experienced musculoskeletal radiologist.

MRI Protocol

Patients and subjects underwent a standardized 1.5-T MRI survey of both thighs (Achieva system, Philips Healthcare) with a quadrature body coil. This survey included the following axial 2D acquisitions: T1-weighted spin-echo (TR/TE, 525/11; bandwidth, 215 Hz/pixel; number of signals averaged, 2; acquisition time, 4 minutes 6 seconds), fast spin-echo STIR (TR, 5700; inversion time, 150 ms, effective TE, 35 ms; echo train length, 9; bandwidth, 242 Hz/pixel; number of signals averaged, 2; acquisition time, 4 minutes 56 seconds); double-echo spoiled gradient-echo at consecutive in-phase and out-of-phase TEs (TR/TE, 250/2.3 and 4.6; flip angle, 90°; bandwidth, 360 Hz/pixel; number of signals averaged, 3; acquisition time, 2 minutes 55 seconds), multiecho spin-echo (TR/TE, 2000/15, 30, 45, 60, 75; bandwidth, 144 Hz/pixel; number of signals averaged, 1; acquisition time, 6 minutes 47 seconds). Slices were obtained at the same locations for all images (slice thickness, 10 mm; interslice gap, 10 mm; FOV, 36–44 cm2; pixel size, 0.9–1.5 mm2).

Fat-Corrected Muscle T2 Measurement

Fat deposition in atrophic muscle is primarily extramyocellular, and the fat preferentially accumulates in interfascicular regions [8]. Volume averaging of fat and muscle protons occurs to varying degrees, depending on image resolution and the pattern of fatty replacement. Conventional T2 maps are generated in multiecho spin-echo MR acquisitions by voxelwise fitting of the observed signal intensity at each echo with the following equation:

where T1m is T1 relaxation of muscle; T2m is T2 relaxation of muscle; Ten is the TE of the nth echo in a multiecho spin-echo sequence; k is a scaling factor correcting for coil sensitivity, receiver gain, and other factors dependent on the MRI system; and Mm is proton density–weighted magnetization of muscle. Mm is also influenced by sequence-dependent magnetization transfer effects. For each voxel, k × Mm × E1m can be fitted as a single constant. For voxels that contain both fat and muscle protons, the signal magnitude for the nth echo in a multiecho spin-echo sequence can be modeled as follows:
where T1f is T1 relaxation of fat, Mf is fat proton density, T2f is T2 relaxation of fat, and FF is the volumetric fat fraction.

When TR is constant,

where R = (Mf × E1f) / (Mm × E1m) and k′ = k × Mm × E1m.

R incorporates differential T1 relaxation of muscle and fat and varies with TR. The signal-to-noise ratio and number of echoes required to effectively fit all the free parameters to this biexponential model would necessitate long, clinically impractical acquisition times. Assuming that the relative T1 and proton density of muscle and fat (R) and the T2 of fat are constant simplifies the model considerably. Then voxelwise fitting of only the one additional parameter k′ yields an estimate of T2m if FF is known. In this study FF was estimated from a separate, dual-echo gradient-echo image (see later, Estimation of Fat Fraction).

Estimation of R

R was estimated for a control group of 10 subjects with MRI with the same multiecho spin-echo sequence used for the patients. For this analysis, neighboring pure fat and lean muscle regions along the deep fascia were manually defined on the multiecho spin-echo images. These regions measured 30–50 mm2. Kf, Km, T2f, and T2m were obtained by fitting the mean signal intensity of these regions to the following equation:


For spatially neighboring regions, the k values should be nearly equal and can be assumed to cancel each other out. Hence, R is estimated as Kf / Km. In the reference group, this region of interest analysis yielded a mean R of 2.34 (standard error of the mean [SEM], 0.024) and mean T2 values for fat and lean muscle of 112.2 ms (SEM, 0.35) and 43.3 ms (SEM, 0.48).

Estimation of Fat Fraction

The fat-corrected T2 estimate (equation 3) incorporates a measurement of fat fraction. In this study, fat fraction was estimated from a separate double-echo gradient-recalled echo image with TE chosen to occur when fat and water protons were consecutively out of phase or in phase. This image was acquired at the same slice location as the multiecho spin-echo image. By Dixon formulation, if the differential relaxation of fat and water protons is ignored, for water dominant voxels, the following applies:

When differential T2* relaxation effects are ignored, the signal intensity of a gradient-recalled echo sequence at in- and out-of-phase TEs can be approximated in terms of fat (Sf) and water (Sw) signal intensity as follows:
The factor kr normalizes fat signal intensity to a volume-equivalent muscle signal intensity, reflecting both differential T1 relaxation and the proton density of fat as opposed to muscle protons. For voxels in which water proton signal predominates, a corrected estimate of fat fraction (FFc) is as follows:
In severely damaged muscle, fat signal can predominate, and the equation is as follows:

Fat- and water-dominant voxels were differentiated in this study by use of a region-growing phase-correction method [9]. In lieu of independent measurements of T1 for muscle and fat, kr can be conveniently estimated from the signal intensity of adjacent pure fat and lean muscle regions as follows:

Analysis of fat and lean muscle regions in the 10 reference subjects yielded an estimate of kr of 1.93 (SEM, 0.007).

Image Analysis

Image postprocessing, including T2 and fat fraction calculation, was implemented in ImageJ (Rasband W, National Institutes of Health), an open-source imaging resource based on Java (Oracle America). T2 and fat-corrected T2 were estimated by nonlinear fitting of multiecho signal data to equations 1 and 3 with a Levenberg-Marquardt algorithm. T2 or fat-corrected T2 values were summarized for each subject for a standardized portion of the both thighs, consisting of 10 contiguous slices. Thigh muscle regions were defined with an automated, adaptive, moving-window, intensity-based segmentation algorithm [10]. This robust automated threshold selection method was applied to T1 spin-echo images acquired through the same standardized portion of the thighs in each subject to yield muscle masks.

Regions of interest were manually defined by an experienced musculoskeletal radiologist for direct comparison of observed and calculated muscle features. On STIR images of each subject, muscle regions were defined that exhibited the following signal characteristics: 0, normal; 1, possibly abnormal; 2, mildly abnormal; 3, moderately hyperintense; and 4, severely hyperintense. Muscle regions were also defined on T1 spin-echo images of each subject that were deemed 0, normal; 1, possibly or mildly atrophic; or 2, definitely atrophic.

TABLE 1: Visual Image Scoring

An experienced musculoskeletal radiologist visually scored STIR and T1 spin-echo images of both thighs using a standardized scoring system (Table 1). Visual scores were assigned to medial, anterior, and posterior compartments. The sum of scores for the three compartments constituted a global visual score for each patient. Active disease was scored on axial STIR images, and disease damage was scored on axial T1-weighted spin-echo images. The score for active disease considers both the 3D spatial extent of disease and the magnitude of changes in signal intensity. For the purposes of this classification, regions exhibiting moderate or severe hyperintensity on STIR images were considered to have marked hyperintensity.

Data Analysis

Muscle masks were used to summarize quantitative muscle features for each patient, including global mean T2 and fat-corrected T2 values and volume histogram entropy [11]. T2 histograms were generated with bin widths of 5 ms. Histogram entropy was calculated as follows:

where pi = ni / N; ni is the number of voxels in bin i; and n is the total number of analyzed voxels. Intrasubject differences in mean muscle T2 and fat-corrected T2 values and in histogram entropy were tested with a paired Student t test. Correlation of these values with visual score categories was tested with Kendall tau.

Previous sectionNext section
Sample Case

Images of the thighs in a patient with polymyositis are shown in Figures 1A, 1B, 1C, 1D, 1E, 1F, and 1G. The T1-weighted spin-echo image (Fig. 1A) shows mild to moderate muscle atrophy manifested by fatty involution in multiple muscles. The STIR image (Fig. 1B) shows extensive inflammation in atrophic and nonatrophic muscles manifested by ill-defined regions of increased muscle signal intensity. Application of a robust automated threshold selection segmentation algorithm [7] to the T1-weighted spin-echo image yields a binary mask, which automates summarization of quantitative muscle features (Fig. 1C) and is applied to the calculated images (Figs. 1D, 1E, and 1F). A masked fat fraction map (Fig. 1D) is derived from a dual-echo gradient-echo acquisition. Areas of muscle atrophy are more conspicuous on the fat fraction maps. The conventional T2 map (Fig. 1E) shows generally higher T2 values than does the fat-corrected T2 map (Fig. 1F). Corresponding histograms are shown in Figure 1G. The conventional T2 map also exhibits diminished contrast or dynamic range and suggests the presence of more extensive disease than does the fat-corrected T2 map in mildly atrophic muscle regions, such as the vastus lateralis.

Visual Scores

The mean visual score for active muscle disease in the 21 patients was 8.6 (SD, 4.4; range, 0–15) based on STIR images. The mean visual score for muscle damage in the 21 patients was 4.5 (SD, 4.6; range, 0–13) based on T1-weighted spin-echo images. Eight of the patients had no muscle damage according to visual assessment, and 10 had damage scores of 2 or less.

View larger version (133K)
Fig. 1A37-year-old man with polymyositis.

A, Axial T1-weighted spin-echo MR image shows ill-defined areas of muscle atrophy, most notably in adductor magnus and vastus lateralis muscles, and relative preservation of muscle bulk.

View larger version (149K)
Fig. 1B37-year-old man with polymyositis.

B, Axial STIR MR image depicts extensive edemalike signal intensity predominating in anterior compartment, consistent with active muscle disease. Mild diffuse inflammatory changes are evident in subcutis.

View larger version (25K)
Fig. 1C37-year-old man with polymyositis.

C, Muscle mask derived from T1-weighted spin-echo image with robust automatic segmentation tool, facilitating summarization of muscle T2 values.

View larger version (94K)
Fig. 1D37-year-old man with polymyositis.

D, Relaxation-corrected fat fraction map calculated from double-echo gradient-echo images (not shown).

View larger version (83K)
Fig. 1E37-year-old man with polymyositis.

E, Conventional T2 map estimated from multiecho spin-echo acquisition.

View larger version (86K)
Fig. 1F37-year-old man with polymyositis.

F, Fat-corrected muscle T2 map integrates fat fraction data and shows lower T2 values in areas of greater fat content and broader range of T2 values. Color table is same as in E.

Standardized Thigh Regions

The mean conventional and fat-corrected T2 values for the standardized thigh regions and the associated volume histogram entropy are summarized in Table 2. The mean values for conventional T2 maps were significantly greater than the mean values for fat-corrected T2 maps. The histogram entropy for fat-corrected T2 maps was significantly greater than that for conventional T2 maps. Differences in mean values and entropy between conventional and fat-corrected T2 maps held for the subgroup of eight patients with visual muscle damage scores of 0 and within the 13 patients with nonzero muscle damage scores (Table 2).

View larger version (14K)
Fig. 1G37-year-old man with polymyositis.

G, Histogram shows fat-corrected (fcT2) and conventional T2 muscle values for standardized thigh muscle regions in this patient. Distribution for fat-corrected T2 values is shifted to lower values and slightly broader. Vertical axis is in unitless proportions; horizontal axis, 5-ms intervals).

Mean conventional and fat-corrected muscle T2 values correlated significantly (p < 0.001) and to a similar degree with visual scores for active myositis (conventional T2, τ = 0.55; SD, 0.057; fat-corrected T2, τ = 0.53; SD, 0.060). Intrasubject differences between mean conventional T2 and fat-corrected T2 values (mean difference, 7.9 [SD, 3.2] ms) did not correlate significantly with global disease activity score (τ = 0.01; p = 0.98) but correlated significantly with global disease damage score (τ = 0.33; p = 0.015). Higher damage score was associated with a larger difference between conventional T2 and fat-corrected T2.

Region of Interest Analysis

Results of the region of interest analysis are summarized in Table 3, which shows significantly higher values for conventional T2 than for fat-corrected muscle T2 for all categories of STIR abnormality. The mean fat fraction for regions exhibiting no atrophy, possible or minimal atrophy, and definite atrophy were 3.76% (SD, 3.68%; n = 20), 9.95% (SD, 4.94; n = 15), and 19.02% (SD, 3.79%; n = 13). Not every visual category was represented in every patient, accounting for the varying number of regions for each category.

Conventional and fat-corrected T2 values appeared to be equivalent for differentiating muscle regions assessed as normal and equivocal for disease on STIR images. The mean intrasubject T2 difference between muscle regions assessed on STIR images as equivocal for disease and as normal was 5.8 (SD, 13.1) ms. The mean intrasubject fat-corrected T2 difference between muscle regions assessed on STIR images as equivocal for disease and as normal was 6.1 (SD, 11.7) ms.

Conventional and fat-corrected T2 values also appeared equivalent in differentiating muscle regions assessed as mild disease and equivocal for disease on STIR images. The mean intrasubject T2 difference between muscle regions assessed on STIR images as mild disease and as equivocal for disease was 4.7 (SD, 15.7) ms. The mean intrasubject fat-corrected T2 difference between muscle regions assessed on STIR images as mild disease and as equivocal for disease was 5.1 (SD, 16.6) ms.

Previous sectionNext section

STIR MRI is useful for assessing active muscle disease in inflammatory myopathy, in part because of robust fat suppression over large regions. For longitudinal assessments of disease activity, and in clinical trials, visual assessment of STIR images for muscle disease can lack precision and be influenced by variation of acquisition parameters.

TABLE 2: T2 and Fat-Corrected T2 for Standardized Thigh Regions and Corresponding Histogram Entropy

Muscle T2 measurements may serve as a more objective marker of active muscle disease in patients with idiopathic inflammatory myopathy [2]. Because the T2 of fat greatly exceeds that of healthy muscle, fatty replacement of muscle can substantially alter bulk muscle T2. For this reason, T2 measurement may be a useful marker of disease in neuromuscular conditions such as Duchenne muscular dystrophy [12], in which T2 prolongation primarily reflects increases in muscle fat. Muscle damage and concomitant fatty replacement of muscle in idiopathic inflammatory myopathy, however, may simulate worsening active disease, even when inflammation is absent or improving.

We present a strategy for correcting muscle T2 measurements for potential fatty replacement of muscle. This approach integrates a separate estimate of muscle fat fraction, accomplished in this study by a traditional two-point Dixon method. The quantitative fat fraction map may also aid in assessment of muscle damage in idiopathic inflammatory myopathy. Fat-corrected T2 values for muscle have been found to be consistently lower than conventional T2 values while the histograms of fat-corrected T2 maps exhibit consistently higher entropy, and thus potential information content, than conventional T2 maps, even in the absence of definite muscle damage based on visual score. Fat-corrected and conventional muscle T2 values are similarly correlated with muscle disease intensity as perceived on STIR images in this preliminary study.

TABLE 3: T2 and Fat-Corrected T2 Versus STIR Signal Intensity

The appearance of active muscle disease can vary on STIR images, depending on the chosen imaging parameters. Signal intensity in damaged muscle can be equivocally elevated by the presence of fat if the inversion time is somewhat shorter than the null point for fat—a common practice for increasing the conspicuity of normal anatomy. For this reason, a fat-corrected T2 map may better resolve disease in muscle that exhibits equivocal signal intensity elevation on STIR images compared with conventional T2 maps. Our analysis of equivocal changes in muscle signal intensity on STIR images of this group of patients, however, did not confirm this potential advantage.

Our approach to fat-corrected T2 estimation is feasible in routine clinical practice with use of common, commercially available pulse sequences. The strategy is also compatible with newer, more sophisticated MRI methods of fat fraction determination, such as iterative decomposition techniques with unconstrained TE [13]. Fat fraction measurements can be adjusted for T2* relaxation by collection of three or more echoes, although in this strategy identical T2* values for fat and water protons are typically assumed [14]. Fat fraction measurements incorporating independent T2* corrections for the lipid and aqueous proton species have also been described [15, 16]. Fat fraction measurements and associated T2* corrections can be further refined by accounting for minor spectral peaks of fat, in addition to the major methylene—(CH2)n—proton peak approximately 3.5 ppm downfield from water. In these refinements, predefined multispectral models are used for fat [15], or the sizes of separate fat peaks are individually calibrated [17].

Potentially larger errors in fat fraction measurement can result from differential T1 relaxation or volumetric density of lipid and aqueous protons. Use of a small flip angle minimizes this error [18] but compromises signal-to-noise ratio, and the resulting images may be less useful for visual assessment. Alternatively, dual flip angle acquisitions can correct for T1-related bias in fat fraction measurements [18, 19], albeit at the cost of additional acquisition time. We introduce a simple, empirical method for adjusting for T1 and proton density bias in traditional two-point Dixon measurements, assuming consistent and constant signal characteristics of fat and muscle compartments, as inferred from reference regions. This simplified approach is susceptible to error, however, in the presence of marked inhomogeneity in the transmit field (B1) or receiver coil sensitivity.

In our proposed method of fat-corrected muscle T2 measurement, constant relative T1 and proton density weighting for fat and muscle is assumed and presents a potential source of error. Muscle T1 relaxation is likely prolonged to some degree in myositis. However, owing to substantial differences in T1 relaxation of normal fat and muscle, any myositis-related prolongation of muscle T1 would incur a relatively less substantial change in the ratio of T1 weighting of muscle versus fat (R, equation 3). A longer TR would minimize this error and errors caused by any myositis-related alterations of muscle proton density, albeit at the cost of longer acquisition times.

Our proposed method of fat-corrected muscle T2 measurement could be further automated by adaptive derivation of the necessary constant parameters. For example, if muscle and fat regions are adequately defined by a segmentation procedure, R could be estimated for each patient or each image from an analysis of lean muscle regions (identified by a suitably low fat fraction) and pure fat regions. This approach would discount spatial variation in k (equation 1), although these variations may average out over large reference regions. The T2 of fat could also be estimated from adipose regions identified in each patient or each image. The unsupervised selection of reference fat regions may be complicated, however, by poorly localized subcutaneous disease that often occurs in dermatomyositis and, less often, polymyositis [20].

A biexponential fitting strategy similar to our approach was described for a study in which the subjects were patients with fascioscapulohumeral muscular dystrophy [21]. The purpose in that study was to improve measurement of muscle fat fraction, rather than muscle T2. In that method, constant T2 values for both muscle and fat were assumed, and the estimated fat fraction was not adjusted for potential bias related to differential T1 relaxation. Measurement of muscle T2 independent of fat proton contributions can be accomplished in a single acquisition by combining a robust iterative fat-water decomposition technique and a Carr-Purcell-Meiboom-Gill sequence [6]. A related hybrid technique with the gradient and spin-echo methods and echo sharing has also been developed. Use of that technique facilitates rapid estimation of fat-corrected T2 measurement in a breath-hold image acquisition [7]. These developments should extend any potential utility of fat-corrected T2 measurement.

Caution should be taken when adopting any T2 mapping strategy as an outcome measure because of the potential influence of magnetization transfer, diffusion effects, and stimulated echoes on measured T2 [22, 23]. Fat signal intensity is also influenced by J-coupling to varying degrees depending on multiecho pulse sequence design [24], which may partially explain the considerable variance in published T2 values for adipose tissue. Hence, standardization of acquisition technique is important for application of these methods in clinical studies.

In summary, muscle T2 measurement may serve as a useful marker of active muscle disease in patients with idiopathic inflammatory myopathy and facilitate a semiautomated approach to disease assessment. Our findings show how muscle T2 values can be corrected for potential muscle damage by integration of information from separate multiecho spin-echo and multiecho chemical shift–sensitive images into a biexponential analysis. In this preliminary study, fat-corrected T2 measures appeared to convey information similar to that on conventional T2 maps with reference to STIR imaging. Because muscle damage and fatty replacement in idiopathic inflammatory myopathy occur over extended periods, a longitudinal investigation that includes patients with idiopathic inflammatory myopathy may clarify whether a fat-corrected muscle T2 measurement strategy improves the utility of clinical T2 measurement as a marker of active disease.

We thank Lisa G. Rider and Frederick W. Miller of the National Institute of Environmental Health Sciences, who provided the patient referrals and the conceptual motivation for this study.

Supported in part by the Intramural Research Program at the Clinical Center, National Institutes of Health.


This is a Web exclusive article.

Previous sectionNext section
1. Rider LG, Miller FW. Deciphering the clinical presentations, pathogenesis, and treatment of the idiopathic inflammatory myopathies. JAMA 2011; 305:183–190 [Google Scholar]
2. Maillard SM, Jones R, Owens C, et al. Quantitative assessment of MRI T2 relaxation time of thigh muscles in juvenile dermatomyositis. Rheumatology 2004; 43:603–608 [Google Scholar]
3. Gold GE, Han E, Stainsby J, Wright G, Brittain J, Beaulieu C. Musculoskeletal MRI at 3.0 T: relaxation times and image contrast. AJR 2004; 183:343–351 [Abstract] [Google Scholar]
4. Poon CS, Szumowski J, Plewes DB, Ashby P, Henkelman RM. Fat/water quantitation and differential relaxation time measurement using chemical shift imaging technique. Magn Reson Imaging 1989; 7:369–382 [Google Scholar]
5. de Bazelaire CM, Duhamel GD, Rofsky NM, Alsop DC. MR imaging relaxation times of abdominal and pelvic tissues measured in vivo at 3.0 T: preliminary results. Radiology 2004; 230:652–659 [Google Scholar]
6. Janiczek RL, Gambarota G, Sinclair CD, et al. Simultaneous T2 and lipid quantitation using IDEAL-CPMG. Magn Reson Med 2011; 66:1293–1302 [Google Scholar]
7. Li Z, Graff C, Gmitro AF, et al. Rapid water and lipid imaging with T2 mapping using a radial IDEAL-GRASE technique. Magn Reson Med 2009; 61:1415–1424 [Google Scholar]
8. Trudel G, Ryan SE, Rakhra K, Uhthoff HK. Extra- and intramuscular fat accumulation early after rabbit supraspinatus tendon division: depiction with CT. Radiology 2010; 255:434–441 [Google Scholar]
9. Ma J. Breath-hold water and fat imaging using a dual-echo two-point Dixon technique with an efficient and robust phase-correction algorithm. Magn Reson Med 2004; 52:415–419 [Google Scholar]
10. Wilkinson MH. Gaussian-weighted moving-window robust automatic threshold selection. In: , Petkov N, Westenberg MA, eds. Computer analysis of images and patterns: lecture notes in computer science. Berlin, Germany: Springer, 2003: 369–376 [Google Scholar]
11. Shannon CE. A mathematical theory of communication. Bell Syst Tech J 1948; 27:379–423, 623–656 [Google Scholar]
12. Kim HK, Laor T, Horn PS, Racadio JM, Wong B, Dardzinski BJ. T2 mapping in Duchenne muscular dystrophy: distribution of disease activity and correlation with clinical assessments. Radiology 2010; 255:899–908 [Google Scholar]
13. Reeder SB, Pineda AR, Wen Z, et al. Iterative decomposition of water and fat with echo asymmetry and least-squares estimation (IDEAL): application with fast spin echo imaging. Magn Reson Med 2005; 54:636–644 [Google Scholar]
14. Yu H, McKenzie CA, Shimakawa A, et al. Multiecho reconstruction for simultaneous water-fat decomposition and T2* estimation. J Magn Reson Imaging 2007; 26:1153–1161 [Google Scholar]
15. Bydder M, Yokoo T, Hamilton G, et al. Relaxation effects in the quantification of fat using gradient echo imaging. Magn Reson Imaging 2008; 26:347–359 [Google Scholar]
16. Chebrolu VV, Hines CD, Yu H, et al. Independent estimation of T2* for water and fat for improved accuracy of fat quantification. Magn Reson Med 2010; 63:849–857 [Google Scholar]
17. Yu H, Shimakawa A, McKenzie CA, Brodsky E, Brittain JH, Reeder SB. Multiecho water-fat separation and simultaneous R2* estimation with multifrequency fat spectrum modeling. Magn Reson Med 2008; 60:1122–1134 [Google Scholar]
18. Liu CY, McKenzie CA, Yu H, Brittain JH, Reeder SB. Fat quantification with IDEAL gradient echo imaging: correction of bias from T(1) and noise. Magn Reson Med 2007; 58:354–364 [Google Scholar]
19. Karampinos DC, Yu H, Shimakawa A, Link TM, Majumdar S. T1-corrected fat quantification using chemical shift-based water/fat separation: application to skeletal muscle. Magn Reson Med 2011; 66:1312–1326 [Google Scholar]
20. Kimball AB, Summers RM, Turner M, et al. Magnetic resonance imaging detection of occult skin and subcutaneous abnormalities in juvenile dermatomyositis: implications for diagnosis and therapy. Arthritis Rheum 2000; 43:1866–1873 [Google Scholar]
21. Kan HE, Scheenen TW, Wohlgemuth M, et al. Quantitative MR imaging of individual muscle involvement in fascioscapulohumeral muscular dystrophy. Neuromuscul Disord 2009; 19:357–362 [Google Scholar]
22. Mendlik T, Faber SC, Weber J, et al. T2 quantitation of human articular cartilage in a clinical setting at 1.5 T: implementation and testing of four multiecho pulse sequence designs for validity. Invest Radiol 2004; 39:288–299 [Google Scholar]
23. Watanabe A, Boesch C, Obata T, Anderson SE. Effect of multislice acquisition on T1 and T2 measurements of articular cartilage at 3T. J Magn Reson Imaging 2007; 26:109–117 [Google Scholar]
24. Stables LA, Kennan RP, Anderson AW, Gore JC. Density matrix simulations of the effects of J coupling in spin echo and fast spin echo imaging. J Magn Reson 1999; 140:305–314 [Google Scholar]

Address correspondence to L. Yao ().

Recommended Articles

Fat-Corrected T2 Measurement as a Marker of Active Muscle Disease in Inflammatory Myopathy

Full Access, ,
American Journal of Roentgenology. 2012;198:1152-1160. 10.2214/AJR.11.7387
Abstract | Full Text | PDF (1119 KB) | PDF Plus (1057 KB) 
Full Access, , , , ,
American Journal of Roentgenology. 2004;183:343-351. 10.2214/ajr.183.2.1830343
Abstract | Full Text | PDF (2732 KB) | PDF Plus (817 KB) 
Full Access, , , , , ,
American Journal of Roentgenology. 2009;192:1708-1716. 10.2214/AJR.08.1764
Abstract | Full Text | PDF (926 KB) | PDF Plus (644 KB) 
Full Access, , , , , ,
American Journal of Roentgenology. 2012;198:1161-1166. 10.2214/AJR.11.7018
Abstract | Full Text | PDF (855 KB) | PDF Plus (781 KB) 
Full Access, , , , ,
American Journal of Roentgenology. 2012;198:1144-1151. 10.2214/AJR.11.7442
Abstract | Full Text | PDF (686 KB) | PDF Plus (766 KB) 
Full Access, , , ,
American Journal of Roentgenology. 2019;213:524-533. 10.2214/AJR.19.21143
Abstract | Full Text | PDF (1201 KB) | PDF Plus (977 KB)