SCIENTIFIC ARTICLE

Muscle fat-fraction and mapping in Duchenne musculardystrophy: evaluation of disease distribution and correlationwith clinical assessmentsPreliminary experience

Michele Gaeta & Sonia Messina & Achille Mileto & Gian Luca Vita & Giorgio Ascenti &Sergio Vinci & Antonio Bottari & Giuseppe Vita & Nicola Settineri & Daniele Bruschetta &

Sergio Racchiusa & Fabio Minutoli

Received: 15 June 2011 /Revised: 5 August 2011 /Accepted: 23 September 2011 /Published online: 10 November 2011# ISS 2011

AbstractPurpose To examine the usefulness of dual-echo dual-flipangle spoiled gradient recalled (SPGR) magnetic resonanceimaging (MRI) technique in quantifying muscle fat fraction(MFF) of pelvic and thighs muscles as a marker of diseaseseverity in boys with Duchenne muscular dystrophy(DMD), by correlating MFF calculation with clinicalassessments. We also tried to identify characteristic patternsof disease distribution.Materials and methods Twenty consecutive boys (meanage, 8.6 years±2.3 [standard deviation, SD]; age range, 5–15 years; median age, 9 years;) with DMD were evaluatedusing a dual-echo dual-flip angle SPGR MRI technique,calculating muscle fat fraction (MFF) of eight muscles inthe pelvic girdle and thigh (gluteus maximus, adductormagnus, rectus femoris, vastus lateralis, vastus medialis,biceps femoris, semitendinosus, and gracilis). Color-coded

parametric maps of MFF were also obtained. A neurologistwho was blinded to the MRI findings performed the clinicalassessments (patient age, Medical Research Council score,timed Gower score, time to run 10 m). The relationshipsbetween mean MFF and clinical assessments were investi-gated using Spearman’s rho coefficient. Positive andnegative correlations were evaluated and considered signif-icant if the P value was<0.05.Results The highest mean MFF was found in the gluteusmaximus (mean, 46.3 %±24.5 SD), whereas the lowest wasfound in the gracilis muscle (mean, 2.7 %±4.7 SD). MeanMFF of the gluteus maximus was significantly higher thanthat of the other muscles (P<0.01), except for the adductormagnus and biceps muscles. A significant positive correla-tion was found between the mean MFF of all muscles andthe patients age (20 patients; P<0.005), Medical ResearchCouncil score (19 patients; P<0.001), timed Gower score(17 patients; P<0.03), and time to run 10 m (20 patients; P<0.001). A positive correlation was also found between themean MFF of the gluteus maximus muscle and the timedGower score. Color-coded maps provided an efficient visualassessment of muscle fat content and its heterogeneousdistribution.

M. Gaeta :A. Mileto (*) :G. Ascenti : S. Vinci :A. Bottari :N. Settineri : S. Racchiusa : F. MinutoliDepartment of Radiological Sciences, Policlinico “G. Martino”,Via Consolare Valeria 1,98125 Messina, Italye-mail: achille.mileto@gmail.com

S. Messina :G. L. Vita :G. VitaDepartment of Neurosciences, Policlinico “G. Martino”,Messina, Italy

D. BruschettaDepartment of Biomorphology and Biotechnologies,Policlinico “G. Martino”,Messina, Italy

Skeletal Radiol (2012) 41:955–961DOI 10.1007/s00256-011-1301-5

Conclusion Muscle fat fraction calculation and mappingusing the dual-echo dual-flip angle SPGR MRI techniqueare useful markers of disease severity and permit patterns ofdisease distribution to be identified in patients with DMD.

Keywords Dual-echo dual-flip angle MRI technique .

Muscle fat fraction . Color-coded maps . Duchennemuscular dystrophy . Clinical assessments . Timed Gowerscore

Introduction

Duchenne muscular dystrophy (DMD) is a recessive X-linked form of muscular dystrophy that affects 1 in 3,600 to6,000 live male births, making it the most prevalent ofmuscular dystrophies. It is characterized by progressivedestruction of the skeletal muscle with subsequent replace-ment by adipose and fibrous tissue, eventually leading toloss of ambulation by the 13th year and to death, usually inearly adulthood [1].

One of the main problems facing clinicians is to evaluatethe precise amount of adipose replacement in dystrophicmuscles as the amount of intramuscular fat is an accuratemarker of disease progression [1]. However, muscle biopsyis invasive and cannot be repeated, especially in children,and it is limited to the site of sampling.

Moreover, following a rapidly increasing number ofpotentially effective therapeutical approaches in DMD,there has been an increased request of validated outcomemeasures able to reflect disease monitoring to be used inclinical trials [2–8]. However, the proposed tools measureonly gross decremental changes in muscle strength orfunction, entirely rely on patient compliance and, with theexception of the segmental muscle strength tests, do notevaluate individual muscles.

Several investigators [9–14] used nonquantitative mag-netic resonance imaging (MRI) to characterize the patternof fatty infiltration in DMD. A correlation betweennonquantitative MRI fatty infiltration score and clinicalassessments has also been demonstrated [10]. However,evaluation of nonquantitative MRI findings of the fattyinfiltration on T1-weighted images is a subjective methodwith limitations in assessing the disease severity.

Wren et al. [15] demonstrated that the muscle fat fraction(MFF) obtained with the three-point Dixon MRI techniquewas accurate in its assessment of disease severity in patientswith DMD and correlated more strongly with diseaseprogression than measurements of muscle strength.

Recently, Gaeta et al. [16] found that MFF evaluationobtained with the dual-echo dual-flip angle SPGR MRItechnique correlated well with histopathology in patientswith neuromuscular disorders. The technique used by theseauthors is simple and available on every routine MRIsystem, and with a very short acquisition time. Mostrecently, the same author used this technique in fivepatients with late-onset Charcot–Marie–Tooth disease type2 and found that it is a valuable tool for obtaining rapidquantification of MFF [17].

The purpose of this study was to examine the usefulnessof the dual-echo dual-flip angle SPGR MRI technique inquantifying muscle fat infiltration of pelvic and thighmuscles as a marker of disease severity in boys withDMD. MRI measurements of MFF were compared with

functional measurements obtained with clinical tests.Moreover, we tried to identify characteristic patterns ofdisease distribution.

Materials and methods

No industry provided support for this study. The entirestudy was approved by our university’s ethics committeebefore patient recruitment; written informed consent of allparents was obtained.

Patient population

Between March 2009 and November 2010 a cohort of 22consecutive boys (mean age, 8.6 years±2.3 [standarddeviation, SD]; age range, 5–15 years; median age, 9 years)with DMD, were recruited at the Neuromuscular DisordersCenter of our University for evaluation.

Inclusion criteria were: genetically proven DMD diag-nosis, patient still ambulant and able to walk independentlyfor at least 75 m, and no severe or moderate learningdifficulties or behavioral problems.

Exclusion criteria were: contraindications to MRI (n=0),and inability to cooperate and participate in the various tests(n=2). Therefore, our final patient cohort included 20subjects (mean age, 8.8 years±2.3 SD; age range, 5–15 years; median age, 9 years). These patients had nohistory of chronic illness other than DMD (including anyneuromuscular, metabolic, or endocrine disorder that couldalter bone or muscle metabolism). They were examinedusing the dual-echo dual-flip angle SPGR MRI technique inorder to estimate the MFF. All clinical assessments wereperformed within 1 month of the MRI examination.

MRI study protocol

All examinations were performed on a 1.5-T superconduc-tive MRI (Gyroscan Intera; Philips, Best, The Netherlands)using a four-channel phased array coil. Each patientunderwent examination of both the pelvic girdle and thethighs. In no cases was sedation necessary.

First, a transverse spin-density-weighted fast SPGR in-phase (IP) and opposed-phase (OP) dual-echo sequence(two-point Dixon technique) was performed with thefollowing parameters: repetition time, 300 ms; echo time,2.3 and 4.6 ms; flip angle, 20°; section thickness, 6 mm;field of view, 350–250; matrix, 512 × 288; number ofsignals acquired, 2. The acquisition time was 1 min and36 s. Fifteen sections were obtained by using theseparameters from the iliac crest to the femoral condyles.Intersection gap was between 0.5 and 2 cm to cover theentire volume of interest. Immediately after this first

956 Skeletal Radiol (2012) 41:955–961

sequence, a T1-weighted dual-echo sequence was per-formed by using the same parameters and the sameacquisition time as for the previous sequence, except withan 80° flip angle, without altering the level of sections.

The total amount of time required for each MRIexamination was less than 5 min.

MR image analysis and MFF calculation

Images were used to bilaterally quantify the MFF in eightmuscles: gluteus maximus, adductor magnus, rectus femo-ris, vastus lateralis, vastus medialis, biceps femoris, semite-ndinosus, and gracilis.

For region of interest (ROI) placement, two sectionlevels were chosen because they contained the largest areaof visible muscles with good differentiation of the differentmuscle compartments.

For each patient, the levels selected for evaluation wererespectively: for the gluteus maximus muscles, a sectionthrough the femoral heads; for thigh muscles, a sectionthrough the mid-thigh.

In all the patients, ROIs were drawn by a radiologist (M.G.), with 20 years’ experience in musculoskeletal radiolo-gy. On T1-weighted images, the size and shape of the ROIswere determined by using the individual muscle size on theaxial images, including the central part of muscles. Size,shape, and position of the ROIs were kept constant betweenthe image sets by applying a copy-and-paste function at theworkstation. Images with ROIs were stored on a secondaryconsole with open-source software (OsiriX, version 3.3.2;the OsiriX Foundation, Geneva, Switzerland) [18].

For each patient, MR images were retrieved from thesecondary console, and MFF calculations were performedin one session by another radiologist (F.M.) with 10 years’experience in musculoskeletal radiology, according to themethod described by Gaeta et al. [16].

After MFF calculation, we arbitrarily stratified patientsinto four classes, based on mean MFF percentages: class 0,patients with mean MFF between 0 and 20%; class 1,patients with mean MFF between 21 and 40%; class 2,patients with mean MFF between 41 and 60%; class 3,patients with mean MFF>60%.

Finally, images were converted to the “Analyze” formatby using Matlab software (version 7.0; Mathworks, Natick,MA, USA). An automated pixel-by-pixel analysis wasperformed to obtain color-coded parametric maps of MFF.

Clinical assessments

All enrolled childrenwere examined by a pediatric neurologist(S.M.), with pediatric neurology fellowship training and10 years of experience, who was blinded to the MRI findings.Patients were evaluated using the following methods: the

Medical Research Council (MRC) score for testing muscularstrength, the timed Gower score (in seconds), and time to run10 m (in seconds) [4]. The MRC was measured in the majorthigh muscles (thigh extensors and adductors and leg flexorsand extensors bilaterally), scored from 0 to 5 points andexpressed as an MRC score (MRCS), the MRC mean valueof all muscles tested. The timed Gower score was the timethe patient needed to rise from a lying position on the floorto standing. This test mainly expresses the function of thegluteus maximus muscles. Three out of 20 patients were notable to perform the test.

Statistical analysis

All continuous variables were expressed as the mean±standard deviation (SD).

The relationship between mean MFF and clinical assess-ments (MRCS, timed Gower score, time to run 10 m) wasinvestigated using Spearman’s rho coefficient. Spearman’srho coefficient was also used to evaluate the relationshipbetween mean MFF of gluteus maximus muscles and timedGower score. Positive and negative correlations were evalu-ated and considered significant if the P value was<0.05.

Ordinal logistic regression was used to examine therelationships among the four MFF classes and the ordinalMRCS.

A two-way analysis of variance (ANOVA)was used to lookat the MFF responses for various muscles. The factors weremuscle and age. Post-hoc comparisons using the Tukey HSDtest were conducted to explore if the fat fraction of a musclewas significantly different from the fat fraction of the others.

All statistical analyses were performed by using astatistical software program (SPSS, version 11.5; SPSS,Chicago, IL, USA).

Results

MFF and disease distribution

The muscle with the highest mean MFF was the gluteusmaximus (mean, 46.3 %±24.5 SD), followed by athedductor magnus muscle (mean, 40.0 %±26.9 SD). Thelowest mean MFFs were documented in the gracilis (mean,2.7 %±4.7 SD), and in the semitendinosus muscles (mean,12.9 %±20.8 SD) respectively (Fig. 1). Using the ANOVAtest a significant main effect for muscle and age (P<0.001)was found. The mean MFF of the gluteus maximus wassignificantly higher than that of the other muscles (P<0.01), except for the adductor magnus and biceps muscles.

Color-coded maps provided an efficient visual assess-ment of muscle fat content and its heterogeneous distribu-tion (Figs. 2, 3).

Skeletal Radiol (2012) 41:955–961 957

MFF and clinical assessments

For each patient, a significant positive correlation was foundbetween the mean MFF of all muscles and the patients’ age(20 patients; P<0.005), MRCS (19 patients); P<0.001);Fig. 4), timed Gower score (17 patients; P<0.03; Fig. 5), andtime to run 10 m (20 patients; P<0.001; Fig. 6). A positivecorrelation was also found between the mean MFF of thegluteus maximus muscle and the timed Gower score.

All 3 patients who were not able to perform the timedGower score had a mean MFF of gluteus maximus muscles>75% vs a range of 12–72% in the 17 patients who wereable to perform it.

There was a significant odds ratio (OR) for the four MFFclasses (class 0, n=11 patients, 55%; class 1, n=2 patients,10%; class 2, n=6 patients, 30%; class 3, n=1 patient, 5%)and the MRCS. The OR was 1.03, meaning that for each1% increase in MFF, the OR of being in an higher categoryfor MRCS increased by 3% (95% confidence interval, forOR: 0.1–11.8).

Discussion

It has been well documented that the amount of intramus-cular fat is an accurate marker of disease progression [1],and it also sensitively correlates with both patient age andclinical stage [10–19].

While there were few survivors of DMD beyond teenageyears in the 1960s and 1970s, the lifespan has beengradually extended in the past few decades, with manyDMD patients now living into their late 20s and 30s [20].Improvement in survival is due to a multidisciplinary careapproach [21], including corticosteroid therapy whichslows evolution of muscular disease, but does not cure it.

There are currently no effective treatments to halt themuscle breakdown in DMD. Genetics-based clinical trialshaving as an end point the restoration of dystrophin inmuscle fibers are being piloted [19, 22–24]. These trialsrequire selection of well-preserved muscles for the intra-muscular injection of drugs.

Current clinical measures of muscle impairment inpatients with DMD show several limitations, as they aresubjective methods and often not sensitive enough to detect

Fig. 2 Color-coded axial MFF maps show a low, b intermediate, andc high degrees of gluteus maximus involvement, in 6-, 10-, and 15-year-old boys respectively

Fig. 1 Bar chart shows mean muscle fat fraction (MFF ± SD) of eightmuscles in 20 patients. The gluteus maximus muscle has the highestmean MFF and the gracilis muscle has the lowest

958 Skeletal Radiol (2012) 41:955–961

small changes in disease status for longitudinal studies ortreatment assessment [2, 15, 25].

Muscle biopsy plays a fundamental role in the evaluation ofthe patient with neuromuscular disease. It has long beenrecognized as the reference standard to establish the diagnosisof DMD, and as a useful tool inmonitoring response to therapyand disease progression as it permits the degree of muscle fatreplacement to be quantified [1]. However, informationobtained from muscle biopsy is limited to the site ofsampling, and cannot be extended to all the muscles involved.Moreover, repeating muscle biopsy is invasive and imprac-tical for monitoring disease, especially in children. Therefore,it is important to have non-invasive tools, both to assessmuscle involvement for recruitment of patients in clinicaltrials, and to evaluate the response to treatments [14–16].

Several investigators [9–14] used nonquantitative MRIto characterize the pattern of fatty infiltration in DMD using

a modified qualitative score. A correlation between the fattyinfiltration grade at MRI and clinical assessments has alsobeen demonstrated [10]. However, evaluation of nonquan-titative MRI findings of the fatty infiltration on T1-

Fig. 5 Graph shows correlation between mean MFF and timed Gowerscore in 17 patients (Spearman’s test: P<0.03)

Fig. 4 Graph shows correlation between mean MFF and MedicalResearch Council score (MRCS) in 19 patients (Spearman’s test: P<0.001)

Fig. 3 Color-coded axial MFF maps show a low, b intermediate, andc high degrees of thigh involvement, in 7-, 9-, and 10-year-old boysrespectively

Skeletal Radiol (2012) 41:955–961 959

weighted images is a subjective method with limitations forassessing disease severity accurately. Similarly, mapping onT2-weighted images, which was recently proposed by Kimet al., is interesting [14]. However, although these authorsclaimed that this technique is quantitative; actually, it is notquantitative as it measures only the signal intensity and notthe amount of muscle fat infiltration [14].

Recently, Gaeta et al. [16] demonstrated in a cohort of 27patients with neuromuscular disorders that the MFFcalculated with the dual-echo dual-flip angle SPGR (two-point Dixon) MRI technique shows good agreement withmuscle histopathology. Dual-echo dual-flip angle SPGR isa rapid technique available on all routine MRI systems.Most recently, the same author applied this technique infive patients with late-onset Charcot–Marie–Tooth diseasetype 2, and found that it is a valuable tool for rapid MFFcalculation [17].

Our study showed that DMD has a characteristicdistribution with more precocious and severe involvementof the gluteus maximus and adductor magnus muscles. Onthe other hand, gracilis muscles are spared or conspicuouslyless involved. Our data confirmed the pattern of diseasedistribution described with nonquantitative MRI [10, 14],but permitted a quantitative evaluation of the degree ofmuscle involvement by MFF calculation. We found thatMRI calculation of MFF strongly correlated with diseaseseverity indicated by clinical assessments in patients withDMD. Two points are worth discussing. First, we foundthat an increase of 20% in MFF is associated with a highrisk of functional reduction. Second, we noted that patients

with an MFF higher than 75% are unable to perform thetimed Gower score. This previously unreported evidencemay suggest that this level of fatty infiltration couldrepresent a break-point for a valuable motor function. Thishypothesis has to be further explored in a larger series ofpatients.

Moreover, this method permits accurate evaluation ofsingle muscle involvement, which is a very importantparameter for genetic therapy. MFF color-coded mapping isa rapid visual method of assessment of muscle involvementthat could turn out to be useful for the injection of drugs.Therefore, we believe that MFF calculation may greatlyfacilitate multicenter genetic therapy trials involvingpatients with DMD.

Other MRI methods for MFF calculation are worthdiscussing, particularly the three-point Dixon technique andits variants [26]. Wren et al. [15] explored the usefulness ofthe three-point Dixon MRI technique in the assessment ofdisease severity in patients with DMD. The MFF calculatedwith this technique correlated more strongly with diseaseprogression than measurements of muscle strength. Themajor limitations of Wren et al.’s study were the smallcohort of patients and the lack of comparison with astandard of reference such as muscle biopsy and/or MRspectroscopy (MRS).

In the last few years, many authors have demonstratedthe usefulness of variants of the three-point Dixontechnique, particularly iterative decomposition of waterand fat with echo asymmetry and least-squares estimation(IDEAL), for the quantification of tissue fat content [9, 22,23]. Recently, Bernard et al. [27] compared MRS, withIDEAL, dual-echo dual-flip angle SPGR (two-point-Dixon), fat saturation, and water-equivalent suppressiontechniques for the quantification of fat content in phantomsat 3.0 T. All methods correlated significantly with MRS,with IDEAL demonstrating the higher precision. Of note,there was a general underestimation of all sequences with ahigh fat fraction (more than 80–90%). However, IDEAL itis not available on all MRI routine systems. Moreover, toour knowledge, this technique has never been used to studypatients with neuromuscular disorders and, consequently,its results have never been compared with those of musclebiopsies. Despite this, IDEAL is a very promising tech-nique for in vivo MFF calculation and studies have to beperformed to explore its potential value.

Some weaknesses of our study must be acknowledged.The major limitation is the small number of patientsenrolled. Although our preliminary data have to beconfirmed with further and larger studies, they are verypromising. Another limitation of our study is the absence ofa comparison with MR spectroscopy (MRS). However, thevalue of MRS is limited in evaluating multiple muscles as itexplores only small portions of muscle, is time-consuming

Fig. 6 Graph shows correlation between mean MFF and time to run10 m in 20 patients (Spearman’s test: P<0.001)

960 Skeletal Radiol (2012) 41:955–961

and operator-dependent, and it is susceptible to static fieldheterogeneity. Moreover, it is not available in many MRIsystems.

In conclusion, our study demonstrated that MFF calcu-lation and mapping obtained with a dual-echo dual-flipangle SPGR MRI technique are useful for evaluating thedegree and pattern of distribution of muscle fatty infiltrationin patients with DMD and, in our opinion, should beconsidered as a substitute for nonquantitative MRI meth-ods. Moreover, we believe that the use of MFF color-codedmaps in choosing the sites of drug injection for genetictherapy is worth exploring in the future.

References

1. Dubowitz V. The muscular dystrophies. In: Muscle disorders inchildhood. 2nd ed. London: Saunders, 1995; 39–42.

2. Mercuri E, Mayhew A, Muntoni F, et al. On behalf of the TREAT-NMD Neuromuscular Network. Towards harmonisation of out-come measures for DMD and SMA within TREAT-NMD; reportof three expert workshops: TREATNMD/ENMC workshop onoutcome measures, 12th–13th May 2007, Naarden, The Nether-lands; TREAT-NMD workshop on outcome measures in experi-mental trials for DMD, 30th June–1st July 2007, Naarden, TheNetherlands; Conjoint Institute of Myology TREAT-NMD Meet-ing on physical activity monitoring in neuromuscular disorders,11th July 2007, Paris, France. Neuromuscul Disord. 2008;18:894–903.

3. Brooke MH, Griggs RC, Mendell JR, Fenichel GM, Shumate JB,Pellegrino RJ. Clinical trial in Duchenne dystrophy. The design ofthe protocol. Muscle Nerve. 1981;4:186–97.

4. Scott OM, Hyde SA, Goddard C, Dubowitz V. Quantitation ofmuscle function in children: a prospective study in Duchennemuscular dystrophy. Muscle Nerve. 1982;5:291–301.

5. Florence JM, Pandya S, King WM, et al. Intrarater reliability ofmanual muscle test (Medical Research Council scale) grades inDuchenne’s muscular dystrophy. Phys Ther. 1992;72:115–22.

6. Mayhew JE, Florence JM, Mayhew TP, et al. Reliable surrogateoutcome measures in multicenter clinical trials of Duchennemuscular dystrophy. Muscle Nerve. 2007;35:36–42.

7. Mazzone ES, Messina S, Vasco G, et al. Reliability of the NorthStar Ambulatory Assessment in a multicentric setting. Neuro-muscul Disord. 2009;19:458–61.

8. Mazzone E, Martinelli D, Berardinelli A, et al. North StarAmbulatory Assessment, 6-minute walk test and timed items inambulant boys with Duchenne muscular dystrophy. NeuromusculDisord. 2010;20:712–6.

9. Schreiber A, Smith WL, Ionasescu V, et al. Magnetic resonanceimaging of children with Duchenne muscular dystrophy. PediatrRadiol. 1987;17:495–7.

10. Liu GC, Jong YJ, Chiang CH, Jaw TS. Duchenne musculardystrophy: MR grading system with functional correlation.Radiology. 1993;186:475–80.

11. Marden FA, Connolly AM, Siegel MJ, Rubin DA. Compositionalanalysis of muscle in boys with Duchenne muscular dystrophyusing MR imaging. Skeletal Radiol. 2005;34:140–8.

12. Huang Y, Majumdar S, Genant HK, et al. Quantitative MRrelaxometry study of muscle composition and function inDuchenne muscular dystrophy. J Magn Reson Imaging.1994;4:59–64.

13. Phoenix J, Betal D, Roberts N, Helliwell TR, Edwards RH.Objective quantification of muscle and fat in human dystrophicmuscle by magnetic resonance image analysis. Muscle Nerve.1996;19:302–10.

14. Kim HK, Laor T, Horn PS, Racadio JM, Wong B, Dardzinski BJ.T2 mapping in Duchenne muscular dystrophy: distribution ofdisease activity and correlation with clinical assessments. Radiol-ogy. 2010;225:899–908.

15. Wren TA, Bluml S, Tseng-Ong L, Gilsanz V. Three-pointtechnique of fat quantification of muscle tissue as a marker ofdisease progression in Duchenne muscular dystrophy: preliminarystudy. AJR Am J Roentgenol. 2008;190:W8–12.

16. Gaeta M, Scribano E, Mileto A, et al. Muscle fat fraction inneuromuscular disorder: dual-echo dual-flip-angle spoiledgradient-recalled MR imaging technique for quantification—afeasibility study. Radiology. 2011;259:487–94.

17. Gaeta M, Mileto A, Mazzeo A, et al. MRI findings, patterns ofdisease distribution, and muscle fat fraction calculation in fivepatients with Charcot-Marie-Tooth type 2 F disease. SkeletalRadiol 2011;doi:10.1007/s00256-011-1199-y.

18. Rosset A, Spadola L, Ratib O. OsiriX: an open-source softwarefor navigating in multidimensional DICOM images. J DigitImaging. 2004;17:205–16.

19. Muntoni F, Bushby KD, Van Ommen G. 149th ENMC Interna-tional Workshop and 1st Treat-NMD Workshop on “Planningphase I/II clinical trials using systematically delivered antisenseoligonucleotides in Duchenne muscular dystrophy”. NeuromusculDisord. 2008;18:268–75.

20. Eagle M, Baudouin S, Chandler C, Giddings DR, Bullock R,Bushby K. Survival in Duchenne muscular dystrophy: improve-ments in life expectancy since 1967 and the impact of homenocturnal ventilation. Neuromuscul Disord. 2002;12:926–9.

21. Wagner KR. Approaching a new age in Duchenne musculardystrophy treatment. Neurotherapeutics. 2008;5:583–9.

22. Yokota T, Takeda S, Lu QL, Partridge TA, Nakamura A, HoffmanEP. A renaissance of antisense oligonucleotide drugs in neurology:exon skipping breaks new ground. Arch Neurol. 2009;66:32–8.

23. Muntoni F, Wels D. Genetic treatments in muscular dystrophies.Curr Opin Neurol. 2007;20:590–4.

24. Kinali M, Arechavala-Gomeza V, Feng L, et al. Local restorationof dystrophin expression with the morpholino oligomer AVI-4658in Duchenne muscular dystrophy: a single-blind, placebo con-trolled, dose-escalation, proof-of-concept study. Lancet Neurol.2009;8:918–28.

25. Sookhoo S, Mackinnon I, Bushby K, Chinnery PF, Birchall D.MRI for the demonstration of subclinical muscle involvement inmuscular dystrophy. Clin Radiol. 2007;62:160–5.

26. Ma J. Dixon techniques for water and fat imaging. J Magn ResonImaging. 2008;28:543–58.

27. Bernard CP, Liney GP, Manton DJ, Turnbull LW, Langton CM.Comparison of fat quantification methods: a phantom study at3.0 T. J Magn Reson Imaging. 2008;27:192–7.

Skeletal Radiol (2012) 41:955–961 961