In vivo muscle morphology comparison in post-stroke survivors using ultrasonography and diffusion tensor imaging

This is a sel -a chi ed e sion o an o iginal a icle. This e sion may di e om he o iginal in pagina ion and ypog aphic de ails. Au ho (s): Ti le: Yea : Ve sion: Copy igh : Righ s: Righ s u l: Please ci e he o iginal e sion: CC BY 4.0 h ps://c ea i ecommons.o g/licenses/by/4.0/ In i o muscle mo phology compa ison in pos -s oke su i o s using ul asonog aphy and di usion enso imaging © The Au ho (s) 2019 Published e sion Kö ing, Cla a; Schlippe, Ma ius; Pe e sson, S en; Penna i, Gaia Valen ina; Ta asso a, Olga; A nd , An on; Finni, Taija; Zhao, Kangqiao; Wang, Ruoli Kö ing, C., Schlippe, M., Pe e sson, S., Penna i, G. V., Ta asso a, O., A nd , A., Finni, T., Zhao, K., & Wang, R. (2019). In i o muscle mo phology compa ison in pos -s oke su i o s using ul asonog aphy and di usion enso imaging. Scien i ic Repo s, 9, A icle 11836. h ps://doi.o g/10.1038/s41598-019-47968-x 2019 1 Scien i ic RepoR S | (2019) 9:11836 | h ps://doi.o g/10.1038/s41598-019-47968-x www.na u e.com/scien i ic epo s In i o muscle mo phology compa ison in pos -s oke su i o s using ul asonog aphy and di usion enso imaging cla a Kö ing1, Ma ius Schlippe1, S en pe e sson2,3, Gaia Valen ina penna i4, olga a asso a5, An on A nd 5,6, aija inni7, Kangqiao Zhao1 & Ruoli Wang 1,8,9 Skele al muscle a chi ec u e signi ican ly in luences he pe o mance capaci y o a muscle. A DTI-based me hod has been ecen ly conside ed as a new e e ence s anda d o alida e measu emen o muscle s uc u e in i o. This s udy sough o quan i y muscle a chi ec u e pa ame e s such as ascicle leng h ( L), penna ion angle (pA) and muscle hickness ( m) in pos -s oke pa ien s using di usion enso imaging (DTI) and o quan i a i ely compa e he di e ences wi h 2D ul asonog aphy (US) and DTI. Muscle ascicles we e econs uc ed o examine he ana omy o he medial gas ocnemius, pos e io soleus and ibialis an e io in se en s oke su i o s using US- and DTI-based echniques, espec i ely. By aligning he US and DTI coo dina e sys em, DTI econs uc ed muscle ascicles a he same scanning plane o he US da a can be iden i ied. The a chi ec u e pa ame e s es ima ed based on wo imaging modali ies we e u he compa ed. Signi ican di e ences we e obse ed o PA and m be ween wo me hods. Al hough mean FL was no signi ican ly di e en , he e we e conside able in a-indi idual di e ences in FL and PA. On he indi idual le el, pa ame e s measu ed by US ag eed poo ly wi h hose om DTI in bo h deep and supe icial muscles. The signi ican di e ences in muscle pa ame e s we obse ed sugges ed ha he DTI-based me hod seems o be a be e me hod o quan i y muscle a chi ec u e pa ame e s which can p o ide impo an in o ma ion o ea men planning and o pe sonalize a compu a ional muscle model. Skele al muscles p o ide s eng h and p o ec ion by dis ibu ing loads and abso bing shocks. They pe o m bo h dynamic wo k, such as locomo ion o posi ioning o he body segmen s in space, and s a ic wo k, such as main- aining body pos u e o posi ion1. Skele al muscle a chi ec u e signi ican ly in luences he pe o mance capaci y o a muscle. Pa ame e s ha a e o en used o desc ibe he muscle a chi ec u e a e ascicle leng h (FL), penna ion angle (PA), muscle hickness ( m) and physiological c oss-sec ional a ea (PCSA)2. I has also been shown ha he mac oscopic a angemen o muscle ibe s can di e signi ican ly be ween indi iduals e.g., s oke pa ien s3, young/old indi iduals4, male/ emale5. In i o measu emen s he e o e limi he accu acy o muscle s uc u e pa ame e s when applied o indi idual pe sons o popula ions. In i o measu emen s o muscle a chi ec u e can p o ide insigh s in o he in e -indi idual a iabili y o muscle unc ion as well as e e ence da a o he de el- opmen o pe sonalized muscle models o compu a ional modeling. Con en ionally, 2D ul asound (US) and magne ic esonance imaging (MRI) a e o en used o quan i y muscle s uc u e in i o. Howe e , 2D US has limi a ions, since i only p o ides 2D images wi h a limi ed ield o iew, and a misalignmen o he US ans- duce can lead o inaccu a e measu emen s6,7. Mo eo e , assessing deepe muscles wi hin he body such as he an e io soleus and ibialis pos e io in he human lowe limb can be di icul using US because image quali y is 1Depa men o Mechanics, Royal Ins i u e o Technology, S ockholm, Sweden. 2Depa men o Medical Radia ion Physics and Nuclea Medicine, Ka olinska Uni e si y Hospi al, S ockholm, Sweden. 3Depa men o Clinical Science In e en ion and Technology, Ka olinska Ins i u e , S ockholm, Sweden. 4Ka olinska Ins i u e , Depa men o Clinical Sciences, Dande yd Hospi al, Di ision o Rehabili a ion Medicine, S ockholm, Sweden. 5The Swedish School o Spo and Heal h Sciences, S ockholm, Sweden. 6Depa men o CLINTEC, Ka olinska Ins i u e , S ockholm, Sweden. 7Facul y o Spo and Heal h Sciences, Uni e si y o Jy äskylä, Jy äskylä, Finland. 8Depa men o Child en’s and Women’s Heal h, Ka olinska Ins i u e , S ockholm, Sweden. 9KTH Biomex Cen e , Royal Ins i u e o Technology, S ockholm, Sweden. Co espondence and eques s o ma e ials should be add essed o R.W. (email: [email p o ec ed]) Recei ed: 11 Ma ch 2019 Accep ed: 23 July 2019 Published: xx xx xxxx open 2 Scien i ic RepoR S | (2019) 9:11836 | h ps://doi.o g/10.1038/s41598-019-47968-x www.na u e.com/scien i ic epo s www.na u e.com/scien i ic epo s/ o en poo 5. The ana omical MRI me hod can o e come some o he limi a ions o US8 bu lacks he esolu ion o examine indi idual muscle ascicles. O e he las decade, app oaches using di usion enso imaging (DTI) o econs uc muscle ascicles ha e e ol ed as an al e na i e e e ence s anda d7,9,10. Muscle DTI uses an MRI p o ocol o measu e he di usion o wa e molecules in he muscle issue and can iden i y muscle ascicle di ec ion because wa e molecules mo e p ima ily along he long axis o ascicles11. S udies ha e demons a ed ha econs uc ed ascicles om DTI aligned well wi h ue muscles ascicles in a 11 and human skele al muscles10. Heemske k e al.12 and Bols e lee e al.10 bo h e alua ed he wi hin-day and be ween-day epea abili y o DTI ac og aphy. In bo h s udies he epea abili y o measu emen s was shown o be accep able. Bols e lee e al. p oposed a alida ion me hod based on DTI ascicle acking compa ed wi h US measu es o FL and PA wi h he same measu es ob ained using DTI in heal hy gas ocnemius10. On a e age, US yielded sligh ly longe measu emen s o FL compa ed o DTI. Al hough mean measu emen s did no di e s a is ically, he in a-indi idual measu es subs an ially a ied. PA measu ed wi h US we e ound o be signi ican ly smalle han he PA measu ed wi h DTI. To con i m he accu acy o mus- cle a chi ec u e pa ame e s ob ained using he DTI me hod, Bols e lee e al. u he compa ed he a chi ec u e pa ame e s o soleus o hose ob ained based on cada e ic dissec ion9. Dissec ion measu emen s o FL we e simila o hose measu ed wi h DTI, bu inconsis en PAs we e obse ed in he pos e io compa men o he soleus. Simila imaging echniques ha e also been applied o de e mine muscle olumes, leng hs, c oss sec ional a ea and ibe o ien a ions o he soleus in child en wi h ce eb al palsy (CP)13. Compa ed o able-bodied con ols, c oss sec ional a ea pe pendicula o he muscle ibe di ec ion we e educed, bu no di e ences we e obse ed in ascicle leng h. The au ho s concluded ha a be e unde s anding o muscle a chi ec u e may mo i a e new he apies and help o p edic hei ou comes in child en wi h CP13. S oke is a majo global heal h p oblem; i is he second-mos common cause o dea h and one o he leading causes o adul disabili y14. The skele al muscle may unde go nume ous s uc u al and unc ional al e a ions ol- lowing such an inju y o he b ain15. Fo ins ance, seconda y changes a e spas ici y a e commonly seen in many neu ological diso de s and impac app oxima ely 30% o s oke pa ien s16. P e ious s udies ha e indica ed ha spas ic muscles may unde go mo phological changes which al e he o ce gene a ion capaci y du ing mo emen and lead o an inc eased join and muscle s i ness in hemiplegic pe sons3,15. Muscles a ound he ankle a e o en a ec ed a e s oke, and his con ibu es conside ably o he disabili y17. The impai ed ankle do si lexo s, spas ic plan a lexo s, and associa ed oo d op make i di icul o he oo o clea om he g ound du ing gai . In addi ion, weak plan a lexo s also esul in a educed o wa d p opulsion18,19. Shea wa e elas og aphy was applied on s oke su i o s and pe sons wi h ce eb al palsy in he biceps b achii, medial gas ocnemius and ibialis an e- io 15,20,21. G ea e shea wa e eloci ies we e ound in he a ec ed muscles, which indica ed highe muscle s i - ness in biceps b achii and medial gas ocnemius on he pa e ic side20. Howe e , no signi ican di e ences we e obse ed in he ibialis an e io 15,20. The e o e, au ho s concluded ha muscles need o be e alua ed indi idually o assess al e a ions20. Yang e al. assessed bila e al di e ences in m, FL and PA o he gas ocnemius in s oke pa ien s using US and ound ha m and PA we e signi ican ly la ge bu ha FL was signi ican ly smalle in he a ec ed side22. Al hough US is widely used o quan i y muscle a chi ec u e ea u es, e idence o he alidi y o 2D US o measu ing a ibu es such as FL and PA in human is limi ed6. The ideli y o hese measu emen in muscles wi h dis up ed ascicles o ib osis is unce ain, which likely occu s in pos -s oke su i o s23. The hype -echoic appea ance in muscle due o a in il a ion and inc eased ib ous issue a e neu ological diso de s, e.g., s oke, makes he muscle bounda y and ascicle less dis inc in US image. This ea u e leads o challenges and po en ial e o s in muscle mo phological pa ame e quan i ica ion in 2D US, especially in he deep muscles such as soleus. Mo eo e , 2D US imaging o muscle a chi ec u es has been sugges ed as a aluable ool o e alua e he unc ional imp o emen o muscles a ec ed by neu ological disease a e an in e en ion p og am24, and he knowledge o speci ic pa ame e s ha a e likely o be un eliable is essen ial o such clinical applica ions. The aim o his s udy is wo- old: i s , we seek o desc ibe muscle a chi ec u e pa ame e s such as FL, PA and m in medial gas ocnemius (GA), pos e io soleus (PSO) and ibialis an e io (TA) in pos -s oke pa ien s in i o using inno a i e DTI-based echniques; second, we seek o de e mine whe he indi idual muscle mo phological pa ame e s in he GA, he PSO and he TA as iden i ied h ough 2D US a e simila o hose ob ained h ough DTI. Ou indings may p o ide impo an in o ma ion in ehabili a ion s a egy planning o s oke and shed ligh on he de elopmen o pe sonalized compu a ional muscle models. Based on he physics o ul asonog- aphy and on he inhe en di e ences in muscle echogenici y in neu ologically pa hological muscle issue, ou hypo hesis was ha he pa ame e s ha desc ibe muscle a chi ec u e measu ed by he wo me hods would ag ee in supe icial muscles bu no in deep muscles. Me hods Pa icipan s. In o al, ele en hemiplegic pos -s oke su i o s we e ec ui ed om a local ehabili a ion clinic (Depa men o Rehabili a ion Medicine, Dande yd Hospi al, S ockholm, Sweden). Fou inclusion c i e ia we e de e mined: (1) s oke >6 mon hs p io o inclusion; (2) no an i-spas ic ea men wi hin h ee mon hs; (3) eligibili y o MR scanning, and (4) absence o o he lowe limb inju ies o diso de s. US measu emen and DTI da a acquisi ion occu ed in wo sepa a e sessions bu wi hin one mon h. All pa icipan s ga e w i en in o med consen acco ding o he Decla a ion o Helsinki. The s udy was app o ed by he Regional E hics Commi ee, S ockholm, Sweden. All expe imen s we e pe o med in acco dance wi h ele an guidelines and egula ions. Th ee pa icipan s d opped ou a e ei he he US o MRI measu emen s. Da a om one pa icipan we e excluded because o he poo image quali y, caused by mo ion a i ac s. The e o e, expe imen al da a om se en pa icipan s we e included o u he analysis (see Table1 and Supplemen a y A TableA1). Da a acquisi ion. Du ing US da a acquisi ion, pa icipan s we e sea ed in a com o able semi-up igh posi- ion and we e ins uc ed o emain elaxed and s a ic du ing he measu emen wi h hei knee lexed a 30° and 3 Scien i ic RepoR S | (2019) 9:11836 | h ps://doi.o g/10.1038/s41598-019-47968-x www.na u e.com/scien i ic epo s www.na u e.com/scien i ic epo s/ ankle a 10° plan a lexion. Thei oo was ixa ed o a oo pla e connec ed o a dynamome e (IsoMed 2000, D&R GmbH, Hemau, Ge many). The US da a we e collec ed using an M9 Mind ay ul asonog aphy sys em (Mind ay M9, Shenzhen, China) wi h a 38 mm wide linea ansduce (6–14 MHz). The imaging in es iga o , a licensed physician, ecei ed sonog aphy aining in a sepa a e ul asonog aphic clinic be o e aking he images o his s udy. The same in es iga o imaged all pa icipan s. In o de o de e mine he coo dina e sys em o US measu emen (in a leg-based e e ence ame), a mo ion cap u e sys em (Qualisys, Go henbu g, Sweden) in com- bina ion wi h he US sys em was u ilized. Eigh e lec i e ma ke s we e placed on ana omical landma ks o he pa icipan ’s a ec ed leg and h ee ma ke s we e placed on he US ansduce (Fig.1A,B). The ma ke loca ions we e la e used o de e mine he loca ion and o ien a ion o he US image wi h espec o a leg-based ame. Subjec Gende Age (Yea s) Weigh (Kg) Heigh (cm) BMI (kg/m2)Lesion Type Pos S oke Time (Mon hs) S1 M 50 85 169 29,8 Haemo hagic 7 S2 M 31 79 178 24,9 Haemo hagic 8 S3 F 71 88 182 26,6 Ischemic 29 S4 F 67 50 153 21,4 Haemo hagic 60 S5 M 58 61 181 18,6 Haemo hagic 84 S6 M 57 86 174 28,4 Haemo hagic 37 S7 M 34 70 168 24,8 Haemo hagic 25 Table 1. Cha ac e is ics o pa icipan s. Figu e 1. (A,B) Ma ke placemen and expe imen al se -up du ing he ul asound measu emen . Ma ke s P1, P2, P3 and P4 we e ixed on o a ec angle silicon pad and a ached o he lowe limb. Ma ke P5 and P6 we e placed on he head o he ibula and he ube osi y o he ibia, espec i ely. Ma ke P7 was placed 10 cm dis al o P6 along he ibia bone. P8 was placed on he medial side o he epicondyle. Ma ke P4, P7 and P6 we e ca e ully adjus ed in o de o align in a s aigh line. AL, AR and AU we e he ma ke s placed on he US ansduce , whe e AL and AR de ined he acquisi ion window o he ansduce . (C) Example o he measu emen o he pa ame e s penna ion angle (PA), muscle hickness ( m) and ascicle leng h (FL) o he ibialis an e io muscle in an US image using MATLAB (g een lines: deep and supe icial aponeu osis, ed lines: selec ed ascicles). The muscle hickness was iden i ied by se ing wo poin s (blue ma ke s) ma king each o he aponeu oses. 4 Scien i ic RepoR S | (2019) 9:11836 | h ps://doi.o g/10.1038/s41598-019-47968-x www.na u e.com/scien i ic epo s www.na u e.com/scien i ic epo s/ Du ing DTI da a acquisi ion, all pa ien s we e scanned bila e ally using a 3T MRI scanne (Siemens T io, Siemens Medical Solu ion, E langen, Ge many) while lying in a supine posi ion whe e he high was es ing on a wedge and he oo s apped in a oo pla e. The con igu a ion o he knee and ankle angles was se o be iden ical o ha in he US measu emen . P io o scanning, eigh ma ke s we e also placed on he landma ks o he pa icipan ’s a ec ed side by he same ins uc o (in he same loca ion as in he US measu emen ), which we e isible in T1-weigh ed MR images. The T1-weigh ed images we e ob ained wi h he ollowing se ings: TSE sequence, TR/TE 605 ms/23 ms, ield o iew (FOV) 430 mm, acquisi ion ma ix 512 × 299 pixel, oxel size 0.84 × 0.84 × 5 mm and scan ime 101 s. The se ings o he DTI images we e: EPI sequence, FOV 350 mm, acqui- si ion ma ix 140 × 140 pixel, oxel size 2.5 × 2.5 × 2.5 mm, 20 DTI g adien di ec ions, numbe o signal a e ages 4, b = 500 s/mm2 (B0 image wi h b = 0 s/mm2, wi h 2 signal a e ages), and scan ime 520 s. Ul asound ascicle econs uc ion. US images we e analyzed using a cus om-w i en MATLAB p og am ( e sion: R2017b) ha was, simila o p e ious epo s25. The deep and supe icial aponeu oses o each espec i e muscle we e i s ly iden i ied. Th ee muscle ascicles ( o as long as he ascicles we e clea ly isible) in he p ox- imal, dis al, and in e media e pa o he muscle we e hen selec ed in he image. Finally, he m was iden i ied as he dis ance be ween he supe icial and deep aponeu oses (Fig.1C). The deep and supe icial PAs we e iden i ied as he angles o med be ween he ascicles and he line o he p oximal and dis al aponeu oses. The epo ed PA esul ed om calcula ing he a e age o he deep and supe icial PAs. The FL was calcula ed as he s aigh -line dis ance by di iding he m by he sine o he deep PA. DTI ascicle econs uc ion. The econs uc ion o muscle ascicles and calcula ion o he DTI-based muscle pa ame e s we e modi ied based on a p e iously epo ed me hod7. B ie ly, i consis ed o ou s eps: (1) segmen a ion o muscles based on T1-weigh ed images; (2) denoising o DTI da a using a Local P inciple Componen Analysis il e 26; (3) ascicle ac og aphy using DSI s udio o gene a e 100 ac s h ough a small egion wi hin he a ge muscle [se ings: 0.7 < ac ional aniso opy <0.1, maximum angle be ween ac seg- men s 10°, 20 mm ≤ ac leng h ≤ 200 mm; s ep size = 1.0 mm)]27; and (4) muscle pa ame e iden i ica ion (Fig.2). The muscle segmen a ion o his p ojec was ca ied ou manually o each pa ien using he open-sou ce pla o m 3D Slice (on a e age 80 slices pe scan wi h a spacing o 5 mm be ween slices). 3D Slice was used o cal- cula e 3D su ace models om he segmen ed image da a which is done by using a pipeline o algo i hms: c ea ing a bina y label map om he segmen a ion, gene a ing a ma ching cubes model, and unning iangle educ ion and iangle smoo hing using he smoo hing ac o o 0.5. To ensu e co ec segmen a ions, he segmen a ions we e e i ied by an independen expe . To calcula e ascicle pa ame e s, he ac s we e o e lain o e he su ace mesh calcula ed om he muscle segmen a ion. To econs uc he end poin s o he ibe ac s, he median x-, y- and z-coo dina es o he 100 endpoin s a ei he end o he ac s we e de e mined and owa ds he muscle su ace along he line connec ing bo h median endpoin s un il he su ace was in e sec ed. The FL was ob ained by calcula ing he Euclidean dis ance be ween he wo endpoin s on he supe icial and deep aponeu osis, and he PA was calcula ed as he a e age angle o he median ac o he no mal ec o s o all o he su ace iangles inside a adius o 5 mm a ound he espec i e endpoin . The epo ed PA esul ed om calcula ing he a e age o he deep and supe icial PAs. The m was measu ed in he T1-weigh ed MRI images wi hin 3D slice by ma king each o he aponeu oses manually. Alignmen o ul asound and MRI coo dina e sys em. In o de o loca e DTI acks wi hin he muscle a he same scanning plane o he US da a, he US coo dina e sys em and he MRI coo dina e sys em we e aligned. The alignmen was based on a ou -ma ke clus e : P1, P2, P3 and P4 on he ibia (Fig.1A,B). The addi ional Figu e 2. P ocedu e o ascicle acking based on MRI and DTI da a. The ana omical MRI images we e manually segmen ed, and a su ace model o each muscle was c ea ed. In pa allel, he DTI da a we e econs uc ed wi hin DSI s udio and 100 ascicle ac s (in ligh blue) we e acked based on a small seeding a ea (in pu ple). The ac s we e impo ed in o MATLAB (5, e sion: R2017b) oge he wi h he su ace models. The median end poin s o all 100 ac s we e calcula ed and ansla ed owa ds he muscle su ace in bo h di ec ions. When he su ace was in e sec ed, he FL was ob ained by calcula ing he dis ance be ween bo h median endpoin s. The supe icial and deep penna ion angle PAS and PAD, ep esen he angle o he median ac o he su ace o he muscle olume in bo h endpoin s. 5 Scien i ic RepoR S | (2019) 9:11836 | h ps://doi.o g/10.1038/s41598-019-47968-x www.na u e.com/scien i ic epo s www.na u e.com/scien i ic epo s/ ma ke s, AL and AR, on he US ansduce we e used o de ine he acquisi ion window o he ansduce . The coo dina e sys em alignmen was pe o med by ans o ming he coo dina es o he ma ke s AL, AR and AU om he US global coo dina e sys em o he MR global coo dina e sys em and c ea ing a scanning plane based on he h ee ma ke s on he p obe. The c ea ed plane was displayed oge he wi h he segmen ed muscle olumes and he ma ke posi ions AR and AL indica ing he posi ion o he ansduce on he pa ien ’s skin. Fo compa ing US and DTI measu emen s, ibe acking was pe o med a he si e o he in e sec ion o he US plane wi h he segmen ed muscle su aces (Fig.3A). Da a analysis. FL, PA and m we e ob ained o each pa icipan in h ee muscle compa men s o he a ec ed limb (GA, PSO and TA) using wo me hods (i.e., US and DTI). Fo muscle pa ame e s based on US da a, h ee muscle ascicles in he p oximal, dis al and in e media e pa o he muscle we e hen selec ed in he image and FL and PA we e a e aged o each muscle. Fo muscle pa ame e s based on DTI da a, h ee seed egions we e chosen, and he esul ing FL and PA alues we e a e aged o each muscle. m was measu ed in he MRI images manually in h ee slices and a e aged (Fig.3B,C). S a is ical analyses we e pe o med using IBM SPSS S a is ics (IBM Co p., A monk, NY, USA). Desc ip i e s a is ics we e epo ed as median and ange alues. Wilcoxon signed- ank es was used o in es iga e he di e ences be ween muscles pa ame e s iden i ied by DTI- and US-based me hods, espec i ely. Di e ences we e conside ed s a is ically signi ican when p ≤ 0.05. A mean (s anda d de ia ion) wi hin-subjec absolu e di e ences by sub ac ing ma ching pai s o US and DTI measu emen s we e also calcu- la ed. In o de o accoun o he e ec s o di e en pa icipan heigh , s a is ical analysis was also pe o med on he ascicle leng h and hickness no malized by heigh . Resul s 3D muscle mo phological pa ame e s o he a ec ed-side in pos -s oke pa icipan s. Th ee dis inc compa men s can be clea ly iden i ied om he MRI and DTI scans o all pos -s oke su i o s, excep he PSO om one pa icipan , which was excluded om u he analysis because he ac s wi hin a bundle did no show a consis en pa e n. Recons uc ed 3D muscle ascicles om an example pa icipan we e illus a ed in Fig.4 and showed a unipenna e s uc u e in he GA and PSO and a mo e complexed ascicle o ien a ions in TA. La ge wi hin-subjec di e ences we e obse ed in he PAs in all h ee muscles (Fig.5). Compa ed o GA and PSO, conside able in a-subjec di e ences in FL we e ound in TA (Table2). A e age FL anged om 16 mm o 59 mm in GA, 22 mm o 45 mm in PSO, and om 29 mm o 78 mm in TA espec i ely. A e age PA anged om 14° o 35° in GA, om 27° o 48° in PSO and om 20° o 41° in TA, espec i ely. Compa ison o muscle mo phological pa ame e s using US and DTI me hods. Almos all cases (app oxima ely 86%) an in e cep ing plane o he US image h ough he muscle olume could be calcula ed and ascicle acking as well as muscle hickness measu emen s could be ca ied ou in he same loca ion as he measu emen s pe o med in he US da a. Compa ed o he DTI-based me hod, signi ican ly smalle PAs we e obse ed in all h ee muscles (GA: p = 0.02, PSO: p = 0.03, TA: p = 0.03) using US-based measu emen s (Fig.6 Figu e 3. (A) Visualiza ion o he muscle olumes o he medial gas ocnemius (blue) and medial-pos e io soleus ( ed) wi h he ans o med US plane cu ing h ough he muscle olumes. The ed ma ke s speci ying he wid h o he US ansduce . (B) The isualiza ion o he US plane in e sec ion wi h he muscle olumes wi hin he ield o iew o he US ansduce ( ed ma ke s). (C) The co esponding MRI image o he muscles. The hickness measu emen s we e done wi hin he 3D slice so wa e using he p o ided ule ool24. 6 Scien i ic RepoR S | (2019) 9:11836 | h ps://doi.o g/10.1038/s41598-019-47968-x www.na u e.com/scien i ic epo s www.na u e.com/scien i ic epo s/ and Table2).The m was also ound o be signi ican ly smalle in medial GA (p = 0.05) and TA (p = 0.02). No signi ican di e ences we e ound in econs uc ed FL be ween he wo me hods. Howe e , a la ge s anda d de i- a ion was obse ed o wi hin-subjec di e ences be ween US and DTI measu emen s in FL in all h ee muscles (Table3, GA: 10.82 mm, PSO: 8.50 mm and TA: 15.47 mm). Compa able obse a ions we e ound by pe o ming analysis using no malized FL and m. Discussion This s udy used bo h US- and DTI-based echniques o examine he muscle ana omy o he lowe limb o s oke su i o s in i o. Signi ican di e ences we e obse ed in bo h PAs and m in all h ee muscles be ween he wo me hods. Al hough he measu ed FL was no signi ican ly di e en be ween he wo me hods, a la ge s anda d de ia ion was obse ed o he wi hin-subjec di e ences in bo h supe icial and deep muscles. These indings only pa ially suppo ou hypo hesis. To ou knowledge, his is he i s s udy compa ing hese wo imaging me hods in measu ing muscle a chi ec u e pa ame e s in pos -s oke pa icipan s. Muscle a chi ec u e is one de e minan o a muscle’s o ce-gene a ing capaci y. Fo ins ance, muscle FL is he p ima y de e minan o muscle excu sion, whe e sho e muscle ascicles ha e a smalle ange o excu sion ha Figu e 4. (A) The su ace model o medial gas ocnemius (GA, in ed), medial-pos e io soleus (PSO, in blue) and ibialis an e io (TA, in g een) we e c ea ed based on ana omical MR images. (B) The ans e se iew o he ana omical MR slice app oxima ely midway be ween he ankle and knee. (C) 3D econs uc ed ascicle acks o GA, PSO and TA based on DTI da a. 7 Scien i ic RepoR S | (2019) 9:11836 | h ps://doi.o g/10.1038/s41598-019-47968-x www.na u e.com/scien i ic epo s www.na u e.com/scien i ic epo s/ can de elop o ce and powe 28. S uc u e changes such as FL and PAs ha e been commonly in es iga ed using ul asonog aphy. Howe e , la ge a iabili y was ound in he epo ed da a in pos -s oke pa icipan s due o se - e al ac o s such as po en ial e o s in he US measu emen , he na u e o he inhomogenei y o he pa ien g oup, and di e en join con igu a ions in he measu emen . The p e iously epo ed mean alue o he FL and PA o he pa e ic GA a e s oke in a ully ex ended knee and a neu al ankle posi ion anged om 36 mm o 51 mm and 16° o 25°, espec i ely29–31. These p e iously epo ed he FL o GA a e sligh ly longe han hose ound in Figu e 5. Indi idual muscle a chi ec u e pa ame e s es ima ed using DTI-based me hod (FL: ascicle leng h, m: muscle hickness, PA: penna ion angle). FL and m we e no malized by indi idual’s heigh . 8 Scien i ic RepoR S | (2019) 9:11836 | h ps://doi.o g/10.1038/s41598-019-47968-x www.na u e.com/scien i ic epo s www.na u e.com/scien i ic epo s/ he p esen s udy. The disc epancy could be due o he lexed-knee posi ion we adop ed du ing he measu e- men . Ve y ew s udies ha e epo ed he muscle s uc u e pa ame e s o he PSO and TA using ul asonog aphy. Ramsay e al. epo ed mean FL and PA o 39.7 ± 10.4 mm and 18.0° ± 3.4° in he PSO29, and 26.38 ± 4.2 mm and 13.4° ± 2.7° in he TA, espec i ely32. Ou ascicle leng h o he PSO was wi hin he ange o epo ed alues. Howe e , he ascicle leng h o he TA was longe . The PAs o all muscles ell wi hin he ange o he epo ed alues. DTI-based muscle a chi ec u e measu es ha e been in oduced in ecen yea s and, can p o ide a h ee-dimensional measu emen o he muscle a chi ec u e. Howe e , he me hodology has only been applied in heal hy pe sons, and muscle s uc u e pa ame e s ha e only been epo ed in he GA and PSO. Compa ed o he epo ed da a om heal hy pa icipan s9,10, he FLs in ou pos -s oke su i o s we e sho e and he PAs we e la ge in he GA, hough he di e ences in he PSO we e no as dis inc as he o he wo muscles. Accu a e assessmen o changes in he muscle a chi ec u e can p o ide c ucial in o ma ion o planning ea - men modali ies such as s e ching and o ho ic managemen 28. The compa ison be ween US and DTI pa ame e s showed a su icien ag eemen o mean alues o FL in h ee muscles, which ag eed wi h a p e ious s udy com- pa ing US and DTI FL es ima ion in he GA in heal hy pa icipan s9. Simila o he abo emen ioned s udy, we also obse ed a la ge de ia ion o he wi hin-subjec di e ences be ween US and DTI FLs in he s oke su i o s (Table3), which wa an cau ion when using US-based measu emen s o guide ea men planning and pe son- alize FL in a musculoskele al model. Muscle hickness is conside ed o be s ongly co ela ed wi h muscle mass and s eng h33,34; hus, i is o en assessed by US in a clinical se ing. We ound signi ican ly la ge m es ima ion in GA and TA using DTI, bu no di e ences we e obse ed in PSO. This is likely due o a comp ession o he muscle by he p essu e applied by he US ansduce on he supe icial muscles. No da a ha e been published ha compa e US and DTI-based muscle hickness es ima ions compa ison ei he in heal hy o in pos -s oke pa ic- ipan s. Compa ed o GA and PSO, he TA hickness showed a sligh ly highe mean wi hin-subjec di e ence o 4.6 mm. Al hough he TA is ana omically di ided in o wo compa men s by a medial aponeu osis, he isibili y o he aponeu osis a ied in pa ien s om he US images, a ec ing he accu acy in calcula ing he hickness o he ibialis an e io . In ou s udy, mean PAs measu ed wi h US we e signi ican ly smalle han angles measu ed wi h DTI in pos -s oke su i o s. Howe e , he di e ences in PA be ween US and DTI a e la ge han he di e ences epo ed by Bols e lee e al.9 and Bé na d e al. in heal hy GA35. This could be due o se e al easons. Fo exam- ple, he muscle could be de o med by he comp ession om he US p obe. Rheological muscle p ope y changes such as ib osis and a ophy a e common in he spas ic muscle36. In p ac ice, inc eased p essu e was some imes necessa y o apply in he p obe in o de o achie e a be e isibili y o he ascicles. In addi ion, unlike he heal hy pa icipan s, join alignmen was di icul in some pa ien s due o spas ici y o pain, which may ha e caused some inconsis encies in join alignmen be ween he US and DTI measu emen s, e en hough he iden ical join align- men se -up was adop ed in bo h measu emen s. In clinical p ac ice, he ehabili a ion plan a e s oke is usually selec ed based on subjec i e clinical expe- ience a he han on an objec i e p edic ion o pos - ea men unc ion de eloped based on indi idual da a. Pe sonalized compu a ional models o he neu omusculoskele al sys em could he e o e acili a e he objec i e p edic ion o pa ien -speci ic unc ional ou come o di e en ea men designs. Using DTI-based muscle a chi- ec u e measu emen s, he complex muscle s uc u e o di e en compa men s and loca ions can be disce ned in i o. Based on ou compa ison in pos -s oke su i o s, FL and PA a e mo e sensi i e o he measu emen me hod, and TA is mo e sensi i e han he o he wo plan a lexo s. The e a e se e al limi a ions in he cu en s udy. Fi s , he signal in ensi y o he pos e io pa o he leg was lowe han he an e io pa o he leg in some pa icipan s. In he u u e, using an al e na i e coil se -up could imp o e he image in ensi y and lead o a be e quali y o he DTI da a. Second, he accu acy o he muscle pa ame e es ima ion could be imp o ed by using mo e han h ee seed poin s o each muscle compa men o by acking ascicles wi hin he whole muscle olume. Thi d, calcula ing FL as Euclidean dis ance be ween he wo endpoin s migh unde es ima e he ac ual ac s. Using a polynomial cu e i ing o accoun o he cu a u e o he ascicles may imp o e he accu acy in FL es ima ion. Fou h, u he in es iga ion including an age- and gende -ma ched con ol g oup would be aluable o p o ide addi ional in o ma ion o muscle mo pho- logical al e na ions a e s oke. Howe e , his was ou o he scope o he cu en s udy. Finally, he sample size o he pa ien s is small, and only he a ec ed side was measu ed using US due o he ime cons ain o he pa ien . Fu u e s udies wi h la ge coho s and mul iple measu emen s will imp o e he eliabili y and gene aliza ion o ou indings o he en i e s oke popula ion. Muscle Fasicle Leng h (FL) (mm) Penna ion Angle (PA) (°) Muscle Thickness ( m) (mm) US DTI US DTI US DTI GA 34.1 (23.8, 42.8) 28.4 (16.3, 59.4) 13.4 (7.8, 16.1) 29.3a (14.2, 34.8) 9.0 (7.6, 15.3) 11.6c (7.6, 15.5) PSO 34.0 (32.1, 51.2) 31.7 (21.6, 45.2) 20.1 (12.8, 27.2) 39.6b (26.7, 48.0) 11.3 (6.5, 15.0) 12.8 (10.8, 13.9) TA 45.0 (33.4, 75.5) 51.4 (28.7, 77.9) 7.4 (5.2, 15.3) 32.8b (22.4, 41.3) 8.2 (5.4, 12.4) 12.8a (9.3, 18.1) Table 2. Compa ison o measu emen s o he medial gas ocnemius (GA), pos e io soleus (PSO) and ibialis an e io (TA) o he a ec ed leg o he s oke su i o s using ul asound- (US) and di usion enso imaging- (DTI) based me hods, exp essed as median (Min, Max) o he g oup. Bold- aced alues indica ed ha he signi ican di e ences we e obse ed be ween wo me hods. ap = 0.02. bp = 0.03. cp = 0.05.