Recei ed: 14 Ma ch 2023 Re ised: 19 July 2023 Accep ed: 10 Augus 2023
DOI: 10.1002/m m.29846
RESEARCH ARTICLE
T1 mapping o myoca dium in a s using sel -ga ed
golden-angle acquisi ion
Ji
í Vi ouš1,2 Rado an Ji
ík1Tibo S aˇ
cina3Michal Hend ych4
Ja osla Nádeníˇ
cek3Ond
ej Macíˇ
cek1Ye Tian5Lucie K á ká1
E a D ažano á1,6 Ma ie No áko á3Pe Babula3
Roman Pano ský7,8 Edwa d DiBella9Zenon S a ˇ
cuk1
1Ins i u e o Scien i ic Ins umen s, Czech Academy o Sciences, B no, Czechia
2Facul y o Elec ical Enginee ing and Communica ion, B no Uni e si y o Technology, B no, Czechia
3Depa men o Physiology, Masa yk Uni e si y, Facul y o Medicine, B no, Czechia
4Fi s Depa men o Pa hology, S . Anne’s Uni e si y Hospi al and Facul y o Medicine Masa yk Uni e si y, B no, Czechia
5Vi e bi School o Enginee ing, Uni e si y o Sou he n Cali o nia, Los Angeles, Cali o nia USA
6Depa men o Pha macology, Facul y o Medicine, Masa yk Uni e si y, B no, Czechia
7In e na ional Clinical Resea ch Cen e , S . Anne’s Facul y Hospi al, Facul y o Medicine, Masa yk Uni e si y, B no, Czechia
81s Depa men o In e nal Medicine/Ca dioangiology, S . Anne’s Facul y Hospi al, Facul y o Medicine, Masa yk Uni e si y, B no Czechia
9School o Medicine, Uni e si y o U ah, Sal Lake Ci y, U ah USA
Co espondence
Ji
í Vi ouš, Ins i u e o Scien i ic
Ins umen s, Czech Academy o Sciences,
B no, Czechia.
Email: [email p o ec ed]
Funding in o ma ion
Eu opean Regional De elopmen Fund,
G an /Awa d Numbe :
CZ.02.1.01/0.0/0.0/16_019/0000868;
G an o á Agen u a ˇ
Ceské Republiky,
G an /Awa d Numbe : GA2210953S;
Léka
ská akul a, Masa yko a uni e zi a,
G an /Awa d Numbe s:
MUNI/11/SUP/09/2022,
MUNI/A/1379/2022; Minis e s o
Škols í, Mládeže a Tˇ
elo ýcho y,
G an /Awa d Numbe s:
EF18_046/0016045, LM2018129,
LM2023050
Abs ac
Pu pose: The aim o his s udy is o design a me hod o myoca dial T1 quan i i-
ca ion in small labo a o y animals and o in es iga e he e ec s o spa io empo al
egula iza ion and he needed acquisi ion du a ion.
Me hods:Wep oposeacomp essed-sensingapp oach oT1quan i ica ionbasedon
sel -ga ed in e sion- eco e y adial wo/ h ee-dimensional (2D/3D) golden-angle
s ack-o -s a sacquisi ionwi himage econs uc ionpe o medusing o al- a ia ion
spa io empo al egula iza ion. The me hod was es ed on a phan om and on a
heal hy a , as well as on a s in a small myoca dium- emodeling s udy.
Resul s: The esul s showed a good ma ch o he T1 es ima es wi h he esul s
ob ainedusing heg ound- u hme hodonaphan omandwi h heli e a u e alues
o a s myoca dium. The p oposed 2D and 3D me hods showed signi ican di e -
ences be ween no mal and emodeling myoca dium g oups o acquisi ion leng hs
down o app oxima ely 5 and 15 min, espec i ely.
Conclusions: A new 2D and 3D me hod o quan i ica ion o myoca dial T1 in
a s was p oposed. We ha e shown he capabili y o bo h echniques o dis in-
guishbe weenno maland emodelingmyoca dial issue.Weha eshown hee ec s
o image- econs uc ion egula iza ion weigh s and acquisi ion leng h on he T1
es ima es.
KEYWORDS
ca diac, ib osis, MRI, quan i ica ion, emodeling, e ospec i e ga ing, T1
This is an open access a icle unde he e ms o he C ea i e Commons A ibu ion-NonComme cial-NoDe i s License, which pe mi s use and dis ibu ion in any medium,
p o ided he o iginal wo k is p ope ly ci ed, he use is non-comme cial and no modi ica ions o adap a ions a e made.
© 2023 The Au ho s. Magne ic Resonance in Medicine published by Wiley Pe iodicals LLC on behal o In e na ional Socie y o Magne ic Resonance in Medicine.
368 wileyonlinelib a y.com/jou nal/m m Magn Reson Med. 2024;91:368–380.
VITOUŠ e al. 369
1INTRODUCTION
T1 quan i ica ion can de ec and quan i y pa hological
changes in issue s uc u e. In ca diology, i has been
success ully used o example in diagnosing myoca dial
ib osis, in lamma ion o in localiza ion o in a c ion.1
In p eclinical imaging, myoca dial T1 quan i ica ion can
show hee ec so d ugson hephysiologyo hehea and
ela ed de elopmen o po en ial pa hologies.2
In i o ca diac imaging has o be synch onized wi h
espec o ca diac and espi a o y ( o ee-b ea hing
acquisi ion) mo ion. This synch oniza ion can be done by
using signals om elec oca diog aphic (ECG) and espi-
a o y senso s.2
In ul a-high- ield (UHF) MRI, o example, clini-
cal MRI a 7T, and in p eclinical imaging, e y s ong
magne ic ields and g adien pulses wi h as slew a es
a e used in o de o ob ain good image esolu ion and
signal- o-noise a io (SNR). The ul a-high magne ic ield
in combina ion wi h blood low leads o he magne-
ohyd odynamic e ec 3and he as -swi ching g adien
pulses cause induc ion o noise in o he ECG leads.
Toge he , hese e ec s ende de ec ion o R wa es
and hence he ECG synch oniza ion di icul i no
impossible.
Becauseo hea o emen ionedp oblemsin UHFMRI,
me hods using e ospec i e na iga ion wi h na iga o
ex ac ion om hemeasu edMRda aha ebeenp oposed
in his ield.4-10
WhileclinicalT1 mapping o myoca dium hasbecome
an es ablished me hod, co esponding echniques o
small animals a e no well es ablished. This is because
o he e y di e en espi a o y and mainly ca diac a es
which a e much highe in mice and a s compa ed o
humans( a s s.humans:app ox.45 s.12b ea hspe min
and 260 s. 60 bea s pe min).11-13 In addi ion, b ea h-hold
acquisi ion (s anda d in clinical ca diac MRI) is no an
op ion in p eclinical MRI. This makes small-animal T1
quan i ica ion o myoca dium, he ocus o his pape , a
speci ic ask on i s own.
Mos o he known ca diac T1 quan i ica ion me h-
ods can be di ided in o wo basic g oups: a iable lip
angle7on one hand and In e sion Reco e y (IR) p e-
pa ed sequences4,6,14,15 (o sa u a ion eco e y p epa ed
sequences16) on he o he hand. Some mo e ad anced
echniques,suchasMR inge p in ing,17 s andasideo he
men ioned g oups.
The a iable lip angle me hods su e om high sen-
si i i y o B1inhomogenei y, p esen in ul a-high ield
MRI o a la ge ex en , he e o e B1maps ha e o be mea-
su ed. This is a complica ion in ca diac MRI as also he B1
mapping sequence has o be synch onized wi h espec o
ca diac (and espi a o y) mo ion.18
The second g oup o me hods is based on IR (o
sa u a ion eco e y) p epa a ion. They a e mos ly based
on heLook-Locke me hodwi hp ospec i esynch oniza-
ion wi h ECG,19 wi h he mos widely known Modi ied
Look-Locke in e sion eco e y20 o clinical ca diac T1
mapping and Small Animal Look-locke In e sion Reco -
e y21 o small-animal ca diac T1 mapping. Also a ech-
nique based on segmen ed 2D modi ied look-locke in e -
sion eco e y acquisi ion was success ully p oposed o
mice.22 Compa ed o a iable lip angle, IR p epa ed
sequences a e less sensi i e o B1inhomogenei y (p o-
nounced in UHF MRI), when using small- lip-angle exci-
a ion pulses and global adiaba ic in e sion pulses. As
his pape aims a UHF MRI ( ypical o small-animal
MRI), we ocus only on IR me hods wi h e ospec i e
na iga ion.
Wi h espec o he abo e-men ioned aspec s, we p o-
pose a me hodology o e ospec i e sel -ga ed UHF-MRI
T1 quan i ica ion o myoca dium in small animals wi h
IR-p epa ed acquisi ion. To he au ho s’ knowledge,
he e ha e been only wo pape s on his opic in
he small-animal a ea.4,6 Win e e al.4show e ospec-
i e sel -ga ing o IR-p epa ed wo-dimensional (2D)
golden-angle adial acquisi ion; howe e , i lacks he
possibili y o synch oniza ion a ound he magne iza ion
ze o-c ossing poin o he in e sion eco e y p ocess.
Fu he mo e, he me hod equi es p ospec i e synch o-
niza ion wi h ECG and espi a o y signals o s a he
in e sion cycle always in he same espi a o y and ca -
diac phase. Han Pei e al.6sugges a simila app oach,
howe e wi h acquisi ion o sepa a e na iga o echoes,
which unnecessa ily p olongs he acquisi ion. Compa ed
o Win e ’s4and ou app oaches, Han Pei’s acquisi ion6
is based on Ca esian k-space sampling, which does no
bene i om he lowe sensi i i y o mo ion o adial ead-
ou and omincohe en p ope ies(noise-likeappea ance
o unde sampling-induceda i ac s)o adialgolden-angle
k-space sampling, exploi ed in comp essed sensing.23
Ou app oach (p elimina y wo k published in
Re e ence 24) equi es no ex e nal ECG/ espi a o y sig-
nals. I elies only on e ospec i e sel -ga ing based on a
na iga o signal de i ed om sho ini ial segmen s o he
ee-induc ion-decay (FID) signals. Independence o he
ca diac and espi a o y ac i i ies om he pulse sequence
and hei pe iodici y in sho segmen s allow us o eliably
de e mine e ospec i e physiological ga ing h oughou
he in e sion eco e y in e al.
Weha eimplemen edanIR-p epa ed2Dgolden-angle
adial acquisi ion me hod (simila o Re e ence 4)and
a h ee-dimensional (3D) golden-angle s ack-o -s a s
me hod.25 Images a e econs uc ed using o al a ia-
ion spa io empo al econs uc ion (no conside ed in
Re e ence 4o 6). A complex-image-domain model o
15222594, 2024, 1, Downloaded om h ps://onlinelib a y.wiley.com/doi/10.1002/m m.29846 by Technical Uni e si y In B no, Wiley Online Lib a y on [28/11/2023]. See he Te ms and Condi ions (h ps://onlinelib a y.wiley.com/ e ms-and-condi ions) on Wiley Online Lib a y o ules o use; OA a icles a e go e ned by he applicable C ea i e Commons License
370 VITOUŠ e al.
Look-Locke IR is hen i ed o he econs uc ed images
o quan i y T1.
Ou mo i a ion o implemen ing a 3D acquisi ion
sequence in addi ion o a 2D e sion was as ollows. Con-
a y o he p oblem o non- ec angula slice p o ile in
2D acquisi ion, a 3D acquisi ion p o ides a cons an lip
anglein he inne slices o heslab.Also,inmo ing issue,
like hea , a p edic able s eady s a e o magne iza ion in a
spoiled g adien echo sequence is mo e easily achie able
wi h a 3D acquisi ion, as he whole ( hick) slab is exci ed
a once. These wo aspec s should, a leas in heo y, lead
o mo e accu a e modeling o he e ec s o he exci a ion
pulses in he T1 quan i ica ion model.
2METHODS
2.1 Acquisi ion and p ocessing me hod
The me hod implemen ed acquisi ion o IR-p epa ed
spoiled g adien echoes wi h adial k-space eadou using
golden-angle azimu h inc emen s. Fo he 3D case, posi-
ion encoding was based on a golden-angle s ack-o -s a s
scheme as desc ibed in Re e ence 25. The IR p epa a ion
was accomplished wi h a sequence o Nin (e.g., 60) nons-
elec i e in e sions epea ed wi h a pe iod TIR ≫T1 (e.g.,
10s),whichwe ein e lea edwi h ainso N ad (e.g.,1500)
low- lip-angle eadou pulses. Execu ed as slice-selec i e
o slab-selec i e o 2D o 3D encoding, espec i ely. The
choice o he small lip angle o he eadou RF pulses (3◦)
was based on he ollowing easoning. Fo low lip angles,
heunce ain yabou he eal lipangledue oB1inhomo-
genei y a ec s he T1 es ima es negligibly. Also, con a y
o high- lip-angle pulses, wi h low- lip-angle pulses he
s eady s a e is eached la e , which allows mo e accu a e
es ima ion o longe T1s.10 On he o he hand, low lip
angle means lowe SNR. The choice o 3◦was a comp o-
mise be ween he abo e-men ioned aspec s based on ou
p e ious expe imen s.
The pulse sequence is shown in Figu e 1and he
ga ing p ocess can be seen in Figu e 2.Thep o-
posed pulse sequences we e implemen ed on a B uke
BioSpec USR 94/30 (B uke BioSpin GmbH) 9.4T scan-
ne wi h he BGA12S-HP (660 mT/m) g adien sys em.
The pulse-sequence pa ame e s used o e alua ion on
bo h he phan om and a s (see below) a e summa ized
in Table 1. Fo each exci a ion, Nacq =168 samples we e
acqui ed (incl. he FID and he echo signal), o which he
i s sample was ex ac ed o na iga ion, he ollowing 39
samples (sampled du ing g adien p ewinding) we e dis-
ca ded and he emaining 128 samples (lying on he adial
k-space ajec o y) we e used o image econs uc ion.
Ass a edabo e, he awna iga o signalwasex ac ed
om he i s sample o he FID signals and p ocessed o
emo e he IR end as ollows. The same polynomial o
he8 h o de was i ed o all Nin acqui edIR–IR in e als
simul aneously, based on he assump ion ha he ca diac
and espi a o y mo ion in he indi idual IR–IR in e als
d
FIGURE 1 Two/ h ee dimensional (2D/3D) golden-angle (s ack-o -s a s) pulse-sequence iming diag am. Fo 3D, he phase-encoding
g adien in heslab-selec iondi ec ionisshowndashed(no usedin he2Dcase).Thena iga o isacqui eda hebeginningo he“Acq.”block
(whe e all g adien s a e o , ma ked by blue dashed lines). The ans e se magne iza ion is spoiled immedia ely a e each in e sion eco e y
pulse as well as be ween he g adien -echo eadou s. Bo h adio equency pulse phase al e na ion and g adien spoiling echniques a e used.
15222594, 2024, 1, Downloaded om h ps://onlinelib a y.wiley.com/doi/10.1002/m m.29846 by Technical Uni e si y In B no, Wiley Online Lib a y on [28/11/2023]. See he Te ms and Condi ions (h ps://onlinelib a y.wiley.com/ e ms-and-condi ions) on Wiley Online Lib a y o ules o use; OA a icles a e go e ned by he applicable C ea i e Commons License
VITOUŠ e al. 371
(s)
FIGURE 2 The pic u e shows he magne iza ion e olu ion and syn he ic hea and espi a o y na iga o (Fo an example o a eal
na iga o see Figu e S1) wi h depic ed ga ing in e als (g een) ha lie in he selec ed espi a o y (bo om ow) and ca diac (middle ow)
phases. Any selec ed p ojec ion mus lie wi hin hese ga ing in e als o be selec ed o econs uc ion. In e als ha sa is y his condi ion a e
shown again g een in he magne iza ion e olu ion cha (uppe ow). The p ojec ions om hese in e als a e hen binned based on hei
posi ion wi hin he in e sion eco e y (IR) cycle. Each IR bin (Dashed ed boxes, op) is hen used o econs uc ion o one image. Fo
illus a ion pu poses, he shown numbe o p ojec ions pe IR cycle is educed signi ican ly compa ed o eali y o imp o e he cla i y,
simila ly o he hea and espi a o y a es. Also he numbe o IR bins is educed o 8 shown as oppose o 20 in eal acquisi ion.
TABLE 1 Two- (2D) and h ee-dimensional (3D) T1 mapping pulse-sequence pa ame e s, ha we e used o all MRI expe imen s
(phan om, heal hy a , ib osis) desc ibed in his pape .
Pa ame e 𝝉 d TE FA Nacq N ad TIR Ma ix FOV
2D 7 ms 12 ms 7 ms 1.4 ms 3◦168 1500 10.5 s 128 ×128 40 ×40 ×2mm3
3D 8.2 ms 12 ms 8ms 1.2 ms 3◦168 1500 10.5 s 128 ×128 ×840 ×40 ×16mm3
Abb e ia ions: FA, lip angle; FOV, ield o iew; TE, echo ime.
is unco ela ed. The o de o he polynomial was selec ed
as a comp omise be ween i s lexibili y and complexi y,
based on ou p elimina y es ing o a ious polynomial
o de s. As i emo ed he IR end su icien ly, o he pos-
sibili ies such as i ing a sum o exponen ials we e no
es ed. The app op ia e coil elemen was hen selec ed by
choosing he elemen wi h he s onges na iga o signal.
Then, he na iga o signals o hea and espi a ion we e
sepa a ed by band-pass il e ing ( ia spec al dele ion) o
heIR- end-compensa edna iga o (sizeNin ×N ad).The
physiological espi a o y and hea a es we e aken in o
accoun (hea a e be ween 240 and 600 bea s pe minu e
and espi a ionin he angeo 18–180b ea hspe minu e).
Then he ancho poin s we e de ec ed in he na iga o
signals,andex apola ed oa eas,whe e hena iga o was
un eliable ( he i s 25% o he IR–IR in e al, co e ing
he i s pa o he IR end up o sho ly a e he mag-
ne iza ion ze o-c ossing poin ). The esul ing espi a o y
na iga o (con aining only he ancho poin s) was hen
con e ed oaga ingsignalbyselec ing55%o each espi-
a o y cycle co esponding o i s pla eau (expi a ion phase
wi h minimal animal mo ion). Fo he ca diac na iga o
(also now con aining only he ancho poin s), each ca -
diac cycle was spli in o subsequen ca diac phases. The
equi ed ca diac phase (in ou case dias ole) was hen
ex ac ed by selec ing 15% o he ca diac cycle’s samples
a ound he equi ed ca diac phase posi ion in he na iga-
o . The ela i e posi ions o he dias ole phase wi hin he
ca diac cycle we e ound om isual inspec ion o images
econs uc ed o a ious posi ions. The p oposed choice
15222594, 2024, 1, Downloaded om h ps://onlinelib a y.wiley.com/doi/10.1002/m m.29846 by Technical Uni e si y In B no, Wiley Online Lib a y on [28/11/2023]. See he Te ms and Condi ions (h ps://onlinelib a y.wiley.com/ e ms-and-condi ions) on Wiley Online Lib a y o ules o use; OA a icles a e go e ned by he applicable C ea i e Commons License
372 VITOUŠ e al.
o posi ionsandleng hso heselec edsegmen s ela i e o
he leng h o each indi idual espi a o y and ca diac cycle
(in pe cen age) makes he sel -ga ing me hod obus wi h
espec o changes in espi a o y and ca diac a es du ing
he acquisi ion. These ga ing signals we e hen combined
ogene a e he esul ingga ingsignal o he whole acqui-
si ion.I was henused o selec iono hep ojec ions ha
we e acqui ed in he app op ia e in e als, ha is, wi hin
hepla eau espi a o yphaseand hedesi edca diacphase
(Figu e 2). The selec ed p ojec ions we e inally binned
in oIRbinsbasedon hei posi ionwi hin heIR–IRin e -
al (Figu e 2). He e, 20 equally long IR bins we e used o
he whole IR–IR in e al.
Image econs uc ion was pe o med o line o bo h
he 2D and 3D acquisi ion using modi ied spa ially
and empo ally cons ained econs uc ion so wa e om
Re e ence 26. The spa io empo al egula iza ion was
o mula ed as o al a ia ion egula iza ion, whe e, he
empo aldimensionwas hein e sion ime,TI.Ingene al,
hedesc ibed e ospec i esel -ga ingmaylead odi e en
(and po en ially low) numbe s o p ojec ions in di e en
IR bins and, o he 3D e sion, also di e en numbe s o
p ojec ions in di e en slice phase-encoding s eps.
This caused poo pe o mance o he o iginal 2.5D
GROG g idding ope a o . Hence, i was eplaced by s an-
da d NUFFT om Re e ence 27 based on J. Fessle ’s
implemen a ion in Re e ence 28,whichled obe e
image econs uc ion. The spa ially and empo ally
cons ained econs uc ion algo i hm inco po a es also
pa allel imaging by p ocessing signals om all coil chan-
nels oge he wi h hei sensi i i ies, es ima ed p io o
he econs uc ion.
The econs uc ed images we e subsequen ly used o
es ima ion o T1 using he ollowing bin-a e aging model
(based on Re e ence 29):
SIRLL
bin (𝜌0,T1)
=1
NPB
n2
∑
n=n1
𝜌0sin(𝛼)[F+(E𝜏cos𝛼)n−1(Q−F)],(1a)
F=1−E𝜏
1−E𝜏cos𝛼,(1b)
Q=−Fcos𝛼E Ed[1−(E𝜏cos𝛼)N−1]−2Ed+E Ed+1
1+cos𝛼E Ed(E𝜏cos𝛼)N−1,
(1c)
E𝜏=exp(−𝜏
T1),(1d)
E =exp(−
T1 ),(1e)
Ed=exp(− d
T1 ).(1 )
I akes in o accoun he pe iod be ween he in e sion
pulse and he i s exci a ion, d, he ime be ween subse-
quen exci a ions,𝜏, he elaxa ionpe iod a he end o he
eadou echo ain, , and he lip angle o he exci a ion
pulses o he eadou ain, 𝛼. The numbe o p ojec ions
in each IR bin is deno ed as NPB and n1,2s and o indices
o he i s and las p ojec ions in each IR bin. The num-
be o all exci a ions pe in e sion pe iod is desc ibed by
N ad. The model assumes pe ec in e sion and was i ed
ocomplex oxel-basedin e sion eco e ycu es, esul ed
in es ima es o he complex ac o 𝜌0( ela ed o p o on
densi y) and he ue T1 elaxa ion ime.
2.2 E alua ion on phan om
measu emen s
The p oposed me hodology was i s es ed on a s a ic
phan om consis ing o eigh ials wi h wa e solu-
ions o Gado is ® (Baye AG) o di e en concen-
a ions (0.1–2.5 mmol/L) and hus in di e en T1
elaxa ion imes in each ial. To e alua e he p o-
posed me hods, he phan om was i s measu ed wi h a
g ound- u h acquisi ion— he s anda d IR me hod wi h
11 TIs (4–9000 ms) wi h one g adien -echo eadou pe
in e sion, TR/TE 10,000/4 ms (MR scanne speci ied
abo e). The T1 quan i ica ion om hese g ound- u h
da a was hen based on he ollowing model i ed o he
complex-image domain:
I=𝜌0[1−2exp(−TI∕T1)+exp(−TR∕T1)].(2)
Then, he same phan om was imaged wi h he p oposed
2D and 3D acquisi ion me hods, wi h he same pa am-
e e s as used la e o a s ( o pa ame e s see Table 1).
To accoun o ga ing in a acquisi ions, we andomly
selec ed 8.25% o he acqui ed p ojec ions, based on
he abo e men ioned segmen leng hs o he espi a o y
(55%) and ca diac (15%) ga ing (0.15 ×0.55 =0.0825). We
assumenoco ela ionbe ween heca diacand espi a o y
ac i i ies.
Then, we compa ed he T1 es ima es o he
g ound- u h me hod wi h he es ima es ob ained by he
p oposed 2D and 3D me hods wi h a long acquisi ion
du a ion, ha is, 18 and 45 min long, espec i ely (he e-
a e e e ed o as ull-leng h acquisi ions) o see i he
p oposed ull-leng h 2D and 3D me hods can be used
as a e e ence when e alua ing he e ec s o a sho e
acquisi ion and spa io empo al image egula iza ion on
a da a (whe e he s anda d g ound- u h IR acquisi-
ion is una ainable). To quan i y he e o , we measu ed
he mean T1 elaxa ion ime in each ial o he phan-
om using all me hods and hen calcula ed he ela i e
15222594, 2024, 1, Downloaded om h ps://onlinelib a y.wiley.com/doi/10.1002/m m.29846 by Technical Uni e si y In B no, Wiley Online Lib a y on [28/11/2023]. See he Te ms and Condi ions (h ps://onlinelib a y.wiley.com/ e ms-and-condi ions) on Wiley Online Lib a y o ules o use; OA a icles a e go e ned by he applicable C ea i e Commons License

VITOUŠ e al. 373
e o o he p oposed 2D o 3D me hod wi h espec o
he g ound- u h IR me hod, which is hen shown in
pe cen age.
2.3 E alua ion on a heal hy a
All measu ing p ocedu es we e pe o med unde EU
Di ec i e no. 2010/63/EU and app o ed by he Ani-
mal Ca e Commi ee o Czech Academy o Sciences,
Czech Republic, and Czech Go e nmen al Animal Ca e
Commi ee, in compliance wi h Czech Animal P o-
ec ion Ac No. 246/1992. Imaging was done using
a olume esona o as he ansmi e and a su ace
a -b ain ou -channel a ay coil as he ecei e . The a
was anes he ized using a mix u e o iso luo ane (2 %)
and oxygen gas (1000 mL/min) and lay p one on he
su ace coil.
The abo e desc ibed 2D and 3D T1-quan i ica ion
scans we e done in an oblique o ien a ion co esponding
o hesho -axis iew.Thesliceswe eposi ionedacco ding
o B uke ’s ecommenda ions based on se e al i e-slice
g adien echo scans wi h low compensa ion.
A e scanning, he da a we e p ocessed o line. The
p ojec ions we e i s selec ed using he abo e desc ibed
ga ingandbinningalgo i hms(dias olicca diacphasewas
chosen, as i is s anda d in clinical p ac ice and also used
ino he pape swecompa eou me hodwi h), ollowedby
image econs uc ion and T1 es ima ion.
To e alua e he e ec s o sho ening he acquisi ion
ime and he e ec s o a ious egula iza ion weigh s,
we ha e c opped he da ase s o se e al sho e da ase s.
As he s anda d g ound- u h IR acquisi ion used o he
phan om canno be applied o in i o ca diac imag-
ing o a s (because o ca diac and espi a o y mo ion),
pe o ming a ull-leng h 2D and 3D acquisi ions wi h
he p oposed me hods (18 and 45 min long, espec-
i ely) yielded a e e ence. Nex , he measu ed da a
we e c opped and econs uc ed wi h mul iple egu-
la iza ion weigh s and he bes egula iza ion weigh s
we e ound.
The T1 es ima ion e o was e alua ed simila ly as in
hephan omexpe imen ,he einside egionsde i ed om
manually d awn le - en icula myoca dium ou lines. In
case o he 3D me hod, six slices ou o eigh we e used
o e alua ion— he ou e slices o he slab we e omi ed.
The ela i e e o o he T1 es ima es was calcula ed o
each myoca dial oxel as absolu e alue o he di e ence
be weenT1es ima esob ainedwi h he ull-leng hand he
sho ened egula ized da ase , di ided by he ull-leng h
T1. Then, hei a e age was epo ed as he mean ela i e
e o .
2.4 E alua ion on he a model
o myoca dial- emodeling
The p oposed me hods we e also alida ed in he ol-
lowing animal expe imen s, ca ied ou acco ding o he
ecommenda ions o he Eu opean Communi y Guide o
he Ca e and Use o Labo a o y Animals and acco ding
o he expe imen al p o ocol (No. MSMT-35972/2020-3)
app o ed by he Commi ee o Ensu ing he Wel a e o
Labo a o y Animals, Masa yk Uni e si y and licensed by
he Minis y o Educa ion, You h and Spo s o he Czech
Republic.
The o al o nine Sp ague-Dawley male a s (6 weeks
old) we e included in he s udy. The animals we e an-
domly di ided in o wo expe imen al g oups: ib osis
(FIB; n=5) and con ols (CON; n=4, o iginally i e
animals – 1 animal died du ing he expe imen ). The ani-
mals we e housed in g oups in a empe a u e-, p essu e-,
and humidi y-con olled en i onmen , wi h ligh cycle
12/12 (ligh /da k), ad libi um access o wa e and s an-
da d die du ing he whole expe imen . The animals
we e allowed o adap o he en i onmen and e e y-day
manipula ion o minimally 7 days. The deoxyco icos-
e one ace a e (DOCA, Sigma-Ald ich)—sal model was
used o induce myoca dial emodeling in he g oup FIB,
as p e iously desc ibed.30 B ie ly, unila e al neph ec omy
was pe o med and o h ee ollowing weeks DOCA was
adminis e ed once a week om he day o su ge y in
he depo dose—20 mg/week, subcu aneously (s.c.); dis-
sol ed in 0.2 mL o peanu oil). Du ing he same pe iod,
sal in ake was inc eased by adding sodium chlo ide and
po assium chlo ide in o he d inking wa e (0.9% NaCl
and 0.3% KCl). In he g oup CON, sham ope a ion was
pe o med. Du ing he ollowing 3 weeks, a ehiculum
(peanu oil; Sigma-Ald ich) was adminis e ed s.c. once a
week (0.2 mL/week s.c.) and no sal was added o he
d inking wa e . All a s we e scanned using he same MR
scanne and he p o ocol desc ibed abo e. The animals
we e scanned a day be o e (baseline) and 14 days a e
he su ge y. A week a e he second MRI scan, each ani-
mal was sac i iced, he hea was apidly excised, washed
om he blood in a cold phospha e bu e ed saline (PBS;
Sigma-Ald ich) and ixed in o malin o 24 h. Then, he
hea was cu pe pendicula ly o he hea axis, ou inely
p ocessed and comple ely embedded in o o malin- ixed
pa a in-embedded specimens. His opa hological exami-
na ion was pe o med by hema oxylin-eosin s aining and
Goldne g een special s aining. The ca diac emodeling
was assessed in acco dance wi h he ecommenda ions
o he Eu opean Socie y o Ca diology.31 The ex en o
ca diac ib osis was quan i ied by expe ’s manual anno-
a ion o he whole slice images in he QuPa h so wa e32
15222594, 2024, 1, Downloaded om h ps://onlinelib a y.wiley.com/doi/10.1002/m m.29846 by Technical Uni e si y In B no, Wiley Online Lib a y on [28/11/2023]. See he Te ms and Condi ions (h ps://onlinelib a y.wiley.com/ e ms-and-condi ions) on Wiley Online Lib a y o ules o use; OA a icles a e go e ned by he applicable C ea i e Commons License
374 VITOUŠ e al.
and epo ed as a io o ib o ic issue a ea o he o al
sample a ea.
The MRI p o ocol was an ex ended e sion o he
one desc ibed in Sec ion 2.3. The p oposed 2D and
3D acquisi ion was ca ied ou wice—be o e and a e
adminis a ion o a con as agen (CA), o subsequen
myoca dial ex acellula olume (ECV – ela i e olume
o he in a ascula plus in e s i ial space) es ima ion,
see below. In addi ion, o acqui e addi ional independen
in o ma ion abou he myoca dial issue s a e, ana omi-
cal scans we e done be o e he T1 quan i ica ion scans
by u ilizing endo ’s In aga e FLASH sequence (a e o-
spec i ely sel -ga ed Ca esian FLASH sequence) o assess
ana omical as well as physiological p ope ies o he le
chambe (myoca dial ac ional olume (MFV) and ejec-
ion ac ion (EF)).
TheMFVwascompu edasa ac iono hemyoca dial
mass in he en i e le chambe measu ed in dias ole. The
EF was compu ed as a a io o ejec ed blood olume (di -
e ence o end-dias olic and end-sys olic olume), and he
end-dias olic olume. Endo- and epi-ca dial ou lines we e
d awn manually in each slice o each animal and mea-
su emen . Only le chambe pa ame e s we e assessed.
A e he ana omical and p econ as T1-mapping
scans (2D and 3D), a CA (Gado is , Baye GmbH) was
adminis e ed in a enously in o he ail ein as a bolus
using a linea in usion pump (Ha a d Appa a us), injec-
ion speed 1 mL/min, dose 0.2 mmol/kg weigh . The
pos con as T1-mappingscans(2Dand3D)we e henpe -
o med s a ing 5 minu es a e he con as -agen admin-
is a ion. E alua ion o emodeling was based on ECV
de i ed om he es ima ed p econ as and pos con as
T1 maps. The images o T1 mapping we e econs uc ed
using he op imal egula iza ion weigh s, ound in he
expe imen s on a heal hy a desc ibed abo e, adjus ed o
heacquisi ionleng husedhe e(10min o 2Dand15min
o 3D). A delay a e he con as -agen adminis a ion
is needed o each he s eady s a e o equal in a- and
ex a- ascula con as -agen concen a ion (he e, he 2D
T1-quan i ica ion acquisi ion s a ed 5 min and he 3D
me hod15mina e heCAadminis a ion).Basedonp e-
ious wo k and clinical p ac ice,19,33 he ECV is expec ed
oinc easewi h he emodelledca diac issue,as hein e -
s i ial space is expanded by ex acellula ma ix allowing
i o accommoda e mo e CA han he heal hy issue. The
ECV was calcula ed as33:
ECV =(1−HCT)R1 issue pos −R1 issue p e
R1blood pos −R1blood p e ,(3)
whe e R1 s ands o elaxa ion a es (1/T1) o he issue
and blood be o e and a e he CA adminis a ion, HCT
is hema oc i . Hema oc i was measu ed o each animal
a each MRI ime poin by aking blood samples om he
ail ein.
3RESULTS
3.1 Phan om measu emen s
The esul s o he phan om expe imen a e summa ized
in Table 2. The ag eemen wi h he e e ence me hod was
wi hin 9 % in he whole in e al o he es ed T1 alues o
he 2D and 11 % o he 3D me hod. Wi h his le el o con-
idence, we ha e used he ull-leng h 2D and 3D da ase s
as a e e ence o he in i o expe imen s.
3.2 Resul s om a heal hy a
The mean ela i e e o s (uppe ows in Figu es 3and 4)
show hee ec s o egula iza ionweigh s o di e en le -
els o sho ening he acquisi ion (columns). The posi ions
o he op imal egula iza ion weigh s ( iangles) show
ha a sho e acquisi ion (highe le el o unde sampling)
equi es highe egula iza ion weigh s.
The T1-es ima ion e o s in Figu es 3and 4sugges
he acquisi ion leng hs equi ed o a gi en choice o
he accep able e o . Fo example, i we se he maxi-
mum accep able T1-es ima ion e o wi h espec o he
ull-leng h acquisi ion o 5%, we can sho en he acqui-
si ion o app oxima ely 5 min o he 2D acquisi ion
and o somewhe e be ween 7 and 9 min o he 3D
(eigh slices) acquisi ion as can be seen in Figu e 5.Fo
TABLE 2 Phan om, T1 es ima es in eigh ials.
Me hod ROI 1 ROI 2 ROI 3 ROI 4 ROI 5 ROI 6 ROI 7 ROI 8
G ound u h 2D IR (ms) 74.5 97.3 148 175 316 668 808 2503
Full-leng h 2D IR (ms) 72.7 94.3 156 188 334 726 873 2465
Rela i e e o 2D IR (%) −2.5 −3.0 5.3 7.2 5.7 8.8 8.1 −1.5
Full-leng h 3D IR (ms) 75 95 144 191 331 741 852 2425
Rela i e e o 3D IR (%) 1.0 −2.5 −2.4 9.3 4.7 11.0 5.4 −3.1
15222594, 2024, 1, Downloaded om h ps://onlinelib a y.wiley.com/doi/10.1002/m m.29846 by Technical Uni e si y In B no, Wiley Online Lib a y on [28/11/2023]. See he Te ms and Condi ions (h ps://onlinelib a y.wiley.com/ e ms-and-condi ions) on Wiley Online Lib a y o ules o use; OA a icles a e go e ned by he applicable C ea i e Commons License
VITOUŠ e al. 375
λ
λ
FIGURE 3 Heal hy a , wo-dimensional acquisi ion. Top: Mean ela i e e o s (%) o di e en spa ial and empo al egula iza ion
weigh s (𝜆spa ial,𝜆 empo al) o ou di e en acquisi ion du a ions (columns): 18, 10, 4, 2 min. The minima a e ma ked wi h a iangle. Bo om:
T1 maps o he op imal egula iza ion weigh s.
FIGURE 4 Heal hy a , h ee-dimensional acquisi ion. Top: Mean ela i e e o s o di e en spa ial and empo al egula iza ion
weigh s (𝜆spa ial,𝜆 empo al) o ou di e en acquisi ion du a ions (columns): 45, 20, 10, 5 min. The minima a e ma ked wi h a iangle.
Bo om: T1 maps o he op imal egula iza ion weigh s.
he myoca dial- emodeling expe imen below, acquisi ion
leng hs o 10 ( o 2D) and 15 ( o 3D) min we e chosen, o
s ay sa ely unde he 5 % e o h eshold.
3.3 Resul s om he a model
o myoca dial- emodeling
Thee ec o myoca dial emodelingwase alua edby i s
compa ing he CON and FIB g oups a he baseline MRI
(day 0, no signi ican di e ence expec ed) and hen a he
second MRI (day 14, signi ican di e ence expec ed). The
unpai edle - ailT- es wasused o allexaminedpa ame-
e s EF, MFV and ECV. The esul s can be seen in Table 3.
Acco ding o he expec a ions, o he EF and MFV,
he e was no signi ican di e ence be ween he con ol
and ea ed g oups a baseline MRI (EF: p=0.50, MFV:
p=0.89) and he e was a signi ican di e ence be ween
he g oups a he second MRI ime poin (EF: p<0.05,
MFV: p<0.05).
15222594, 2024, 1, Downloaded om h ps://onlinelib a y.wiley.com/doi/10.1002/m m.29846 by Technical Uni e si y In B no, Wiley Online Lib a y on [28/11/2023]. See he Te ms and Condi ions (h ps://onlinelib a y.wiley.com/ e ms-and-condi ions) on Wiley Online Lib a y o ules o use; OA a icles a e go e ned by he applicable C ea i e Commons License
376 VITOUŠ e al.
FIGURE 5 Mean ela i e e o s o he es ed acquisi ion
leng hs and hei co esponding op imal egula iza ion weigh s.
The chosen 5% e o limi is shown in g ay colo .
TABLE 3 Table summa izing he esul ing ejec ion ac ion
(EF), myoca dial ac ional olume (MVF) and ex acellula
olume (ECV).
Exam MRI 1 MRI 2
G oup CON FIB CON FIB
EF 80 ±579±481±288±2
MVF 61 ±260 ±160 ±367 ±2
ECV 2D, 10 min 26 ±524±423±528±7
ECV 2D, 5 min 27 ±526 ±522 ±527 ±4
ECV 3D, 15 min 27 ±625±624±627±6
ECV 3D, 7.5 min 30 ±726 ±731 ±728 ±6
No e: All p esen ed alues shown in pe cen age, mean ±SD.
The ECV was e alua ed o a 10-min 2D and a
15-min 3D T1-quan i ica ion acquisi ion. In line wi h he
expec a ions, he e was no signi ican di e ence be ween
he con ol and ea ed g oups a baseline MRI (2D: p=
0.99, 3D: p=0.99) and he e was a signi ican di e ence
be ween he g oups a he second MRI ime poin (2D:
p<0.05,3D:p<0.05).ExampleECVmapscanbeseenin
Figu e 6.
Secondly he same assessmen was made o sho e
acquisi ions, whe e bo h he 2D and 3D we e sho ened
o one hal (5 and 7.5 min, espec i ely). The e was no
signi ican di e ence be ween he con ol and he ea ed
g oups a baseline MRI (2D: p=0.58, 3D: p=0.99), bu
he ewasasigni ican di e encebe ween heg oupsa he
second MRI ime poin only o he 2D (p<0.05), no o
3D (p=0.75) case.
His ological examina ion e ealed eg essi e changes
in he myoca dium a a ious s ages, including oci o
nec osis and ma u ing sca issue (classi ied as eplace-
men ,pe i ascula o in e s i ial ib osis31).Thesechanges
we e only obse ed in he FIB g oup. The spa ial dis ibu-
ion o emodeling lesions is a he nonuni o m and can
be seen on example in Figu e 7.
4DISCUSSION
The phan om expe imen clea ly e ealed ha he
T1-quan i ica ion e o was lowe in 2D han in 3D expe -
imen s. This co esponds o a lowe numbe o adials pe
econs uc ion o one slice in he 3D case. In gene al no
sys ema ic end o T1 de ia ion was obse ed. I migh ,
howe e , appea when quan i ying longe T1 compo-
nen s, which would be unde es ima ed, as was shown o
example in Re e ence 29. On he o he hand such long T1
componen sa eunlikely oappea inca diacapplica ions.
The expe imen s on he eco dings om a heal hy
a showed he e ec o acquisi ion leng h and egula -
iza ion on he accu acy o he T1 es ima es (Figu es 3
and 4). As expec ed, highe egula iza ion weigh s we e
needed o sho e acquisi ion leng h, ha is, mo e
unde sampled da a.
Ou p econ as T1es ima eso myoca diumincon ol
a s (1550 ±120 ms) we e in line wi h some li e a u e al-
ues o MR scanne s wi h he same B0: 1534 ±151 ms o
a sel -ga ed IR algo i hm desc ibed abo e.6Howe e , s ill
o hesameB
0 he T1 es ima es a e known o depend on
he acquisi ion scheme and me hod. The hea a e o he
animals in he FIB g oup du ing exam 2 was o e all lowe
han ha o he CON g oup. This migh be an addi ional
ac o a ec ing myoca dial T1, as co ela ion o T1 and
hea a e (al hough only in he ange o se e al pe cen )
has been epo ed in mice p e iously.22
Ou measu ed ECV was o e all sligh ly la ge ( o he
secondMRIexamina ion:23 ±4%in CONg oupand 28 ±
6%in FIB g oup, Figu e 6) han o example in6(18 ±
2.1%in heal hy a s and 22.43 ±2.51%in FIB g oup).
This may be possibly a ibu ed o he young age o ou
a s (9 weeks in ou case s. 14 weeks in Re e ence 6).
This hypo hesis is suppo ed by he esul s o ECV in he
baseline expe imen s, (done a he age o 7-weeks), whe e
we ob ained e en la ge ECVs (25 ±4%). The dec ease
o ECV in he CON g oup wi h inc easing age can also
be obse ed in Table 3. This age- ela ed dec ease o ECV
migh ac agains he emodeling- ela ed ECV inc ease in
he FIB g oup.
15222594, 2024, 1, Downloaded om h ps://onlinelib a y.wiley.com/doi/10.1002/m m.29846 by Technical Uni e si y In B no, Wiley Online Lib a y on [28/11/2023]. See he Te ms and Condi ions (h ps://onlinelib a y.wiley.com/ e ms-and-condi ions) on Wiley Online Lib a y o ules o use; OA a icles a e go e ned by he applicable C ea i e Commons License