Report on the translation of in vitro safety testing results to realistic exposure conditions considering combined GC and RF induced heating

Con iden iali y S a us: PU - Public, ully open ( emembe o deposi public deli e ables in a us ed eposi o y)
Deli e able Co e Shee
Funded by he Eu opean Union. Views and opinions exp essed a e
howe e hose o he au ho (s) only and do no necessa ily e lec hose
o he Eu opean Union o EURAMET. Nei he he Eu opean Union no
he g an ing au ho i y can be held esponsible o hem.
The p ojec has ecei ed unding om he Eu opean Pa ne ship on
Me ology, co- inanced om he Eu opean Union’s Ho izon Eu ope
Resea ch and Inno a ion P og amme and by he Pa icipa ing S a es.
1 o 46
21NRM05 STASIS
D5: Repo on he ansla ion o in i o sa e y es ing esul s o ealis ic
exposu e condi ions conside ing combined GC and RF induced hea ing
O ganisa ion name o he lead pa icipan o he deli e able:
Fo schungss i ung ü In o ma ions echnologie und Gesellscha (IT’IS)
Due da e o he deli e able: 30 Sep embe 2025
Ac ual submission da e o he deli e able: 30 Sep embe 2025
2 o 46
Glossa y
MRI Magne ic Resonance Imaging
MR Magne ic Resonance
RF Radio equency
SAR Speci ic Abso p ion Ra e
DBS Deep B ain S imula o
IPG Implan able Pulse Gene a o
AIMD Ac i e Implan able Medical De ice
CP Ci cula Pola iza ion
GC G adien Coil
RMS Roo Mean Squa e
TSM Tissue Simula ing Medium
TABLE OF CONTENTS
21NRM05 STASIS.............................................................................................................................1
1 Summa y.....................................................................................................................................3
2 Mo i a ion....................................................................................................................................4
3 Ac i i y A3.3.1.............................................................................................................................5
4 Ac i i y A3.3.2...........................................................................................................................10
5 Ac i i y A3.3.3...........................................................................................................................13
5.1 Passi e Implan s.................................................................................................................14
5.2 Ac i e Implan .....................................................................................................................15
6 Ac i i y A3.3.4...........................................................................................................................17
7 Ac i i y A3.3.5...........................................................................................................................18
8 Ac i i y A3.3.6...........................................................................................................................19
9 Summa y o A3.3......................................................................................................................28
10 Ac i i y A3.4.1...........................................................................................................................29
11 Ac i i y A3.4.2...........................................................................................................................32
11.1 Valida ion and Con idence In e al o RF-Hea ing.......................................................... 32
11.2 Valida ion and Con idence In e al o GC Hea ing..........................................................38
12 Ac i i ies A3.4.3 and A3.4.4, and Conclusions.........................................................................44
3 o 46
1 Summa y
In his deli e able, g adien -induced hea ing concep s and p ocedu es p oposed in ISO/TS
10974 a e ex ended o passi e o hopedic implan s and ixa ion de ices. A Tie 2 app oach
is de eloped o educe he o e es ima ion o he sa e y limi s in case o aniso opic implan s,
and o de ine es ing p ocedu es mo e ele an o he ac ual clinical condi ions a which such
implan s a e scanned. The oles o g adien and RF as implan hea ing sou ces a e
in es iga ed in silico; ecommenda ions o combining esul s o RF and g adien es ing a e
p o ided in iew o s anda dized p ocedu es. The esul is a obus me hodology o
s eamline he in i o e alua ion o bulky passi e implan s and ansla e in o ac ual clinical
MRI exposu es wi h sui able con idence le els, which is p oposed in Deli e able 7 “S anda d
Tes Me hod o Measu emen o G adien Magne ic Field Induced Hea ing On o Nea Non-
ac i e Implan s Du ing Magne ic Resonance Imaging” o he STASIS p ojec as a d a
ASTM es s anda d.
Key ou comes o he wo k desc ibed in his deli e able epo include:
demons a ion o signi ican supe posi ion o RF and g adien hea ing o bulky passi e
implan s unde clinically ele an condi ions
o65% in phan om, 10% in i o wo s condi ions and 0.35K in ealis ic condi ion
 e ined RF E- ield me ics de eloped wi h lowe o e es ima ion han Tie 2
 alida ed ep esen a i e con idence in e al o in silico RF and g adien hea ing
assessmen s
These ou comes a e summa ized in his epo , and mos a e also published in he ollowing
open-access pee - e iewed scien i ic jou nal a icles:
Zano ello Umbe o, Fuss Ca ina, A duino Alessand o, Bo auscio O iano. E icien
p edic ion o MRI g adien -induced hea ing o guiding sa e y es ing o conduc i e
implan s. Magne ic Resonance in Medicine 2023, doi:10.1002/m m.29787
Zano ello Umbe o, A duino Alessand o, Fuss Ca ina, Go en Tolga, , Bo auscio
O iano, Impac o simul aneous exposu e o RF and g adien elec omagne ic ields
on implan MR sa e y labeling. Magne ic Resonance in Medicine 2025,
doi:10.1002/m m.70059
A duino Alessand o, Bo auscio O iano, G appein Denise, Scialó S e ano, Vicini
Fabio, Zano ello Umbe o, Zilbe i Luca, 3D–1D modelling o c anial mesh hea ing
induced by low o medium equency magne ic ields, Compu e Me hods and
P og ams in Biomedicine 2025, doi: 10.1016/j.cmpb.2025.109009
4 o 46
1A duino Alessand o, Zano ello Umbe o, Hand Je , e al. Hea ing o hip join implan s in MRI: The combined e ec o RF and swi ched-g adien ields.
Magne ic Resonance in Medicine. 2021;85(6):3447
2Clemen i Vale ia, Zano ello Umbe o, A duino Alessand o, e al. Classi ica ion Scheme o Hea ing Risk du ing MRI Scans on Pa ien s wi h O hopaedic
P os heses, Diagnos ics. 2022;12(8):1873.
2 Mo i a ion
The hea ing o implan s du ing an MR examina ion due o he adio equency (RF) ield and
he g adien ield a e po en ially addi i e12, ye he wo phenomena a e ypically assessed
independen ly; hei supe posi ion is a ely conside ed in sa e y assessmen s, as i is no
di ec ly equi ed in any s anda d o egula o y guidance.
The e a e se e al easons o his:
1. RF ields deposi ene gy di ec ly in he biological issues su ounding he implan ,
whe eas g adien -induced ields deposi ene gy in he me allic implan . The spa ial
dis ibu ions o he esul ing empe a u e inc eases, as well as he ho spo loca ions
a ound a gi en implan , a e signi ican ly di e en .
2. GC hea ing is usually s onge when he implan is ela i ely a om he isocen e ,
whe eas RF hea ing is o en s onge when he implan is close o he isocen e o o
he end- ing o he RF coil;
3. MR sequences which a e he wo s -case o GC-hea ing a e also no ypically he
wo s -case o RF-hea ing, and ice e sa;
4. The es equipmen o bench op hea ing assessmen , as well as he selec ed MR
p o ocols o hea ing assessmen s in clinical scanne s and he s anda ds speci ying
he es p o ocols, a e di e en and independen ly de eloped, encou aging hei
independen ea men . These me hods a e necessa ily summa ized down o hea ing
a es, wo s -case empe a u es, o he mal doses o local ho spo s, o implan s unde
un ela ed es exposu e scena ios. The e o e, no in o ma ion abou he 3D
spa ial/ empo al dis ibu ions o hea is ypically a ailable;
5. The MR sequence pa ame e s which can limi RF-hea ing (SAR, B1+ ms), GC-
hea ing (dB/d ms), o bo h (scan du a ion), a e independen ly speci ied, hus RF-
GC hea ing supe posi ion can be mi iga ed by adjus ing any o h ee pa ame e s,
c ea ing di icul ies o de eloping guidance.
These easons can be subdi ided in o isk-based conside a ions (i ems 1–3) and p ac ical
conside a ions (4–5). The inc easing use o high- ideli y nume ical modeling o pa ien
exposu es using human body models and implan su oga es, in lieu o o in supplemen o
phan om-based physical es ing, is ele an o se e al o hese poin s. The mos signi ican
p ac ical conside a ion, i em 4, is su moun ed by access o he ull 3D spa ial dis ibu ion o
local ho spo s g an ed by nume ical me hods, inc easing he easibili y o e icien ly
assessing combined RF-GC hea ing. Nume ical me hods now allow po en ial sa e y isks o
be eliably iden i ied ahead o ime h ough mul i-physics in i o simula ions, a he han
wai ing o a clinical epo o an inju y di ec ly a ibu ed o he phenomenon in ques ion.
Unde hese ci cums ances, he ques ion is whe he he (now easible) e o o conside ing
combined RF-GC hea ing is jus i ied by eal po en ial isks, i.e., conside ing poin s 1–3, do
ealis ic MR examina ions p oduce simul aneous RF and GC hea ing which supe poses
5 o 46
su icien ly a he wo s -case loca ions o jus i y hei conside a ion in implan sa e y labeling?
The i s objec i e o his epo is o iden i y gene al condi ions unde which he hea ing o
GC and RF can be conside ed as independen o in e ac ing, causing highes -hea ing
scena ios. The en i e analysis is pe o med in silico, simula ing he exposu e o ana omical
human models modi ied by inse ing he implan h ough i ual su ge y. Fo he si ua ions
whe e supe posi ion is ound o be signi ican , ecommenda ions o combining esul s o
RF and GC es ing we e p o ided in iew o s anda dized p ocedu es.
The second goal o his epo is o de ine a obus me hodology o s eamline he in i o
e alua ion o bulky passi e implan s, o ansla e in i o analysis in o clinical MRI exposu e
and o p o ide sui able con idence le els. The analysis will include de elopmen o e e ence
app oaches o RF hea ing es ima ion o bulky passi e implan , b idging he gap be ween
ISO 10974 Tie 2 and Tie 3- ype app oaches o RF hea ing es ima ion; analysis o he
p esence o pe usion in RF and GC hea ing; possibili y o de ine op imal phan om he mal
p ope ies o GC hea ing o ma ch in i o esul s; possibili y o combine es s pe o med on
single implan componen s o ge he inal beha io o he en i e implan o GC hea ing.
The esul s ep esen a jus i ica ion o s eamline he in i o e alua ion o bulky passi e
implan s and ansla e in o ac ual clinical MRI exposu es wi h sui able con idence le els, as
well as he alida ed ools and me ics equi ed o do so; his app oach is p oposed in
Deli e able 7 “S anda d Tes Me hod o Measu emen o G adien Magne ic Field Induced
Hea ing On o Nea Non-ac i e Implan s Du ing Magne ic Resonance Imaging” submi ed
o ASTM F04.15 as a s anda d d a i em.
3 Ac i i y A3.3.1
INRIM, ITIS and MRC will de ine a se o ealis ic exposu e scena ios o pe o ming he
analysis o supe posi ion o RF and GC hea ing. The conside ed scena ios will include
di e en passi e and ac i e implan s (a leas wo bulky o hopaedic implan s, wo
o hopaedic pla es, one ac i e implan ), a selec ion o MRI sequences (a leas wo classes),
wo ha dwa e con igu a ions (RF coil and GC), implan posi ions wi hin he scanne (a leas
six posi ions).
The ollowing implan s we e used:
Hip implan : Implan manu ac u ed by Adle O ho SpA, I aly (model APTA-FIX +
FIXA Ti-Po ) wi h CoC Mo sphe e, Ti+HA s em, Ti6Al4V cup and UHMW-PE line
(Figu e 1a).
Knee implan : Implan manu ac u ed by Adle O ho SpA, I aly (model GENUS MB)
wi h CoC Mo emo al and ibial componen s and UHMW-PE line (Figu e 1b).
Shoulde implan : Implan manu ac u ed by LimaCo po a e SpA, I aly (model
TRAUMA SMR Ana omic) made in Ti6Al4V alloy (Figu e 1c).
C anial pla e: Implan manu ac u ed by Meda is AG, Swi ze land (model M2-7093S)
made in i anium (ASTM F67), semi- igid. The pla e is 100mm × 100mm and 0.6mm
hick (Figu e 1d).
Ankle pla e: Implan manu ac u ed by Meda is AG, Swi ze land (model 4954 25S
(le )) made in i anium (ASTM F136), alloy. Toge he wi h he ankle pla e, wo

6 o 46
i anium (ASTM F136) sc ews a e conside ed (models 5901 50 and 5800 60) (Figu e
1e).
Ac i e implan : The SAIMD-U implan manu ac u ed by ITIS is conside ed. SAIMD-
U is a e i ica ion and alida ion de ice o e alua ions acco ding o ISO/TS 10974
(Figu e 2).
The elec ic and he mal p ope ies o he ma e ial composing he passi e abo e implan s
a e lis ed in Table 1; he ma e ials o he SAIMD-U a e gi en in Annex U o ha s anda d.
Table 1: Ma e ial elec ic and he mal p ope ies
(a) Adle O ho SpA hip (b) Adle O ho SpA knee (c) LimaCo po a e SpA shoulde
Implan CAD implan CAD implan CAD
Ma e ial
El. Cond.
(MSm−1)
El. Rel. Pe m.
Th. Cond.
(Wm−1 K−1)
Sp. Hea
Cap.
(Jkg−1 K−1)
Mass
Dens.
kg/m3
Ti anium
0.58
-
7.2
520
4420
CoC Mo
1.16
-
14
450
8445
UHMW-PE
0
2.3
0.41
1840
930
7 o 46
(d) Meda is AG c anial pla e (e) Meda is AG ankle pla e
CAD CAD
Figu e 1: CAD models o he implan s
Figu e 2: ZMT SAIMD-U AIMD e i ica ion de ice
The ollowing MRI sequences we e conside ed along he nex p ojec ac i i ies:
Echo-plana imaging (EPI): Selec ed o s ess he hea ing e ec s due o swi ched
g adien ields (Figu e 3)
Balanced s eady-s a e (T ueFISP): Selec ed o s ess bo h he hea ing due o
adio equency and o swi ched g adien ields (Figu e 4)
Tu bo Spin Echo (TSE): Selec ed o s ess he hea ing due o adio equency ields
(Figu e 5)
8 o 46
As equi ed by he p o ocol, six imaging posi ions we e analyzed:
Femu
Knee
Pel is
Abdomen
Tho ax
Head
Fo each imaging posi ion mul iple implan posi ions we e simula ed acco ding o a ixed
7cm spa ial s ep along he z-axis (see Figu e 6).
Finally, a h ee-axial ac i ely shielded g adien coil wi h whole-body access manu ac u ed
by Nanjing Cichen Medical Technology Co., L d, Nanjing, China (model: Sola is-R) we e
used o he analyses o swi ched g adien ield hea ing. 16-leg high-pass bi dcage body
coils we e used o in es iga e he e ec s o adio equency hea ing bo h a 1.5T and 3T.
Figu e 3: Echo-plana imaging (EPI)
9 o 46
Figu e 4: Balanced s eady-s a e (T ueFISP)
Figu e 5: Tu bo Spin Echo (TSE)
16 o 46
Table 3: Ta ge deposi ed powe s a SAIMD-U lead ip (in W).
Posi ion /
F equency
64 MHz
128 MHz
30
0.003413
0.006014
100
0.007563
0.009639
170
0.011134
0.012712
240
0.01513
0.023597
310
0.016356
0.032438
380
0.01643
0.03659
450
0.015529
0.030868
520
0.012891
0.019755
590
0.008001
0.009391
660
0.003252
0.003414
730
0.000808
0.000712
800
0.000237
3.78E-05
Posi ion /
F equency
64 MHz
128 MHz
30
0.003413
0.006014
100
0.007563
0.009639
170
0.011134
0.012712
240
0.01513
0.023597
310
0.016356
0.032438
380
0.01643
0.03659
450
0.015529
0.030868
520
0.012891
0.019755
590
0.008001
0.009391
660
0.003252
0.003414
730
0.000808
0.000712
800
0.000237
3.78E-05
Figu e 11: Ta ge deposi ed powe s a SAIMD-U lead ip a 64 MHz.

17 o 46
3A duino Alessand o, Bo auscio O iano, B ühl Rüdige , Chiampi Ma io, Zilbe i Luca. In silico
e alua ion o he he mal s ess induced by MRI swi ched g adien ields in pa ien s wi h me allic
hip implan . Physics in Medicine
& Biology, 2019, doi: 10.1088/1361-6560/ab5428
Figu e 12: Ta ge deposi ed powe s a SAIMD-U lead ip a 128 MHz.
The dis ibu ion a ound he SAIMD-U lead ip has been ex ac ed on a g id o 2mm in all
di ec ions and placed in a box wi h a g id co esponding o he g id used o he esul s o
he passi e implan s so ha he cen e o he dis ibu ion lies a he posi ion o he lead ip
o a gene ic pacemake ou ing. The dis ibu ion has been scaled so ha he deposi ed
powe ex ac ed wi hin he -30dB con ou co esponds o he a ge powe s shown in he
abo e able.
Addi ionally, simula ions wi h he SAIMD-U IPG only ha e been done in he same way as
desc ibed abo e o he passi e implan s.
6 Ac i i y A3.3.4
INRIM will pe o m in silico GC simula ions o he exposu e scena ios iden i ied in A3.3.1,
using he digi al models de eloped in A3.3.2.
All GC simula ions ha e been pe o med complying wi h he posi ions, implan s and
sequences selec ed in ac i i y A3.3.1. Each simula ions has been epea ed wice o compa e
he e ec o he eac ion (skin e ec ) on he powe deposi ion wi hin each implan . Due o
he negligible ole played by issues in GC hea ing, all he simula ions ha e been pe o med
in ai in a spa ially homogeneous magne ic ield co esponding o ha gene a ed by he GCs
in he ele an implan posi ion.
G adien simula ions we e pe o med in Sim4Li e and wi h a home-made alida ed code3
when simula ions in ol ed he ASTM phan om and ana omical body models, espec i ely.
A 2mm oxel esolu ion was seen o be su icien o achie e nume ically s able esul s, and
18 o 46
4Bo auscio O iano, Chiampi Ma io, Hand Je , Zilbe i Luca. A GPU Compu a ional Code o
Eddy-Cu en P oblems in Voxel-Based Ana omy. IEEE T ansac ions on Magne ics, 2015, doi:
10.1109/TMAG.2014.2363140
he simula ion domain was es ic ed o he egion o he me allic implan s since he powe
deposi ed by he GCs in he phan om and body issues is negligible.
In simula ions in ol ing he ASTM phan om, he implan s we e exposed o a ha monic,
spa ially homogeneous, magne ic ield di ec ed along he di ec ion which maximizes he
induced hea ing. An algo i hm implemen ed in Sim4Li e de e mined he wo s exposu e
di ec ion, and simula ions we e pe o med a 1750 Hz as sugges ed by he ISO 10974 d a
IS1 s anda d. These simula ions exposed he implan s o an a bi a y powe , as he
empe a u e inc eases we e scaled o a e e ence peak alue a a la e s age.
Simula ions in ol ing MR pulse sequences and ealis ic body models ollowed he s uc u e
discussed in A duino e al.,3and complied wi h he implemen a ion desc ibed by Bo auscio
e al.4As a consequence o he equency con en o he g adien wa e o m ha monic
spec um in he conside ed MR sequences, elec omagne ic (EM) simula ions modeled he
skin e ec in o de o achie e mo e eliable esul s.
7 Ac i i y A3.3.5
S a ing om he da a collec ed in A3.3.3 and A3.3.4, INRIM will pe o m he mal simula ions
ei he compa ing he esul s ob ained wi h RF and GC ac ing sepa a ely o supe posing hei
con ibu ions.
The mal simula ions ha e been pe o med in h ee di e en scena ios:
1. Only powe om RF
2. Only powe om GC
3. Bo h RF and GC powe s
Gi en a speci ic sequence, he RF powe ha e been scaled o he a e age powe
associa ed o he equi ed o ob ain he lip angle αineeded by he sequence. The
e e s o he spa ial a e age wi hin he slice a he isocen e o he coils and i is ob ained
as:
whe e γis he gy omagne ic a io and τiis he i- h pulse du a ion. The RF powe PRF
associa ed o he conside ed scena io is he e o e ob ained as:
PRF
whe e P1µTis he powe associa ed o a o 1T a e aged on he isocen e slice and TR
is he epe i ion ime. Since he he mal egula ion has no been accoun ed o in
19 o 46
5A duino Alessand o, Bo auscio O iano, Chiampi Ma io, Zilbe i Luca. Douglas–Gunn Me hod
Applied o Dosime ic Assessmen in Magne ic Resonance Imaging. IEEE T ansac ions on
Magne ics. 2017, doi: 10.1109/TMAG.2017.2658021
simula ions, he RF empe a u e dis ibu ions associa ed o he same implan posi ion and
di e en sequences simply scales wi h he powe alue PRF.
The powe densi y dis ibu ions esul ing om GC and RF simula ions we e ex ac ed on a
egula 2mm g id. Du ing GC exposu e, he issues hea up indi ec ly, due o he mal
di usion om he me allic componen s o he implan , whe eas du ing exposu e o RF, he
hea di uses om he issues o he implan , whe e he di ec he mal e ec o RF ields is
negligible.
Pennes' biohea equa ion was exp essed in e ms o empe a u e ele a ion wi h espec o
he empe a u e a es and sol ed, by means o a home-made code3,5. In he case o he
ana omical body models, he biohea equa ion also modeled he blood pe usion in he
issues, acco ding o he pa ame e s epo ed in he da abase o he IT'IS Founda ion. This
led o wo empe a u e inc ease dis ibu ions, one due o GC alone and ano he o RF alone.
Finally, he empe a u e inc ease due o he simul aneous p esence o g adien and RF
e ec s was ob ained as summa ion o he p e ious GC and RF empe a u e dis ibu ions.
Thanks o he linea i y o he Pennes' equa ion, his co esponded o sol e he p oblem using
he sum o he wo powe dis ibu ions as he o cing e m.
The analysis wi h he ASTM phan om combined, poin by poin , he spa ial dis ibu ions o
he empe a u e inc eases p oduced by he GC and RF ields, scaled o he same peak alue
wi hin he phan om.
Simula ions in ol ing he MR pulse sequences and ealis ic body models scaled he g adien
powe acco ding o he unning sequence wa e o m and he RF powe acco ding o he
equi ed lip angle, RF pulse du a ion and modula ion en elope, i.e. a sinc unc ion wi h 0.5
apodiza ion coe icien .
Bo h powe s we e a e aged o e he epe i ion ime (TR) o he sequence and used as a
sou ce o he he mal p oblem. The GC and RF empe a u e inc eases we e combined in
wo ways: he i s by summing GC and RF empe a u e inc eases scaled o he same
maximum alue ac oss he landma ks, and he second by summing GC and RF empe a u e
inc eases as ob ained om he ac ual powe s gene a ed by he ele an MR pulse
sequence.
All empe a u e inc eases a e p o ided as peak alues ac oss he compu a ion domain, o
a 30 min exposu e o he di e en powe sou ces.
8 Ac i i y A3.3.6
INRIM and ITIS will de ine co ec ion ac o s o accoun o he e ec o supe posi ion o RF
and GC hea ing o be conside ed o an ex ension o he ASTMF2182 S anda d ela ed o
RF hea ing o passi e de ices.
20 o 46
Table 4 collec s he ela i e enhancemen s o he peak empe a u e inc eases in he ASTM
phan om due o he simul aneous exposu e o GC and RF ields, wi h espec o he peak
empe a u e inc ease ob ained wi h GC o RF alone scaled o he same e e ence alue.
Rela i e enhancemen s ange om 17-67% a 1.5T and om 24-51% a 3T, depending on
he implan .
Table 4: Rela i e enhancemen s o he maximum empe a u e inc ease due o he simul aneous applica ion
o GC and RF ields wi h he ana omical body models, wi h espec o he empe a u e inc ease due o GC o
RF alone. Fo each implan and MR pulse sequence, RF and GC empe a u e inc eases we e scaled o he
same maximum alue ac oss he landma ks.
The exposu e condi ions o he implan s inside he ASTM phan om a e qui e di e en om
hose in a clinical se up. This e lec s on he empe a u e inc ease enhancemen s ob ained
when he analysis in ol ed MR pulse sequences and ealis ic body models, and he GC and
RF empe a u e inc eases we e scaled o he same maximum alues ac oss all landma ks
be o e being combined (Figu e 13 and Figu e 14) . In his ega d, Table 5 collec s he ela i e
enhancemen s o he peak empe a u e inc eases in he cases o ealis ic MR pulse
sequences (Table 6) and ana omical human body models.
The esul s show ha he maximum ela i e enhancemen s ne e exceed 9% and 11% a
1.5T and 3T, espec i ely. I is in e es ing o no ice ha , mos o he ime, he maximum
empe a u e inc ease due o he simul aneous exposu e o GC and RF occu s a he same
landma k as in he case o exposu e o he GC ield alone.
21 o 46
Figu e 13: Peak empe a u e inc ease, in kel in, on he six implan s exposed o di e en MR pulse sequences
as a unc ion o mul iple axial landma ks. Resul s e e o 1.5T exposu e o he RF EM ield alone ( ed line),
GC EM ield alone (g een line) and simul aneous applica ion o RF and GC EM ields (blue line). The g ay a ea
highligh s he enhancemen o he peak empe a u e inc ease due o he simul aneous exposu e o RF and
GC EM ields wi h espec o he maximum be ween he peak empe a u e inc ease due o RF o GC alone.

22 o 46
Figu e 14: Peak empe a u e inc ease, in kel in, on he six implan s exposed o di e en MR pulse sequences
as a unc ion o mul iple axial landma ks. Resul s e e o 3T exposu e o he RF EM ield alone ( ed line), GC
EM ield alone (g een line) and simul aneous applica ion o RF and GC EM ields (blue line). The g ay a ea
highligh s he enhancemen o he peak empe a u e inc ease due o he simul aneous exposu e o RF and
GC EM ields wi h espec o he maximum be ween he peak empe a u e inc ease due o RF o GC alone.
23 o 46
Table 5: Rela i e enhancemen s o he maximum empe a u e inc ease due o he simul aneous applica ion
o GC and RF ields wi h he ana omical body models, wi h espec o he empe a u e inc ease due o GC o
RF alone. Fo each implan and MR pulse sequence, RF and GC empe a u e inc eases we e scaled o he
same maximum alue ac oss he landma ks.
24 o 46
Table 6: Main pa ame e s o he sequences conside ed o he analysis. Maximum Whole-Body (WB) a e aged
SAR e e s o he maximum including all implan s and exposu e landma ks o he speci ic sequence. The
h ee dB/d ms alues epo ed o he EPI sequence e e o he X, Y and Z eadou di ec ions, espec i ely.
Figu es 13 and 14 show he peak empe a u e inc ease o 1.5Tand 3% scena ios,
espec i ely, esul ing om he exposu e o he speci ic MR pulse sequences whose
pa ame e s complied wi h hose lis ed in Table 6. The igu es epo he empe a u e
inc eases as a unc ion o he axial posi ion o he implan ba ycen e wi h espec o he
scanne isocen e . The selec ion o he epo ed esul s was pe o med on a case-by-case
basis, acco ding o he combina ion o implan and sequence leading o he highes
enhancemen s o he empe a u e inc ease due o he simul aneous e ec o RF and GC
EM ields wi h espec o RF and GC alone. This choice did no necessa ily lead o show he
cases whe e he absolu e maximum empe a u e was eached. Ne e heless, he maximum
peak empe a u e inc eases also played a ole in excluding hose scena ios whe e he
hea ing due o bo h RF and GC we e less p ominen . The g ay a eas in he igu es highligh
he enhancemen o he peak empe a u e inc ease when he RF and GC EM ields a e
applied simul aneously wi h espec o he maximum be ween he peak empe a u e
inc ease due o he applica ion o RF o GC EM ield alone. Scena ios wi h he g ea es
dis ance be ween he blue and he dashed lines ep esen hose whe e he addi ion o he
wo e ec s has he la ges e ec on he combined empe a u e inc ease. In his ega d,
Table 7 collec s he maximum peak empe a u e inc ease enhancemen s o each exposu e
scena io. The able epo s he enhancemen s as he peak empe a u e inc ease ollowing
he combined applica ion o GC and RF, epo ed nex o he maximum be ween he peak
empe a u e inc ease due o RF o GC alone.
25 o 46
Table 7: Maximum peak empe a u e enhancemen s, among all he landma ks o each speci ic scena io.
Each able en y epo s he empe a u e inc ease due o RF o GC alone ( he highes o he wo as epo ed
in he b acke s) on he le o he empe a u e inc ease due o he combined e ec o RF and GC. The
empe a u e enhancemen is added in squa e b acke s nea he empe a u e inc ease alues. All empe a u e
inc eases a e epo ed in kel in. Fo each implan , he i s ow e e s o esul s a 1.5T and he second ow a
3T. When he empe a u e inc ease enhancemen esul ed o be negligible o all he landma ks, a hyphen is
epo ed in he co esponding able en y. Resul s shown in Figu es 13 and 14 a e epo ed in bold.
The U.S. Food & D ug Adminis a ion (FDA) sugges s o include, along wi h he MR
Condi ional labeling o a Medical De ice, he maximum allowable g adien slew a e pe
axis, he maximum pe mi ed Whole-Body (WB) a e aged SAR and/o maximum pe mi ed
head a e aged SAR and he maximum B1+ ms admissible alue. Cons aining he
maximum g adien slew a e educes excessi e GC hea ing, whe eas limi ing he maximum
SAR and/o B1+ is use ul o limi RF hea ing. The e o e, he p e ious quan i ies a e es ic ed
on he basis o es ing esul s and a sa e y h eshold in e ms o maximum accep able
empe a u e inc ease. Independen ly o he ac ual alue o he empe a u e inc ease
h eshold, when GC and RF es ing a e pe o med sepa a ely, applying independen limi s
is app op ia e as long as he wo hea ing e ec s do no combine signi ican ly. When his
holds ue, i an MRI scan is pe o med e.g., a he maximum B1+ alue, he peak o he RF-
induced empe a u e inc ease may be close o he es ablished h eshold, i he clinical
exposu e is close o wo s -case. O cou se, i a he same maximum B1+ alue, he g adien
also con ibu es signi ican ly o he hea ing, he ac ual peak empe a u e inc ease may be
highe han he sa e y h eshold. As a consequence, i he in e ac ion o RF and GC hea ing
is no p ope ly accoun ed o , maximum empe a u e inc eases could each unexpec ed and
unsa e alues, exposing he pa ien o a po en ial haza d.
The analyzed implan s we e no labeled as MR condi ional, he e o e he abo e in o ma ion
was no a ailable; no is he e any gene ally accep ed published alue o maximum
accep able empe a u e inc ease. Fo hese easons, he s udy abs ac ed om hese
inpu s, i s by scaling he peak empe a u e inc eases ob ained om GC and RF exposu e
o he same e e ence alue, and hen by conside ing an exposu e o he EM ield gene a ed
by MR sequences complying wi h no mal ope a ing mode and ha dwa e wa chdog.
32 o 46
While he conse a i eness o Tie 4 is limi ed by he access o he CAD model o he si e
o he wo s -case inciden ield, he p oposed me ic can achie e ull co e age o he a ge
su ace o olume and gene a e he desi ed s a is ics using he same ools and e o as
Tie s 2 / 3.
11 Ac i i y A3.4.2
ITIS will pe o m alida ion SAR/ empe a u e RF measu emen s on selec ed gene ic
implan s om A3.4.1 wi h de ined condi ions o assess he accu acy and conse a i eness
o he p edic ion app oaches.
11.1 Valida ion and Con idence In e al o RF-Hea ing
The alida ion o he accu acy and conse a i eness o he “Tie 2+” app oach de ined in
A3.4.1 ex ends he con idence in e al o he Tie 3 lib a y and oolse , IMAnaly ics wi h
MRIxViP, om whe e he inciden ields a e de i ed. The con idence in e al o he use o
IMAnaly ics wi h he MRIxViP lib a y o a gene ic elonga ed implan , which was es ima ed
by ollowing he s anda d guidelines de ined by he Guide o he Measu emen o Unce ain y
(GUM) and by he ASME V&V 10, is shown in Table 8. The p ocedu e is based on desc ibing
he o al unce ain y o he e alua ion as he con olu ion o independen unce ain y sou ces
using Type B e alua ion. Each unce ain y e m was e alua ed in e ms o i s con ibu ion o
a ia ion o he inal esul (powe deposi ion a a ho spo o induced ol age a lead
e minal).
The e o e, he unce ain y con ibu ion o each e m comp ises bo h he in insic a ia ion o
he sou ce o unce ain y, and he local sensi i i y o he inal esul o ha a ia ion − hus
he unce ain ies a e p esen ed as uni less ac o s, in decibels (as hey a e applicable o
deposi ed powe and o ol age).
The unce ain y associa ed wi h o he IMAnaly ics inpu s, such as unce ain y o he ans e
unc ion models o ou ings/ egions o in e es , mus be assessed sepa a ely and included
in he inal con idence in e al.

33 o 46
Table 8: Con idence in e al o IMAnaly ics + MRIxViP ou pu (deposi ed powe o induced ol age) due o
unce ain y o in i o exposu e condi ions.
34 o 46
The con idence in e al o ISO 10974 Tie 3 e alua ions o deposi ed powe a ound SAR
ho spo s in issue, o induced ol age in a channel, was assessed o es s pe o med by he
IT'IS Founda ion on AIMD in p e ious EMPIR p ojec (MIMAS 17IND01). In ha epo , he
calcula ion o empe a u e ise, he mal dose, and issue damage co esponding o he
assessed in i o deposi ed powe we e no discussed as being ou side he scope o he ISO
10974 app oach. Unlike deposi ed powe , he local empe a u e ise depends s ongly on
he shape o he lead ip as well as he issue en i onmen a ound he ho spo ( issue
p ope ies and dis ibu ion).
In his wo k, he MIMAS epo was ex ended o co e in i o empe a u e ise unce ain y
a ound s ong g adien s such as ound nea sc ew ips o AIMD elec odes.
To app op ia ely measu e he empe a u e ise, a sys em check has o be conduc ed o
gua an ee p ope unc ionali y o he equipmen . I possible, he ace ack in he icini y o
he elec ode should be cu away o minimize i s con ibu ion o he measu emen . The
sys em check consis s o he ollowing s eps:
measu e he elec ical and he mal p ope ies o he TSM oge he wi h he
co esponding empe a u e;
 e i y he co ec ope a ion o he measu emen sys em,
by conduc ing a backg ound empe a u e ise measu emen and compa ing i o he
expec ed alue om he measu ed E- ield. In o de o minimize e o s om dis u bing he
es i em and slu y, he sys em check is pe o med wi h he es i em in place. The
backg ound empe a u e ise measu emen loca ion should be selec ed o ha e he leas
enhancemen due o he implan (<-30dB) and be a leas 2cm away om any in e ace.
1. Fill up he desi ed phan om wi h esh TSM in slu y o m. TSM should no con ain ai
bubbles.
2. Measu e he elec ical and he mal conduc i i y o he TSM using a he mome e , he
DAK, and he he mal conduc i i y me e . All alues should be close o nominal, i.e. wi hin
he ange de ined by he unce ain y budge o you es p o ocol.
3. Le he co e ed phan om es (e.g. o e -nigh ) o equilib a e and o ai bubbles o
dissol e.
4. Be o e s a ing he measu emen , check ha TSM p ope ies ha e no d i ed.
5. Moun he phan om in he exposu e sys em, and con igu e i o expose he es i em o a
uni o m inciden E- ield
7. Valida e he inciden B- ield as app op ia e.
9. Moun he empe a u e p obe and posi ion i a he de ined loca ion o backg ound
empe a u e ise measu emen .
10. Reco d he baseline empe a u e; empe a u e d i should be ≥10dB below he
measu ed empe a u e ise o he sample.
35 o 46
12. Engage he exposu e. The measu emen p o ocol should eco d a empe a u e ise o
a leas 5 × senso esolu ion; he o al empe a u e ise should be <3K o a oid change o
he TSM p ope ies.
13. Moun he SAR p obe and measu e he o al E- ield a he same loca ion as he
empe a u e, o a ing he p obe 360◦ and a e aging o al E.
14. The measu ed ∆T has o be equal o σE2/(ρc)× meas wi hin he con idence in e al de ined
in Table 9. I his is he case hen he measu emen sys em is alida ed.
15. Mo e o he i s measu emen poin de ined o he es i em (e.g., da k blue in Figu e
17). Repea 10 - 12.
Repea o he desi ed numbe o measu emen poin s.
The backg ound empe a u e ise should be alida ed agains he analy ical solu ion (SAR=
C dT/d ). The SAIMD-1-200mm empe a u e ise can be alida ed agains nume ical
simula ions, as shown in Figu es 18 and 19. Ta ge alues can also be ob ained om Annex
I o ISO 10974.
Figu e 17: Schema ic o SAIMD-1-200mm wi h he loca ions o in e es (da k and ligh blue) o empe a u e
measu emen .
Figu e 18: Nume ical benchma k o p1 and p2, no malized o backg ound inciden ield. The backg ound
empe a u e ise has been emo ed om he empe a u e ise due o SAIMD-1-200mm. The nume ical da a
has been a e aged o e he dimensions o he ip o he p obe (d = 0.88mm).
36 o 46
A co ec ion ac o has o be applied o compensa e o he physical size o he
empe a u e p obe and he empe a u e g adien in z-di ec ion (g adien compensa ion
sec ion in Table 9). Figu e 19 shows he co ec ion ac o , de i ed om a linea i o
SAIMD-1 measu emen s wi h wo di e en models o empe a u e p obe.
Table 9: Example con idence in e al o empe a u e ise alida ion a he highes hea ing poin (poin 1, da k
blue) o SAIMD-1-200mm. Re e ences ma ked wi h [?] a e no gene alized and should be upda ed as
app op ia e o he TSM selec ed.
37 o 46
Figu e 19: Co ela ion be ween measu emen de ia ion and he empe a u e g adien in he p obe di ec ion
a bo h 64 (big symbols) and 128 MHz (small symbols). The ou lie s (ma ked wi h black ci cles) a e excluded
om he linea i .

38 o 46
11.2 Valida ion and Con idence In e al o GC Hea ing
In his ac i i y, he wo k was ex ended o include his unce ain y o in i o RF and GC
exposu es. The measu emen s we e conduc ed on a semi- igid i anium (ASTM F67) c anial
mesh manu ac u ed by Meda is AG (Basel, Swi ze land). The implan was 93 mm×93 mm
and i s hickness was 0.6 mm. The hickness o he implan co esponds o he diame e o
i s one-dimensional s uc u e, so ha 𝑟= 0.3 mm. I was equipped wi h h ee op ical ib e
empe a u e p obes, wo posi ioned in he pe iphe al egions, whe e he la ges empe a u e
inc ease was expec ed o be induced, and one in he cen al egion, as shown in Figu e 20.
The p obes we e connec ed o an eigh -channel AccuSens signal condi ione p oduced by
Opsens (Québec, Canada) sampling a 20 Hz. The manu ac u e decla ed an o e all
accu acy o ±0.30 ◦C and a esolu ion o 0.01 ◦C.
39 o 46
Figu e 20: Expe imen al se up and i ual modelling. (a) Semi- igid c anial mesh in i anium equipped wi h he
op ical ib e empe a u e p obes. (b) Sys em o GCs o cylind ical MRI scanne s. (c) Phan om illed wi h
expanded polys y ene g ains. (d) Compu a ional model o he c anial mesh wi h depic ed he boxes o he
ac i e pa s o he empe a u e p obes. (e) Compu a ional domain including he phan om wi h he c anial mesh
and he model o he GCs.
40 o 46
In a i s expe imen , simula ing he hea ans e om he c anial mesh owa ds so issues,
he implan wi h he posi ioned p obes we e plunged in o a cuboid con aine (base o 13 cm
× 44 cm, heigh o 20 cm) illed wi h gel simula ing a e age issue pa ame e s p oduced by
Zu ich MedTech AG (Zu ich, Swi ze land) in acco dance wi h ISO/TS 10974. In a second
expe imen , simula ing an almos adiaba ic condi ion, he same cuboid con aine was illed
wi h expanded polys y ene g ains whose diame e a ies be ween 1 mm and 3 mm. The
base o he implan was ixed in a slo cu in a spongy suppo o keep i in he co ec posi ion
wi hin he con aine . In pa icula , he implan was posi ioned a 45◦ wi h espec o he
con aine walls o ha e i pe pendicula o he gene a ed magne ic ield, which is he
con igu a ion expec ed o maximize he induced empe a u e inc ease.
The magne ic ield was gene a ed by he sys em o ac i ely shielded GCs wi h whole-body
access o cylind ical MRI scanne s. P ecisely, i is he model Sola is-R manu ac u ed by
Nanjing Cichen Medical Technology Co., L d (Nanjing, China). The sys em, ea u ing h ee
coils o gene a e g adien magne ic ields along h ee o hogonal di ec ions, has an in e nal
diame e o abou 67 cm and a o al leng h o abou 150 cm. The g adien di ec ions
de e mine he used e e ence sys em, wi h 𝑧di ec ed longi udinally wi h espec o he coils,
𝑦di ec ed e ically and 𝑥pe pendicula o he o he wo di ec ions.
The GCs we e supplied by a NG500 1.3 g adien ampli ie buil by P od i e Technologies
(Eindho en, The Ne he lands) able o d i e each coil independen ly and o p o ide a peak
cu en o 1000 A and a maximum ol age o 940 V. To dissipa e he loss powe in he coil
conduc o s due o he high cu en alues, a wa e cooling sys em was connec ed o he
coils. The cooling sys em made negligible he impac o he hea ing dissipa ed by he GCs’
conduc o s on he empe a u e alues measu ed on he implan , as e i ied h ough a ou h
op ical ib e empe a u e p obe posi ioned a he bounda y o he phan om, whe e no hea ing
due o he cu en s induced on he c anial mesh was expec ed. In he expe imen , he coils
gene a ing g adien s di ec ed along 𝑥and 𝑧we e d i en wi h wo in-phase sinusoidal
cu en s a he equency o 2 kHz. Acco ding o he nume ical model desc ibed in Sec ion
2.2.2, he peak cu en in ensi ies o 150 A we e such ha a peak magne ic ield o 3.5 mT
was gene a ed a he implan ba ycen e, loca ed a 𝑥= 14 cm, 𝑦= 0 and 𝑧= 30 cm wi h
espec o he coil isocen e (i.e., he cen al poin whe e all he coils gene a e a null magne ic
ield). The empe a u e de ec ed by he op ical ib e empe a u e p obes we e eco ded
e e y 2 s o 900 s o con inuous exposu e o he c anial mesh o he ha monic magne ic
ield gene a ed by he GCs. Tempe a u e eco ding s a ed 30 s be o e swi ching on he
powe ampli ie o acqui e he alue a which he sys em and he p obes we e he malized.
Since he ac ual measu and o be compa ed wi h he simula ion esul s is he empe a u e
inc ease, an o se equal o he a e age o he empe a u e alues eco ded in he
p elimina y 30 s we e applied o he measu emen esul s.
The measu emen s we e epea ed wo imes a day apa wi hou mo ing nei he he p obe
no he phan om, o assess he epea abili y o he expe imen .
Figu e 21 depic s he measu emen pipeline. The empe a u e p obes a ached o he
implan we e connec ed o he empe a u e acquisi ion.
41 o 46
Figu e 21: Expe imen pipeline. The expe imen was moni o ed by a con olle PC ha collec ed he acqui ed
empe a u e da a h ough a se ial connec ion wi h he empe a u e acquisi ion sys em. The same con olle
PC con olled he g adien ampli ie h ough a LAN connec ion and moni o ed he gene a ed cu en s h ough
a Thunde bol connec ion and an acquisi ion boa d connec ed h ough coaxial cables o he g adien ampli ie .
The g adien ampli ie supplied he GCs h ough high cu en cables and he c anial mesh was exposed o he
gene a ed magne ic ield.
Figu e 22: Compa ison be ween measu emen s and simula ions. The noisy pale o ange and pale ed lines
epo he measu emen esul s in he i s and second epe i ion, espec i ely. The solid g een and blue lines
epo he simula ed esul s wi h he he mal seed model (pu ely 3D FEM) and he 3D–1D coupling,