P ep in No ice
This manusc ip is a p ep in and has no been ce i ied by pee e iew.
I has been made publicly a ailable ia Zenodo o acili a e ea ly dissemina ion and
eedback. The inal e sion o his wo k may di e a e pee e iew.
This manusc ip is in ended o submission o Magne ic Resonance in Medicine
(MRM).
© 2025 Ali Ajouz, Ola Jansen, Lynn Johann F ohwein, S ea Seeha e , Naomi La sen,
Jan-Be nd Hö ene .
This wo k is licensed unde a C ea i e Commons A ibu ion-NonComme cial-No
De i a i es 4.0 In e na ional License (CC BY-NC-ND 4.0).
Fo de ails, see: h ps://c ea i ecommons.o g/licenses/by-nc-nd/4.0/.
P ep in DOI: h ps://doi.o g/10.5281/zenodo.15063820.
Au oma ic de e mina ion o glympha ic
low wi h he DTI-ALPS-index along he
p incipal axis sys em in na i e imaging
space co ec s o head and ib e
o ien a ion
Ali Ajouz,1,2,3 Ola Jansen,1 Lynn Johann F ohwein,3 S ea Seeha e ,1 Naomi La sen,1 and Jan-
Be nd Hö ene 1,2*
1 Depa men o Radiology and Neu o adiology, UKSH, CAU Kiel, Kiel, Ge many.
2 Depa men o Radiology and Neu o adiology, SBMI, CAU Kiel, Kiel, Ge many.
3 Siemens Heal hinee s AG, Fo chheim, Ge many.
Co esponding au ho : Jan-Be nd Hö ene ; Mail: jan.hoe ene @ ad.uni-kiel.de
Abs ac
Keywo ds: Neu o luids, DTI-ALPS-index, glympha ic sys em, b ain clea ance, image
egis a ion
Pu pose: This wo k aims o dec ease he dependence o he DTI-ALPS-index on he
head and ib e o ien a ion, calcula ing he DTI-ALPS-index along he p incipal
di usion di ec ions. I is in es iga ed i an au oma ed ROI placemen could lead o a
mo e obus DTI-ALPS-index wi hou na i e da a egis a ion.
Me hods: We calcula ed he DTI-ALPS-index along he p incipal di usion di ec ions
(ALPS-PAS) and compa ed i wi h he o iginal DTI-ALPS-index along he scanne ’s
labo a o y ame (ALPS-LAB) using simula ion and in i o measu emen s.
To calcula e he DTI-ALPS-index in na i e space, we de eloped a no el calcula ion
algo i hm o an au oma ic ROI placemen echnique and compa ed i o a manual
ROI placemen and exis ing a las-based ROI placemen . Fu he , a whole-b ain DTI-
ALPS-map was in oduced.
Resul s: Simula ions demons a ed he dependence o he DTI-ALPS-index on he
head and ib e o ien a ion and he in i o measu emen s yielded highe ALPS-PAS
han ALPS-LAB. The no el ROI placemen led o a mo e obus DTI-ALPS-index
e alua ion in meaning ul egions han he manual ROI placemen . The DTI-ALPS-
map indica ed an aniso opy be ween he second and hi d p incipal di usion
di ec ion in o he b ain a eas besides he ALPS- ib e- egions.
Conclusion: ALPS-PAS elimina es he head and ib e o ien a ion dependence, which
enables he calcula ion in he na i e da a space wi hou egis a ion o he acqui ed
da a. The au oma ic ROI placemen educes he ope a o dependen ROI selec ion,
which could be bene icial o longi udinal s udies.
In oduc ion
The clea ance o me abolic was e and solu e anspo a e essen ial o main aining
p ope b ain unc ion1. E idence sugges s ha his p ocess is acili a ed by a glial
lympha ic (glympha ic) sys em, hough i emains di icul o iden i y bo h in i o
and h ough his ological analysis. Me abolic was e p ima ily o igina es om
neu onal ac i i y, including byp oduc s o neu o ansmi e me abolism, mis olded
p o eins, and he b eakdown o cellula componen s1,2.
I was sugges ed ha he glympha ic sys em ope a es wi hin he pe i ascula space
(PVS), whe e aquapo in-4 (AQP4) channels suppo wa e exchange wi h he b ain
pa enchyma, clea ing solu es and p o eins such as amyloid-be a3.
Se e al MRI based me hods we e sugges ed o in es iga e selec ed aspec s o his
sys em4. These include me hods wi h5 and wi hou Gadolinium-based con as
agen s (GdCA). While GdCA me hods allow acking he pa hway o a molecule ( he
Gd-complexes) o minu es o hou s wi h compa ably high signals, he abili y o
GdCA o p obe AQP4 channels is unclea 1. Me hods wi hou GdCA, on he o he
hand, image wa e molecules, bu may su e om less signal and a sho e
obse a ion window ( o example a e ial spin labeling (ASL)6, chemical exchange
sa u a ion ans e (CEST)7, in a oxel incohe en mo ion (IVIM)8 , and di usion
enso imaging along he pe i ascula space (DTI-ALPS-index)9–11.
The la e ound ha in a speci ic b ain a ea, he di usion (o slow low) o wa e
molecules pe pendicula o he dominan di ec ion o neu onal ib es is aniso opic.
I was shown ha he di usion weigh ed signal pe pendicula o he dominan ib e
o ien a ion, along he pos ula ed mic o ascula u e, was weake han he di usion
weigh ed signal pe pendicula o he dominan ib e o ien a ion and pe pendicula
o he mic o ascula u e, o example by 40% in heal hy con ols9. This aniso opy
was ound o dec ease wi h demen ia and age9, sugges ing ha his “ low” dec eases
wi h hese ac o s.
These esul s may sugges a glympha ic low along he pe i ascula spaces (ALPS) in
selec ed a eas o he human b ain. The o iginal and cu en implemen a ions o
DTI-ALPS-index o en use he di usion in he Ca esian di ec ions o he labo a o y
ame (LAB) o de e mine he index (o low). Howe e , he labo a o y ame
(de ined by he physical o ien a ion o he g adien coils) is no necessa ily aligned
wi h he ne e ib es in he b ain. Head and ib e o ien a ion will a ec he di usion
me ics9 and hus he DTI-ALPS-index.
Di e en head o ien a ion may be add essed by ans o ming he measu ed da a o
a s anda d head o ien a ion, bu equi es hea y p ocessing12,13 and does no
accoun o indi idual ib e o ien a ions ( o example he FMRIB58 empla e in
Mon eal Neu ological Ins i u e (MNI) space). A me hod o calcula e he DTI-ALPS-
index in he un ans o med pa ien space, while co ec ing o he head o ien a ion
and aking he ib e o ien a ion in o accoun , howe e , was no epo ed.
He e, we p opose o measu e he aniso opy wi h espec o he main ib e
o ien a ion ins ead o he Ca esian coo dina es o he scanne . We de e mined he
main ib e o ien a ion by calcula ing he main di usion di ec ions in a p incipal axis
sys em (PAS)14,15 (Figu e 1B) and calcula ed he DTI-ALPS-index acco dingly.
Ano he key ac o o de e mine he DTI-ALPS-index eliably is he ( ep oducible
and p ecise) selec ion o he egion whe e he index is calcula ed. T ans o ming he
acqui ed da a o an a las wi h p eselec ed egions13 in ol es non- igid
ans o ma ions wi h many deg ees o eedom and s ongly a ec s he measu ed
MRI da a16 (Figu e 1C). He e, we p opose o calcula e DTI-ALPS-index in egions
de ined on an a las and ans e ed o he measu ed images. Thus, we de eloped an
a las-based me hod o au oma ically selec egions on he acqui ed da a, calcula ed
he DTI-ALPS-index bo h in he LAB and in he PAS, and compa ed i o he alues
ob ained in an a las space. We hypo hesized ha hese me hods a) inc ease he DTI-
ALPS-index and b) educe he a iabili y.
Me hods
DTI-ALPS-index compu a ion
The di usion enso 𝐷
can be desc ibed in he LAB o an MRI-scanne as
𝐷
=[Dxx Dxy Dzz
Dyx Dyy D𝑦𝑧
D𝑧𝑥 D𝑧𝑦 D𝑧𝑧] , (1)
whe e x, y and z a e he indices o he di usion coe icien s and e e o he
Ca esian coo dina es o he scanne s’ g adien s. In his ame o e e ence (LAB),
he DTI-ALPS-index (ALPS-LAB) was in oduced in Re .9 as (Figu e 1A):
ALPS-LAB =𝐷𝑥𝑥,𝑝𝑟𝑜𝑗,𝐷𝑥𝑥,𝑎𝑠𝑠𝑜𝑐
𝐷𝑦𝑦,𝑝𝑟𝑜𝑗,𝐷𝑧𝑧,𝑎𝑠𝑠𝑜𝑐 , (2)
whe e 𝑝𝑟𝑜𝑗 desc ibes he supe io co ona adia a (SCR) egion in he b ain whe e
p ojec ion ib es a e dominan , and 𝑎𝑠𝑠𝑜𝑐 desc ibes he supe io longi udinal
asciculus (SLF) egion in he b ain whe e associa ion ib es a e dominan . SCR and
SLF e e o he ALPS- ib e- egions.
Diagonalising 𝐷
yields 𝛬, which ansla es o a ans o ma ion o he coo dina e
sys em, esul ing in he p incipal axis sys em (PAS) o p incipal di usion di ec ions
as he eigen basis o he new coo dina e sys em17. 𝛬 is de ined as
𝛬=[𝜆10 0
0 𝜆20
0 0 𝜆3] , (3)
whe e he eigen alues o 𝐷
a e in he co esponding o de : 𝜆1≥ 𝜆2≥𝜆3 (Figu e
1B). The diagonalised ma ix is used o calcula e ALPS-PAS14,15:
ALPS-PAS = 𝜆𝑥,𝑝𝑟𝑜𝑗,𝜆𝑥,𝑎𝑠𝑠𝑜𝑐
𝜆𝑦,𝑝𝑟𝑜𝑗,𝜆𝑧,𝑎𝑠𝑠𝑜𝑐 ,(4)
whe e 𝜆𝑥 is 𝜆2 o 𝜆3 depending on whose eigen ec o has he la ge x componen ,
and 𝜆𝑦 is he emaining 𝜆2 o 𝜆3 (no e ha 𝜆1 is he main di usion di ec ion along
he dominan ib es and no used o calcula e ALPS-PAS). In he ALPS-LAB and
ALPS-PAS calcula ions, he di usion coe icien s (Dxx, Dyy 𝑎𝑛𝑑 Dzz) o , in PAS
(𝜆1,𝜆2 𝑎𝑛𝑑 𝜆3) a e selec ed om he enso s in he oxels o he ROIs desc ibed la e
in his chap e . The o e ba s indica e he a e ages.
Figu e 1: (A) Idealized scheme o he ascula u e (g ey) in he p ojec ion (blue), associa ion (g een) and
subco ical ( ed) ib e a ea. In classical ALPS-LAB, he di ec ions o hese ib es a e assumed o coincide wi h he
Ca esian coo dina e sys em in he labo a o y ame o e e ence (𝒙
, 𝒚
, 𝒛). (B) Analysing he di usion enso
shows ha i s p incipal componen s (𝒆1, 𝒆2, 𝒆3) do no necessa ily align wi h he Ca esian sys em.
(C1) Na i e ac ional aniso opy (FA) map o he olun ee , (C2) a e ans o ma ion in o he space o an a las
(MNI152, FMRIB58), which is shown in (C3). No e how s ongly he ea u es be ween C1 and C2 changed.
(D) Schema ic iew o he wo k low used o he au oma ic e alua ion o ALPS-LAB and ALPS-PAS in he
olun ee DWI and a las MNI spaces. In olun ee space, he ma ix R/R-1 was used o ans o m he b ain mask
om T1w space o DWI space. The ma ix F / F-1 was used o ans o m be ween he olun ee DWI and a las
MNI space. Fou olumes o size 4x4x4 mm3 in he SCR and SLF we e selec ed in he MNI space.
Simula ion expe imen s and Ka dan-angles
We pe o med simula ions, o in es iga e, how s ongly ALPS-LAB is a ec ed by a
misma ch be ween ib e and g adien o ien a ion. We assumed a di usion enso 𝐷
o he p ojec ion a ea diagonal in he labo a o y ame wi h i s eigen alues: 1.7, 0.4,
0.2 µm2/ms. We applied selec ed o a ions o he coo dina e sys em o 𝐷
( o a ions
o indi idual o all axes, de ined by Ka dan-angles (Ψ, Θ, Φ)18 o he in insic
o a ion (ZY’X’’), Suppo ing Figu e S1.1, Suppo ing Figu e S1.3) and calcula ed
ALPS-LAB. No e, o he simula ion, we c ea ed one enso in one oxel. The e o e,
we calcula ed ALPS-PAS by di iding he second eigen alue by he hi d o ALPS-
PAS and Dxx by Dyy o one enso only. Fo ALPS-LAB, we always di ided Dxx by
Dyy, as he simula ed ini ial enso has an o ien a ion like he enso s o he
p ojec ion ib es.
Wi hou o a ion, ALPS-LAB is “ideal” and iden ical o ALPS-PAS. We used he
a io o ALPS-LAB / ALPS-PAS o assess he simila i y o bo h indices.
In i o expe imen s
In i o measu emen s we e acqui ed a 3T (MAGNETOM Cima.X scanne , Siemens
Heal hinee s AG, Fo chheim, Ge many) wi h a maximal g adien ampli ude o 200
mT/m, a maximum slew a e o 200 T/m/s and a wo-channel body ansmi coil,
and 64-channel ecei e head neck coil. Using a slice selec i e, di usion weigh ed
g adien echo sequence wi h EPI eadou , 66 di usion olumes (b = [0,1000]
s/mm², di usion g adien di ec ions = [6, 30], a e ages = [1, 2]) plus six b = 0
s/mm² images wi h e e sed phase encoding di ec ion we e acqui ed (FOV = 220 x
220 mm2, 50 slices wi h 2 mm hickness, 2 mm iso opic esolu ion, ma ix size
110x110x50, TE = 41 ms, olume TR = 6600 ms, eadou bandwid h = 1698 Hz/Px,
lip angle = 90°, GRAPPA ( ac o 2, 28 e e ence lines), phase pa ial Fou ie 0.75,
e ec i e echo spacing = 0.3 ms). In addi ion, a 3D T1-weigh ed (T1w) MP-RAGE
image was acqui ed (FOV = 256 x 256 x 192 mm3, 1 mm iso opic esolu ion, TE =
2.45 ms, TR = 1900 ms).
The p o ocol was applied o nine heal hy subjec s (age: 23-49, six emales
and h ee males, Suppo ing Table S1). Tes - e es measu emen s we e conduc ed
on wo subjec s (3, 9) wi h a i e-minu e b eak in be ween. Subjec one was
measu ed wi h and wi hou delibe a e head o a ion o he op- igh hand co ne .
Image p ocessing
T1w images we e co ec ed o bias ields, denoised and used o gene a e a
b ain mask (ANTs19, Figu e 1D). DWI pos -p ocessing included denoising, un inging
(M ix20), dis o ion and eddy cu en co ec ions (FSL16). A ans o ma ion ma ix
R be ween T1w and DWI images was calcula ed and used o ans o m he b ain
mask o he DWI da a (Epi eg, Vec eg)16. The di usion enso and colo -coded FA
(RGB) maps (Dipy21) we e calcula ed on he masked di usion da a.
In addi ion, he b ain-masked DWI da a was ans o med o he MNI a las
space by calcula ing a ans o ma ion ma ix F using he FA maps o bo h spaces
wi h linea and non-linea ans o ma ions ( li and ni )16.
ROIs
We calcula ed ALPS-LAB and ALPS-PAS o he le and he igh hemisphe e
by c ea ing ROIs in he ALPS- ib e- egions in h ee di e en ways. All ROIs ha e he
same olume (4x4x4 mm3). Fo each way, ou ROIs we e de ined, wi h one ROI in
each ALPS- ib e- egion. This esul ed in wo ROIs pe hemisphe e: one in he SCR
and one in he SLF, on bo h he le and igh sides (Figu e 1D).
A) MNI ROIs: The ou ROIs we e placed on a b ain empla e
(FMRIB58_FA_1mm, MNI152 space22–24) (exis ing a las based
au oma ed way13).
B) Manual ROIs in DWI space: An expe ienced adiologis selec ed he
ROIs manually on he olun ee RGB maps ( he way DTI-ALPS-index
was in oduced9).
C) MNI ROIs in DWI space: he MNI-ROIs om A) we e ans o med o
he DWI space using F-1 by a cus om algo i hm, which was applied o
e ain he s uc u al in eg i y o he ROIs. This algo i hm ensu es ha
he ROIs emain squa e, wi h a oxel size o 2x2x2 oxels
(co esponding o 4x4x4 mm3). I ans o ma ion esul s in mo e han
wo oxels in a gi en di ec ion, he wo oxels wi h he highes
in ensi y a e e ained. I only one oxel emains, he missing oxel is
de e mined by selec ing he neighbou ing oxel wi h he highes FA
alue (no el echnique o his wo k).
In he ollowing, he ROIs de ined by he h ee di e en ROI placemen s ‘MNI
ROIs’, ‘Manual ROIs in DWI space’ and ‘MNI ROIs in DWI space’ a e e e ed o as
‘ROI-op ions’.
Fo each subjec and in each o he ou ALPS- ib e- egions, he indi idual
dis ance be ween he cen es o mass (Δ) o ROI-op ions B and C in DWI space was
calcula ed, ollowed by he mean and s anda d de ia ion ac oss subjec s.
Angles and ALPS-LAB/ ALPS-PAS
The o ien a ion o he di usion enso was desc ibed by in insic Ka dan-
angle o a ions o each ROI- oxel and as mean o all oxels in one ROI (ZY’X’’ wi h
he angles Ψ, Θ, Φ, Suppo ing Figu e S1.1, Suppo ing Figu e S1.3).
Simila ly, he absolu e angles be ween he labo a o y axes and he p incipal
axes o he di usion enso s we e calcula ed (Suppo ing Figu e S1.2).
ALPS-LAB and ALPS-PAS we e calcula ed o each subjec and each
hemisphe e (Equa ion 2, Equa ion 4), o all ROI-op ions.
S a is ical e alua ion
The mean ALPS-LAB/ ALPS-PAS and ela i e s anda d de ia ion (coe icien
o a iance, CV) was compu ed o all subjec s and all ROI-op ions (A, B, C).
Two pai ed - es s enabled a compa ison o ALPS-LAB and ALPS-PAS o each
ROI-op ion independen ly. To inc ease es alidi y, indices o bo h b ain
hemisphe es we e combined (signi icance p = 0.05; wo-sided and one-sided
(al e na i e hypo hesis H1: ALPS-LAB<ALPS-PAS)).
ALPS-maps
To gene a e a map o ALPS-LAB and ALPS-PAS o he en i e b ain, he indices
we e calcula ed o each oxel whe e FA was unequal o ze o. Ins ead o compa ing
he di usi i y o ROIs placed in he wo ALPS- ib e- egions, he second and hi d
la ges di usion componen s wi hin one oxel we e used, simila o he simula ed
enso abo e.
Discussion
We in oduced no el me hods o make he DTI-ALPS-index less suscep ible o
indi idual head and ib e o ien a ions, less suscep ible o indi idual ROI selec ion,
as well as a me hod o calcula e and isualise a DTI-ALPS- ela ed-index o he
en i e b ain.
Simula ing he dependency o he DTI-ALPS-index on o a ions indica ed ha a
misma ch be ween b ain ana omy and scanne coo dina e sys em can ha e a non-
negligible impac on he DTI-ALPS-index. These indings we e con i med in i o,
whe e he DTI-ALPS-index was ound o de ia e up o 20% depending on he
alignmen o ib es, desc ibed by he Ka dan-angles. This e ec was appa en in he
ROIs in he DWI space (ROI-op ions B and C), bu also a e ans o ma ion o he
da a in o MNI space (ROI-op ion A), indica ing ha his ans o ma ion alone may
no be enough o co ec o a ying head and ib e o ien a ions in he b ain.
In p inciple, his a iabili y should be co ec ed by e alua ing DTI-ALPS-index along
he p incipal di usion di ec ions (ALPS-PAS) ins ead o he scanne di ec ions.
Indeed, he ALPS-PASs we e s a is ically signi ican ly highe han all ALPS-LABs in
his s udy. Likely, his e ec can be a ibu ed o ib e-bundles, which a e no
o ien ed along he labo a o y ame. In addi ion, es - e es measu emen s showed
highe ep oducibili y o ALPS-PAS han ALPS-LAB. Thus, ALPS-PAS appea s o be
mo e bene icial han ALPS-LAB.
I should be no ed, hough, ha nei he ALPS-PAS no ALPS-LAB necessa ily
p oo he exis ence o glympha ic low in PVS, no he o ien a ion o PVS along he
second axis o he PAS. The ALPS-PAS calcula ion is s ill based on he DTI app oach,
which a e ages all issue di usi i ies in o one enso . This ques ion should be
in es iga ed wi h addi ional ials in he u u e. S ill, a measu able aniso opy
be ween he second and hi d p incipal di usion di ec ion exis s, no only in he
classical ALPS- ib e- egions, bu as well, when calcula ed oxel-wise, in many o he
egions o he b ain (Figu e 5). These maps may sugges mo e egions whe e DTI-
ALPS can be e alua ed.
Manual placemen o he ROIs (ROI-op ion B) is ope a o depended on and hus
ano he sou ce o a iabili y. Using s anda dised, a las-based ROIs (ROI-op ion C)
appea s o be a sui able solu ion, bu equi es ei he ans o ming he images o an
a las, o he a las-based ROIs o he na i e images.
E alua ing he DTI-ALPS-index in he MNI space, compa ed o DWI space, educed
he mean alue and hus he measu ed DTI-ALPS e ec o “seconda y aniso opy”
signi ican ly o ALPS-LAB. Using PAS educed he e ec so ha he di e ence was
no longe s a is ically signi ican . These indings may be explained by he ac ha
ans o ming he DWI da a o an a las space (he e: MNI) is a qui e in asi e
ope a ion, whe e many ana omical ea u es and hus ib e o ien a ions a e al e ed
(Figu e 1C). Using he PAS appea s o compensa e o his e ec pa ially.
T ans o ming he a las-based ROIs o he na i e imaging (DWI) space did no
signi ican ly change he indices, nei he o PAS no LAB. He e, he mean dis ance
be ween he manually placed and au oma ically placed ROIs was abou 3 mm.
These indings sugges ha i is bene icial (wi h espec o e ec size) o calcula e
he DTI-ALPS-index in he na i e imaging space on s anda dised ROIs ( ha we e
ans o med in o he space om an a las (ROI-op ion C)), a he han o ans o m
he whole imaging da a o he a las space (whe e he s anda dised ROIs we e
de ined (ROI-op ion A)). The conduc ed es - e es measu emen s suppo his idea.
Thus, calcula ing he DTI-ALPS-index in s anda dised ROIs ans o med om an
a las in he imaging space may allow o educe ope a o a iabili y.
The oxel-wise calcula ion o DTI-ALPS o e s an in e es ing app oach o e alua e
and isualise DTI-ALPS no only in selec ed egions, bu he en i e b ain. He e, he
classical ALPS- ib e- egions as well as o he a eas showed a s ong seconda y
aniso opy. No ably, ana omical ea u es we e subjec i ely much mo e consis en
o PAS han o LAB. Whe he o no he DTI-ALPS-index is o in e es has o be
e alua ed and emains o be seen.
Conclusion
Using he p incipal di usion di ec ions (PAS) ins ead o he scanne di ec ions
(LAB) o calcula e he DTI-ALPS-index inc eased i s alue by up o 20% in his
s udy. E alua ing bo h indices (ALPS-LAB and ALPS-PAS) in s anda dised ROIs
ans o med om an a las o he imaging space was no ound o be in e io o
manual ROI placemen bu elimina es po en ial in e -ope a o a iabili y. Toge he ,
bo h ALPS-PAS and s anda dised ROIs inc eased he DTI-ALPS e ec and made i
mo e obus . Whole b ain ALPS-LAB-map and ALPS-PAS-map show in e es ing
ea u es which will ha e o be in es iga ed sepa a ely.
ACKNOWLEDGMENTS
We since ely hank D . Tho s en Feiweie o e iewing his pape and p o iding
aluable and cons uc i e eedback ha helped imp o e he inal e sion.
Financial disclosu e
None epo ed.
Con lic o in e es
The au ho s decla e no po en ial con lic o in e es s. Ali Ajouz is an employee o
Siemens Heal hinee s AG and UKSH.
Suppo ing in o ma ion
Subjec s
Side
MNI ROIs
Manual ROIs in DWI
( olun ee ) space
MNI ROIs in DWI
( olun ee ) space
ALPS-
LAB
ALPS-
PAS
ALPS-
LAB
ALPS-
PAS
ALPS-
LAB
ALPS-
PAS
Sub1 ( , 24)
L
Re es
a io
1.60
1.40
114.3%
1.82
1.83
99.5%
1.71
1.32
129.5%
1.93
1.52
126.9%
1.70
1.34
126.9%
1.94
1.71
113.4%
R
Re es
a io
1.61
1.53
105.2%
1.72
1.71
100.5%
1.68
1.82
92.3%
1.72
1.69
101.7%
1.54
1.44
106.9%
1.71
1.65
103.6%
Sub2 (m, 29)
L
1.49
1.64
1.72
1.88
1.50
1.62
R
1.55
1.65
1.60
1.70
1.64
1.75
Sub3 (m, 27)
L
Re es
a io
1.56
1.56
100%
1.72
1.73
99.4%
1.63
1.59
102.5%
1.76
1.73
101.7%
1.64
1.64
100%
1.79
1.80
99.4%
R
Re es
a io
1.36
1.36
100%
1.43
1.44
99.3%
1.28
1.49
85.9%
1.35
1.59
84.9%
1.41
1.37
102.9%
1.46
1.46
100%
Sub4 ( , 23)
L
1.62
1.85
1.69
1.83
1.69
1.92
R
1.55
1.63
1.62
1.63
1.59
1.68
Sub5 (m, 27)
L
1.51
1.72
1.60
1.78
1.67
1.86
R
1.41
1.48
1.56
1.60
1.35
1.40
Sub6 ( , 49)
L
1.55
1.97
1.71
2.14
1.51
1.96
R
1.71
1.85
1.71
1.90
1.73
1.90
Sub7 ( , 26)
L
1.39
1.74
1.53
1.73
1.42
1.71
R
1.33
1.53
1.35
1.41
1.43
1.60
Sub8 ( , 26)
L
1.50
1.68
1.42
1.63
1.52
1.72
R
1.56
1.75
1.54
1.67
1.57
1.80
Sub9 ( , 30)
L
Re es
a io
1.39
1.40
99.2%
1.65
1.65
100%
1.46
1.61
90.7%
1.66
1.78
93.2%
1.37
1.42
96.4%
1.64
1.64
100%
R
Re es
a io
1.34
1.30
103.1%
1.52
1.51
100.6%
1.51
1.80
83.9%
1.74
1.96
88.8%
1.37
1.32
103.8%
1.60
1.49
107.4%
Mean
L
1.51
1.75
1.61
1.82
1.56
1.80
R
1.49
1.62
1.54
1.64
1.51
1.66
Rela i e
mean [%]
L
100
116
100
113
100
115
R
100
109
100
106
100
110
CV ( ela i e
o mean) [%]
L
5.10
5.76
6.68
8.07
7.36
6.84
R
8.55
8.10
8.76
9.71
8.17
9.07
Suppo ing Table S1: Fo bo h, DWI (Manual ROIs in DWI ( olun ee ) space, MNI ROIs in DWI ( olun ee )
space) and MNI (a las) space, he DTI-ALPS-index is calcula ed in he labo a o y ame (ALPS-LAB) and in he
p incipal axis sys em (ALPS-PAS) o each b ain side (Le -L, Righ -R). Subjec in o ma ion is gi en in he i s
column including he gende ( emale- , male-m) and age in yea s. Fo subjec s one, h ee and nine, ALPS-indices
o a e es measu emen a e ep esen ed in an addi ional line o each b ain hemisphe e. The line ‘ a io’ de ines
he es -index di ided by he e es -index in pe cen . In addi ion, he mean, ela i e mean and he coe icien o
a iance (CV) ac oss subjec s, excluding he e es -measu emen s, a e included.
Suppo ing Figu e S1: Illus a ion o an exempla y di usion enso om a oxel in he SCR a ea o a olun ee
wi h eigen alues in [µm2/ms] 𝜆1=1.06, 𝜆2=0.54 , 𝜆3=0.35 and eigen ec o s 𝑒1 = (-0.93, -0.34, 0.04)T, 𝑒2 =
(0.29, -0.84, -0.44)T, 𝑒3= (-0.19, 0.39, -0.89)T. The ans o ma ion be ween he Ca esian labo a o y axes and he
eigen ec o s (o p incipal axis) o he di usion enso can be desc ibed by h ee subsequen Ka dan- o a ions Z,
Y’, X’’ (3a-c). The minimal (absolu e) angels be ween he PAS and labo a o y ame a e shown in 2).
Suppo ing Figu e S2: (A) In addi ion o Figu e 3.1, he ela ion o ALPS-LAB o ALPS-PAS o mul iple Ka dan-
angle combina ions wi h a simul aneous change o all angles is plo ed. (B) Simila o (A) bu se ing wo
Ka dan-angles o ze o while a ying he hi d in he en i e Ka dan-angle ange.
Suppo ing Figu e S3: The cu es ep esen he mean and s anda d de ia ion alues o he h ee absolu e
angles compu ed o each oxel wi hin he ROIs, which a e de ined pe ALPS- ib e- egion. Fo each angle, he
mean and s anda d de ia ion we e calcula ed ac oss oxels wi hin each ROI o each subjec . Subsequen ly,
hese alues we e a e aged ac oss subjec s. The esul s shown a e o he igh hemisphe e as an illus a ion.
The colou s blue, g een and ed cha ac e ize(𝜶,𝜷,𝜸) in DWI space (A1,2: Manual ROIs in DWI space; B1,2: MNI
ROIs in DWI space) and ligh blue, ligh g een and ligh pu ple cha ac e ize (𝜶,𝜷,𝜸) in MNI space (MNI ROIs).
Suppo ing Figu e S4: Exempla y RGB maps in MNI space (uppe ) wi h MNI ROIs (whi e squa es ROI-op ion A)
and in DWI space (down) wi h Manual ROIs in DWI space ( ed squa es ROI-op ion B) and MNI ROIs in DWI
space (whi e squa es ROI-op ion C) o subjec one, wi h ( igh ) and wi hou (le ) head o a ion. In all cases, he
algo i hm placed he ROIs (ROI-op ion C) in ana omically meaning ul a eas (g een and blue a eas). No e he
smoo hing e ec o ans o ming he da a o he MNI space.
ORCID
Ali Ajouz 0009-0006-2251-0361
Ola Jansen 0000-0002-7330-1942
Lynn Johann F ohwein 0000-0003-3474-115X
S ea Seeha e 0009-0001-2573-5770
Naomi La sen 0000-0002-8782-8278
Jan-Be nd Hö ene 0000-0001-7255-7252
Re e ences
1. Ili JJ, Wang M, Liao Y, e al. A Pa a ascula Pa hway Facili a es CSF Flow Th ough
he B ain Pa enchyma and he Clea ance o In e s i ial Solu es, Including Amyloid β. Sci
T ansl Med. 2012;4(147). doi:10.1126/sci anslmed.3003748
2. Boh T, Hjo h PG, Hols SC, e al. The glympha ic sys em: Cu en unde s anding
and modeling. iScience. 2022;25(9):104987. doi:10.1016/j.isci.2022.104987
3. Smi h AJ, Ve kman AS. The “glympha ic” mechanism o solu e clea ance in
Alzheime ’s disease: game change o unp o en specula ion? FASEB j. 2018;32(2):543-551.
doi:10.1096/ j.201700999
4. Kamaga a K, Sai o Y, Andica C, e al. Nonin asi e Magne ic Resonance Imaging
Measu es o Glympha ic Sys em Ac i i y. Magne ic Resonance Imaging. 2024;59(5):1476-
1493. doi:10.1002/jm i.28977
5. Taoka T, Naganawa S. Glympha ic imaging using MRI. Magne ic Resonance Imaging.
2020;51(1):11-24. doi:10.1002/jm i.26892
6. Joseph CR, Benha zel CM, S e n LJ, Hoppe OM, Lockwood MD. Pilo s udy u ilizing
MRI 3D TGSE PASL (a e ial spin labeling) di e en ia ing clea ance a es o labeled p o ons
in he CNS o pa ien s wi h ea ly Alzheime disease om no mal subjec s. Magn Reson Ma e
Phy. 2020;33(4):559-568. doi:10.1007/s10334-019-00818-3
7. Chen Y, Dai Z, Fan R, e al. Glympha ic Sys em Visualized by Chemical-Exchange-
Sa u a ion-T ans e Magne ic Resonance Imaging. ACS Chem Neu osci. 2020;11(13):1978-
1984. doi:10.1021/acschemneu o.0c00222
8. Gomolka RS, Habli z LM, Mes e H, e al. Loss o aquapo in-4 esul s in glympha ic
sys em dys unc ion ia b ain-wide in e s i ial luid s agna ion. eLi e. 2023;12:e82232.
doi:10.7554/eLi e.82232
9. Taoka T, Masu ani Y, Kawai H, e al. E alua ion o glympha ic sys em ac i i y wi h
he di usion MR echnique: di usion enso image analysis along he pe i ascula space
(DTI-ALPS) in Alzheime ’s disease cases. Jpn J Radiol. 2017;35(4):172-178.
doi:10.1007/s11604-017-0617-z
10. Taoka T, I o R, Nakamichi R, e al. Rep oducibili y o di usion enso image analysis
along he pe i ascula space (DTI-ALPS) o e alua ing in e s i ial luid di usi i y and
glympha ic unc ion: CHanges in Alps index on Mul iple condi iON acquIsi ion eXpe imen
(CHAMONIX) s udy. Jpn J Radiol. 2022;40(2):147-158. doi:10.1007/s11604-021-01187-5
11. Taoka T, I o R, Nakamichi R, Nakane T, Kawai H, Naganawa S. Di usion Tenso
Image Analysis ALong he Pe i ascula Space (DTI-ALPS): Re isi ing he Meaning and
Signi icance o he Me hod. MRMS. 2024;23(3):268-290. doi:10.2463/m ms. e .2023-0175
12. Ta ekawa H, Ma sushi a S, Ueda D, e al. Imp o ed ep oducibili y o di usion enso
image analysis along he pe i ascula space (DTI-ALPS) index: an analysis o eo ien a ion
echnique o he OASIS-3 da ase . Jpn J Radiol. 2023;41(4):393-400. doi:10.1007/s11604-
022-01370-2
13. Sai o Y, Kamaga a K, Andica C, e al. Rep oducibili y o au oma ed calcula ion
echnique o di usion enso image analysis along he pe i ascula space. Jpn J Radiol.
2023;41(9):947-954. doi:10.1007/s11604-023-01415-0
14. Ajouz A. Impac o image egis a ion and ib e o ien a ion on he DTI-ALPS-index.
In: 26 h Ann. Mee . Ge . Chap e In . Soc. Magn. Reson. Med. 2024, Tübingen. Tübingen,
Ge many; 2024:129-130. doi:h ps://plan.e en s.mpg.de/e en /204/o e iew
15. Ajouz A. Using a p inciple axis sys em allows au oma ic e alua ion o glympha ic
low wi h ALPS in imaging space and co ec s o head and ib e o ien a ion. In: 2025
ISMRM & ISMRT Annual Mee ing & Exhibi ion. Honolulu, Hawai’i, USA; 2025.
h ps://www.ism m.o g/25/accep ed_abs ac s.pd .
16. Jenkinson M, Beckmann CF, Beh ens TEJ, Wool ich MW, Smi h SM. FSL. Neu oImage.
2012;62(2):782-790. doi:10.1016/j.neu oimage.2011.09.015
17. Basse PJ, Ma iello J, LeBihan D. MR di usion enso spec oscopy and imaging.
Biophysical Jou nal. 1994;66(1):259-267. doi:10.1016/S0006-3495(94)80775-1
18. Yang F, Zhu YM, Kingsley PB. A u he in es iga ion o he Eule angle calcula ion in
di usion enso imaging. In: 2016 IEEE 13 h In e na ional Con e ence on Signal P ocessing
(ICSP). Chengdu, China: IEEE; 2016:39-43. doi:10.1109/ICSP.2016.7877792
19. A an s B, Tus ison NJ, Song G. Ad anced No maliza ion Tools: V1.0. The Insigh
Jou nal. July 2009. doi:10.54294/u nhin
20. Tou nie JD, Smi h R, Ra el D, e al. MR ix3: A as , lexible and open so wa e
amewo k o medical image p ocessing and isualisa ion. Neu oImage. 2019;202:116137.
doi:10.1016/j.neu oimage.2019.116137
