Human en icula ac i a ion sequence and he
simula ion o he elec oca diog aphic QRS
complex and i s a iabili y in heal hy and
in a en icula block condi ions
Louie Ca done-Noo
1,†
, Al onso Bueno-O o io
1,
*
,†
, Ana Minchole´
1
,
Nejib Zemzemi
2,3
, and Blanca Rod iguez
1
1
Depa men o Compu e Science and B i ish Hea Founda ion Cen e o Resea ch Excellence, Uni e si y o Ox o d, Ox o d OX1 3QD, UK;
2
INRIA Bo deaux Sud-Oues ,
200 a enue de la ieille ou , Talence Cedex 33405, F ance; and
3
IHU Li yc, Elec ophysiology and Hea Modeling Ins i u e, ounda ion Bo deaux Uni e si e´, F-33600 Pessac
Bo deaux, F ance
Recei ed 15 Ap il 2016; accep ed a e e ision 9 Augus 2016
Aims To in es iga e how a iabili y in ac i a ion sequence and passi e conduc ion p ope ies ansla es in o clinical a i-
abili y in QRS bioma ke s, and gain no el physiological knowledge on he in o ma ion con ained in he human QRS
complex.
............................ ............. ............. .................. ............. ............. ............................... ............. ............. .................. ............. .........
Me hods
and esul s
Mul iscale bidomain simula ions using a de ailed hea - o so human ana omical model a e pe o med o in es iga e
he impac o ac i a ion sequence cha ac e is ics on clinical QRS bioma ke s. Ac i a ion sequences a e buil and
alida ed agains expe imen ally-de i ed ex i o and in i o human ac i a ion da a. R-peak ampli ude exhibi s he
la ges a iabili y in e ms o QRS mo phology, due o i s simul aneous modula ion by ac i a ion sequence speed,
myoca dial in acellula and ex acellula conduc i i ies, and p opaga ion h ough he human o so. QRS wid h,
howe e , is egula ed by endoca dial ac i a ion speed and in acellula myoca dial conduc i i ies, whe eas QR
in e als a e only a ec ed by he endoca dial ac i a ion p o ile. Va iabili y in he apico-basal loca ion o ac i a ion
si es on he an e io and pos e io le en icula wall is associa ed wi h S-wa e p og ession in limb and p eco dial
leads, espec i ely, and occasional no ched QRS complexes in p eco dial de i a ions. Va iabili y in he numbe o
ea ly ac i a ion si es success ully ep oduces pa hological abno mali ies o he human conduc ion sys em in he
QRS complex.
............................ ............. ............. .................. ............. ............. ............................... ............. ............. .................. ............. .........
Conclusion Va iabili y in ac i a ion sequence and passi e conduc ion p ope ies cap u es and explains a la ge pa o he clinical
a iabili y obse ed in he human QRS complex. Ou physiological insigh s allow o a deepe in e p e a ion o
human QRS bioma ke s in e ms o QRS mo phology and loca ion o ea ly endoca dial ac i a ion si es. This migh
be used o a ain a be e pa ien -speci ic knowledge o ac i a ion sequence om ou ine body-su ace
elec oca diog ams.
䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏䊏
Keywo ds Elec oca diog am •QRS complex •Ac i a ion sequence •Va iabili y •Compu e modelling and simula ion
†The i s wo au ho s con ibu ed equally o he s udy.
*Co esponding au ho . Tel: þ44 1865 610737; Fax: +44 1865 273839. E-mail add ess: [email p o ec ed]
V
CThe Au ho 2016. Published by Ox o d Uni e si y P ess on behal o he Eu opean Socie y o Ca diology.
This is an Open Access a icle dis ibu ed unde he e ms o he C ea i e Commons A ibu ion License (h p://c ea i ecommons.o g/licenses/by/4.0/), which pe mi s
un es ic ed euse, dis ibu ion, and ep oduc ion in any medium, p o ided he o iginal wo k is p ope ly ci ed.
Eu opace (2016) 18, i 4–i 15 SUPPLEMENT PAPER
doi:10.1093/eu opace/euw346
Downloaded om h ps://academic.oup.com/eu opace/a icle/18/suppl_4/i 4/2737164 by Uni e sidad de Za agoza. Biblio eca Uni e si a ia use on 26 Feb ua y 2021
In oduc ion
Since i s in en ion by Ein ho en a he beginning o he 20 h cen u y,
he body-su ace elec oca diog am (ECG) emains as he mos ex-
ensi ely used clinical ool o he non-in asi e diagnosis o ca diac
diso de s. A speci ic challenge is o disc imina e be ween he e ec s
on he ECG bioma ke s o he di e en componen s o he en icu-
la ac i a ion and epola iza ion sequences (in ol ing simul aneous
ansmu al, apico-basal, pos e io -an e io , and in e - en icula
p opaga ion), and/o he con ibu ion o concu en pa hological
s a es. This hampe s ou abili y o e ec i ely ex ac he maximum
in o ma ion abou he hea om he ECG.
In his con ex , mul iscale human hea models a e powe ul pla -
o ms o in eg a e elec ophysiological and s uc u al in o ma ion
om he ionic o he whole o gan le els, and o di e en ia e be ween
key ac o s de e mining ECG bioma ke s in disease
1
o unde
pha macological ac ion.
2
Such an imp o ed knowledge may u he
aid in he de elopmen o mo e selec i e bioma ke s o speci ic dis-
eased condi ions o d ug-induced isk s a i ica ion.
3
Mos compu a-
ional s udies o da e, howe e , ha e mainly concen a ed on
eplica ing he epola iza ion sequence o he human en icles and
placed less a en ion on he ac i a ion sequence, o en esul ing in
poo ly- eco e ed QRS complex mo phology and lead pola i ies (see
o example
1,2
) This highligh s he need o accu a e ep esen a ions
o he human ac i a ion sequence in ECG simula ions, due o i s di -
ec in luence in de e mining epola iza ion, and o p ope ly add ess
he compu a ional s udy o ca diac conduc ion diso de s.
In his wo k, we in es iga e he e ec o a iabili y in cha ac e is ics
o he human en icula ac i a ion sequence in he simula ed QRS
complex using a human o so-hea bidomain model. The simula ed ac-
i a ion sequences a e buil and alida ed using expe imen ally-de i ed
human ac i a ion da a, anging om ex i o mic oelec ode eco dings,
non-in asi e in i o elec omechanical wa e and elec oca diog aphic
imaging, up o he body-su ace ECG. Quan i a i e in es iga ions on
a iabili y in ac i a ion sequence and passi e conduc ion p ope ies a e
hen p esen ed o gain no el insigh s on he impac o issue-le el
p opaga ion in clinical QRS bioma ke s, as well as o demons a e he
easibili y o ou app oach o model pa hophysiology o he human
conduc ion sys em. Ou esul s can he e o eha eimpo an implica-
ions in ad ancing he ealis ic modelling o ca diac conduc ion unde
heal hy and diseased condi ions, and aid in un a elling he ole o a i-
abili y in modula ing esponse o he apy a he popula ion le el.
Me hods
Expe imen al da a
The p esen in es iga ions build on h ee main sou ces o human expe i-
men al da a. Fi s ly, en icula ac i a ion sequences we e based on ex i o
mic oelec ode eco dings by Du e e al.
4
(Figu e 1A), in i o elec o-
mechanical wa e imaging by P o os e al.
5
(Figu e 1B), and in i o epica -
dial ac i a ion sequences as epo ed by Ramana han e al.
6
h ough
non-in asi e elec oca diog aphic imaging (Figu e 1C). Secondly, dis ibu-
ions o body-su ace po en ials (BSPs) du ing en icula ac i a ion in
heal hy condi ions we e analysed om Tacca di e al.
7
(Figu e 1D).
Thi dly, 12-lead body-su ace ECG signals (Figu e 1E) in heal hy and bundle
b anch block condi ions we e ob ained om he PTB Diagnos ic ECG
Da abase,
8
eely a ailable in Physione (www.physione .o g), whils
hemiblock ECGs (no p esen in he PTB Diagnos ic ECG Da abase)
we e analysed om Eliza i e al.
9
Ana omical hea - o so model
A human bi en icula mesh, embedded in a o so olume con aining
lung and bone egions, was gene a ed om compu e omog aphy
images as desc ibed in Supplemen a y Me hods. Vi ual elec odes we e
posi ioned on he o so a s anda d elec ode loca ions o he calcula-
ion o he 12-lead ECG (Supplemen a y ma e ial online, Figu e S1).
Elec ophysiological and issue model
Compu e simula ions we e conduc ed using he ully coupled hea -
o so bidomain equa ions, which a e he gold-s anda d o he desc ip-
ion o ca diac elec ical p opaga ion, including de ailed desc ip ion o
human en icula cellula elec ophysiology. Myoca dial and o so con-
duc i i ies we e based on he li e a u e, as de ailed in Supplemen a y
Me hods, oge he wi h speci ica ions on ou nume ical sol e . Resea ch
ma e ials a e a ailable upon eques .
Ac i a ion sys em models
As in Kelle e al.
10
and due o he a iabili y and lack o high- esolu ion
da a on which o build ana omically-de ailed models o he human ee-
unning Pu kinje ne wo k, we assume ha i s b anches couple wi h and ex-
ci e he endoca dial laye a se e al si es o ea lies ac i a ion ( oo
poin s) om which exci a ion quickly p og esses. We base he posi ions
o hese oo poin s on he no ion o he human i ascicula conduc ion
sys em
11
(le an e io -supe io , le pos e io -in e io and igh la e al
ascicula b anches). To model he igh ly-packed endoca dial Pu kinje ne -
wo k, pai wise dis ances be ween all endoca dial su ace nodes we e p e-
p ocessed using Dijks a’s algo i hm. A un ime, each endoca dial su -
ace node is assigned a s imulus ime p opo ional o he dis ance o i s
closes oo node. The cons an o p opo ionali y allows a s aigh o -
wa d con ol o conduc ion eloci y in he abs ac ed endoca dial
Pu kinje ne wo k. This was adjus ed o yield a dis ibu ion o endoca dial
Wha ’s New?
•Mul iscale human hea - o so models wi h ana omical and
unc ional de ail based on mul imodal human ac i a ion da a
a e p esen ed o in es iga ions on a iabili y in clinical QRS
bioma ke s.
•Ac i a ion sequence, myoca dial conduc i i ies and o so
p opaga ion ha e dis inc e ec s in modula ing R-peak ampli-
ude, QRS wid h and QR in e al.
•Va iabili yin heana omicalloca ion o le en icula ac i a-
ion si es is associa ed wi h S-wa e p og ession in limb and
p eco dial leads, and no ched QRS complexes in p eco dial
de i a ions.
•Va iabili yin henumbe o ea lyac i a ionsi es eplica es he
clinical QRS mani es a ions o all hemiblocks and bundle
b anch blocks o he human conduc ion sys em.
•The ools p esen ed he e ad ance he ield o compu a ional
ca diac elec ophysiology owa ds mo e p edic i e echnolo-
gies o isk assessmen a he popula ion le el, as well as
o he be e unde s anding o clinical ECG bioma ke s o
disease.
Simula ion o he human QRS complex i 5
Downloaded om h ps://academic.oup.com/eu opace/a icle/18/suppl_4/i 4/2737164 by Uni e sidad de Za agoza. Biblio eca Uni e si a ia use on 26 Feb ua y 2021
Figu e 1 Non-diseased human en icula ac i a ion da a conside ed in his s udy. (A)Ex i o mic oelec ode eco dings o ac i a ion sequence in
a heal hy subjec . (B)Non-in asi ein i o elec omechanic wa e imaging isoch ones o h ee no mal subjec s. A ows indica e en icula si es o
ea ly ac i a ion. (C) Non-in asi e in i o elec oca diog aphic imaging econs uc ion o epica dial ac i a ion sequence in h ee heal hy subjec s.
(D) Body su ace po en ial maps in a heal hy subjec du ing p og ession o en icula ac i a ion. (E) QRS complexes in he 12-lead ECG o a ep e-
sen a i e heal hy indi iduals o he PTB Diagnos ic ECG Da abase
8
(subjec 198). Clinical ECG g id esolu ion: 40ms/0.1 mV. Panels A–D ep oduced
wi h pe mission om Du e e al.,
4
P o os e al.,
5
Ramana han e al.
6
and Tacca di,
7
espec i ely.
i 6 L. Ca done-Noo e al.
Downloaded om h ps://academic.oup.com/eu opace/a icle/18/suppl_4/i 4/2737164 by Uni e sidad de Za agoza. Biblio eca Uni e si a ia use on 26 Feb ua y 2021
ac i a ion imes in acco dance wi h he epo ed ex i o mic oelec ode
eco dings by Du e e al.
4
Resul s
Model cons uc ion and alida ion
Human en icula ac i a ion da a exhibi a high le el o in e -subjec
a iabili y in he si es o ea lies ac i a ion and epica dial b eak-
h ough, as illus a ed in he mul i-modali y da a compa ison
summa ized in Figu e 1. The ollowing commonali ies we e howe e ,
iden i ied ac oss he s udied human da ase s: (i) ac i a ion wi hin he
le en icle (LV) usually begins on he basal an e io pa asep al, he
mid-sep um and he pos e io apex egions, and hen p og esses
ansmu ally (Figu e 1AandB); (ii) ea lies ac i a ion in he ee wall o
he igh en icle (RV) occu s ‘nea he inse ion o he an e io pap-
illa y muscle’,
4
he posi ion o which is known o be highly a iable
(Figu e 1B and C); and (iii) ac i a ion o he RV epica dium usually
occu s be o e ha o he LV epica dium (Figu e 1BandC).
Based on he abo e, we designed a baseline con igu a ion o he
human ac i a ion sequence (see Figu e 2A, black a ows) consis ing o
ou LV ea lies ac i a ion si es (LV mid sep um, LV basal an e io
pa asep al, and wo LV mid-pos e io ) and h ee in he RV (RV mid
sep um, wo RV ee wall). Such a con igu a ion allows o a mul i-
scale in es iga ion (up o he body-su ace ECG) o in e -subjec a i-
abili y in ac i a ion speed and passi e issue conduc i i ies, as well as
in he posi ion and numbe o ea lies ac i a ion si es, wi hou a ou -
ing a speci ic da a modali y in ou analysis.
Simula ed en icula ac i a ion imes o he baseline con igu -
a ion a e p esen ed in Figu e 2A. The LV endoca dial su ace is ully
ac i a ed wi hin 30 ms, ansmu al p opaga ion occu s in 35 ms in he
LV basal ee wall, wi h anssep al p opaga ion o 25 ms. This is in ac-
co dance wi h he ex i o mic oelec ode eco dings by Du e e al.
4
(Figu e 1A), epo ing 35 ms in endoca dial LV ac i a ion, and 25 and
35 ms in anssep al and ansmu al p opaga ion, espec i ely. Sligh ly
sho e ac i a ion imes a e ob ained in he RV base in simula ions,
possibly due o unce ain ies in he a ia- en icula segmen a ion.
C oss-sec ional slices o en icula ac i a ion imes a e shown in
Figu e 2B o compa ison wi h in i o elec omechanical wa e imaging
da a by P o os e al.
5
(Figu e 1B), indica ing ag eemen in he loca ion
o ea lies and la es ac i a ion si es. Speci ically, ea ly ac i a ion
occu ed on ei he side o he sep um, RV ee wall, and an e io and
pos e io LV pa asep al si es. La es ac i a ion ook place in he basal
sep um, basal la e al LV wall, and apex.
An e io and pos e io iews o epica dial ac i a ion imes a e p o-
ided in Figu e 2C o ease o compa ison wi h he in i o elec oca -
diog aphic imaging da a o Ramana han e al.
6
Simula ion esul s a e in
close ag eemen wi h Subjec s 1 and 3 in Figu e 1C, wi h ea lies
b eak h ough in he uppe pa asep al an e io RV, apical LV ac i a-
ion 20 ms la e , and basal LV ac i a ion 28 ms a e ha . The ac i a-
ion o he RV epica dium comple ed in 30 ms, compa ed o 25 ms in
heal hy subjec s. The LV epica dial b eak h ough occu ed abou
15 ms a e he RV b eak h ough, consis en wi h bo h Subjec s 1
and 3. The pa e n o LV epica dial ac i a ion is well ep oduced due
o he la e al basal LV oo poin (leading o he epica dial b eak-
h ough on he LV an e io pa asep al egion) and he hickening o
he LV ee wall om apex o base (which gene a es ac i a ion imes
ha leng hen om apex o base).
Figu e 2Dshows isopo en ial maps o BSPs, exhibi ing he same
spa io empo al e olu ion du ing en icula ac i a ion as hose e-
po ed in heal hy subjec s by Tacca di e al.
7
(Figu e 1D). Ini ially a
local maximum appea s on he ches due o he i s RV epica dial
b eak h ough (10 ms). Once he ac i a ion o he RV is comple ed,
his is eplaced by a nega i e egion. The posi i e pole mo es le -
wa ds and downwa ds as ac i a ion p og esses ansmu ally h ough
he LV. This o ms he cha ac e is ic dipole pa e n on he on o
he ches wi h a ze o isopo en ial be ween he lowe - igh lank and
he le shoulde (20–30 ms). The maximum is inally d awn owa ds
he la es a ea o ac i a e, he pos e io pa asep al LV (40 ms).
As a consequence o he dis ibu ions o BSPs desc ibed abo e, a
on al plane QRS axis (a e age o global en icula depola iza ion)
be ween 30and þ90is conside ed o be no mal, whe e 0is le -
wa ds in he ans e se plane. This de e mines he limb (I, II, III) and
augmen ed (aVR, aVL, aVF) leads, and yields nega i e p eco dial leads
in V1, posi i e in V4–V6, wi h R-wa e p og ession in be ween, as
illus a ed in Figu e 1E o a ep esen a i e heal hy subjec . Fo com-
pa ison, Figu e 2Eillus a es he simula ed QRS complexes associa ed
wi h ou baseline ac i a ion sequence, exhibi ing a QRS wid h o
70 ms wi hin he no mal ange (60–100 ms) and a QRS axis o 60,
bo h wi hin he heal hy ange, and ypical lead pola i ies in all limb,
augmen ed limb and p eco dial leads, as well as R-wa e p og ession
in he la e .
Va iabili y in subendoca dial ac i a ion
speed and myoca dial conduc i i ies
modula e QRS wid h and ampli ude
De ailed simula ion s udies we e conduc ed o in es iga e how a i-
abili y in he di e en de e minan s o dis ibu ions o BSPs modu-
la es he human QRS complex. Whils he ex en o his a iabili y in
he human popula ion is di icul o de e mine,
12
we hypo hesize ha
a iabili y o 650% in ac i a ion sequence and issue p ope ies en-
capsula es a majo ac ion o he obse ed clinical a iabili y in he
heal hy human QRS complex. Changes on baseline QRS bioma ke s
a e hence p esen ed in he ollowing sec ions (a e ages epo ed
ac oss all leads) o hese bounds o plausible physiological a iabili y.
In ac , his could be a modes es ima e o o al a iabili y ac oss he
popula ion, since a iabili y bounds om 2 o 12- old ha e been e-
po ed in issue conduc i i ies in human,
13
including blood (2.3),
lungs (3.4), a (5.5), so muscle (7.5), hea (9.0) and bone
(12).
Va iabili y in endoca dial ac i a ion speed di ec ly impac s he ull
en icula ac i a ion sequence (Figu e 3A, le panel), a ec ing he
wid h and ampli ude o he econs uc ed QRS complexes. Inc eased
ac i a ion speeds (Figu e 3A, solid ed aces) esul ed in ea lie R-
wa e imes- o-peak, sho e QRS wid hs, and inc eased R-wa e
ampli udes (75 63%, 9064%, and 125 630% o baseline, espec -
i ely). Dec eased ac i a ion speeds had he opposi e e ec s (Figu e
3A, solid blue aces), exhibi ing p o ac ed R-wa e imes- o-peak,
wide QRS wid hs, oge he wi h dec eased R-wa e ampli udes
(17867%, 182 612%, and 8368% o baseline, espec i ely).
Va iabili y in in acellula myoca dial conduc i i ies (as modelled
by globally scaled conduc i i y enso s) subs an ially in luences he
Simula ion o he human QRS complex i 7
Downloaded om h ps://academic.oup.com/eu opace/a icle/18/suppl_4/i 4/2737164 by Uni e sidad de Za agoza. Biblio eca Uni e si a ia use on 26 Feb ua y 2021
Figu e 2 Simula ion esul s o he heal hy human en icula ac i a ion sequence. (A) Volume ic isoch ones o en icula ac i a ion imes.
A ows indica e ac i a ion sys em oo poin s. (B) C oss-sec ional slices o en icula ac i a ion imes. A ows indica e en icula si es o ea ly ac i-
a ion. (C) An e io and pos e io iews o epica dial ac i a ion imes. Di e en colou maps a e shown in Panels A–C o acili a e he compa ison
wi h expe imen al da a in Figu e 1. (D) Isopo en ial body su ace po en ial maps du ing p og ession o en icula ac i a ion. (E) QRS complexes in
he simula ed 12-lead ECG. Simula ed ECG g id esolu ion: 40ms/0.1 mV.
i 8 L. Ca done-Noo e al.
Downloaded om h ps://academic.oup.com/eu opace/a icle/18/suppl_4/i 4/2737164 by Uni e sidad de Za agoza. Biblio eca Uni e si a ia use on 26 Feb ua y 2021

Figu e 3 Va iabili y in ac i a ion speed and myoca dial conduc i i ies modula e QRS wid h and ampli ude. (A) Va iabili y in ac i a ion speed
impac s he en icula ac i a ion sequence, a ec ing R-wa e imes- o-peak, R-wa e ampli udes, and QRS wid h. (B) Va iabili y in in acellula
myoca dial conduc i i ies in luences he ansmu al dispe sion o en icula ac i a ion imes, yielding scaled R-wa e ampli udes and wide and less-
symme ic QRS complexes unde condi ions o dec eased in acellula coupling. (C) Va iabili y in ex acellula myoca dial conduc i i y yields a scaling
e ec s on QRS ampli udes. In all cases, dashed g ey ECG aces ep esen baseline condi ions, whe eas solid ed and blue ECG aces ep esen con-
di ions o inc eased and dec eased myoca dial pa ame e s, espec i ely. Simula ed ECG g id esolu ions: 40 ms/0.1mV.
Simula ion o he human QRS complex i 9
Downloaded om h ps://academic.oup.com/eu opace/a icle/18/suppl_4/i 4/2737164 by Uni e sidad de Za agoza. Biblio eca Uni e si a ia use on 26 Feb ua y 2021
ansmu al dispe sion o en icula ac i a ion imes (Figu e 3B,le
panel). This yielded a p edominan scaling e ec on R-wa e ampli-
udes (121 613% and 65612% o baseline o inc eased/dec eased
in acellula conduc i i ies, espec i ely), wi h a smalle impac on
imes o R-peak (95 63% and 10563%, espec i ely). Wide QRS
wid hs (12767%) and less symme ic QRS complexes we e also
associa ed wi h condi ions o dec eased in acellula coupling, in pa -
icula in he p eco dial leads (Figu e 3B, solid blue aces).
On he o he hand, a iabili y in ex acellula myoca dial conduc-
i i ies has a pu ely scaling ole on he QRS complex, wi h negligible
e ec s on he en icula ac i a ion sequence (Figu e 3C). Ac oss all
leads, peak R-wa e ampli udes we e 6966% and 183 625% o
baseline o a 50% inc ease and dec ease in ex acellula conduc i i y
enso s, espec i ely. Times o R-peak we e wi hin 2% in all cases,
wi h negligible changes o QRS wid hs.
The obse ed e ec s can be explained by ecalling a simpli ied
co e-conduc o model o elec ical p opaga ion in myoca dial
ibe s,
14
wi h ex acellula po en ials gi en by
ue ;xðÞ/
i
eðX
ui ðÞ 1

dX;
whe e u
e
is he ex acellula po en ial a ime and body su ace
poin x,
i
and
e
a e he in acellula and ex acellula conduc i i ies,
and he in eg al ep esen s he sum o he con ibu ion o all g adi-
en s in in acellula po en ials ( u
i
) in he olume o he hea (X),
weigh ed by hei espec i e dis ances o he measu e poin ( ).
Changes in ac i a ion speed he e o e a ec he ime cou se o endo-
ca dial and ansmu al g adien s in in acellula po en ials, has ening
o p o ac ing en icula depola iza ion as e lec ed in he QRS
wid h and imes o R-peak, while hei spa ial dis ibu ion (mo e
sp ead g adien s o la ge ac i a ion speeds) impac s R-wa e ampli-
udes (Figu e 3A). In con as , a iabili y in in acellula conduc i i ies
unde he imposi ion o he same endoca dial ac i a ion sequence
only modula es he ansmu al p opaga ion o in acellula po en ials,
hence impac ing he e minal pa o he QRS complex wi h mode -
a e in luence on imes o R-peak, oge he wi h a p opo ional scaling
o ex acellula po en ials (Figu e 3B). Co espondingly, ex acellula
po en ials a e in e sely p opo ional o ex acellula conduc i i ies,
modula ing R-wa e ampli udes wi h negligible e ec s on in acellula
g adien s (Figu e 3C). Changes in QRS wid h by myoca dial conduc i -
i ies can also be in e p e ed by conside ing he e ec i e conduc i i y
enso ,
m
¼
i
e
/(
i
þ
e
), and i s impac on myoca dial p opaga ion
speed, gi en by he squa e oo o he quo ien be ween al e ed and
baseline e ec i e conduc i i ies. Fo ou choice o issue p ope ies,
a ia ions o þ50% and 50% in in acellula conduc i i ies, espec -
i ely yield changes o þ20% and 27.5% in ansmu al p opaga ion
speed, ye only o þ1.7% and 4.6% o equi alen a ia ions in he
ex acellula enso , which he e o e explains he mino con ibu ion
o he la e o he wid h o he QRS complex.
Va iabili y in body conduc i i ies
modula es QRS ampli udes
Simila o he ole o ex acellula myoca dial conduc i i ies, a iabil-
i y in body conduc i i ies ( o so/lungs/bones) exe ed a scaling e ec
on QRS ampli udes by a ec ing he magni udes o BSPs. Resul s on
he e ec s o his a iabili y in o so conduc i i y a e illus a ed in
Figu e 4, as his exhibi ed he la ges con ibu ion in he human QRS
complex compa ed o lung and bone conduc i i ies (Supplemen a y
ma e ial online, Figu e S2).
Al hough he spa ial pa e ns o BSPs emained mos ly unal e ed
unde a ia ions o body conduc ion p ope ies, he la ge di usion
o ex acellula po en ials associa ed wi h la ge conduc i i ies
yielded BSPs wi h lowe magni udes a imes o R-peak, whils
Figu e 4 Va iabili y in body conduc i i ies modula es QRS ampli udes. Di e ences in he di usion o elec ical po en ial ac oss he human o so a -
ec he magni udes o BSPs, wi hou signi ican ly al e ing hei spa ial pa e n (le panel; BSP dis ibu ions shown a ime o R-peak in lead II).Thisis
ansla ed in in e sely p opo ional QRS ampli udes o body conduc i i ies, and inc eased R-wa e p og ession ( igh panel). Dashed g ey ECG aces:
baseline condi ions; solid ed and blue ECG aces: inc eased and dec eased o so conduc i i y, espec i ely. Symbols indica e maximum wa e ampli-
udes unde a iabili y in o so conduc i i y (same colou code). Simula ed ECG g id esolu ion: 40 ms/0.1 mV.
i 10 L. Ca done-Noo e al.
Downloaded om h ps://academic.oup.com/eu opace/a icle/18/suppl_4/i 4/2737164 by Uni e sidad de Za agoza. Biblio eca Uni e si a ia use on 26 Feb ua y 2021
smalle body conduc i i ies accen ua ed he magni ude o he
obse ed g adien s (Figu e 4, le panel). This was hus ansla ed in in-
e sely p opo ional QRS ampli udes o body conduc i i ies in all
leads (R-peak ampli udes o 9266% and 112 610% o baseline o
inc eased/dec eased o so conduc i i ies, espec i ely). In he p e-
co dial leads, educed body conduc i i ies inc eased R-wa e p og es-
sion (Figu e 4, solid blue aces), and ice e sa.
Va iabili y in loca ion o endoca dial oo
poin s a ec s S-wa e p og ession
Addi ional simula ion s udies we e pe o med in o de o e alua e
how ana omical a iabili y in he loca ion o endoca dial ac i a ion
oo poin s (a coupling si es o he ee- unning Pu kinje sys em) im-
pac s he human QRS complex. Eigh ac i a ion sequences we e de-
signed based on he ana omical a iabili y epo ed on he human
i ascicula sys em,
9
as well as he ea ly ac i a ion si es in he ex i o
mic oelec ode s udies by Du e e al.
4
and in i o elec omechanical
wa e imaging by P o os e al.
5
These included six ac i a ion se-
quences al e ing he posi ion o LV an e io and pos e io coupling
si es owa ds mo e apical and basal loca ions, and wo o mo e ap-
ical and/o la e al RV bundle coupling si es.
Va iabili y in he loca ion o ea ly ac i a ion on he an e io LV
wall, om mo e basal owa ds mo e apical posi ions (Figu e 5A, op
o bo om), was associa ed wi h inc easing S-wa e p og ession in
helimbandaugmen edlimbleads(Figu e 5A,bluea ows).This
p og ession was especially ma ked in leads II, III and aVF, exhibi ing
inc easingly mo e nega i e S-wa es. Va iabili y in he apico-basal
loca ion o he LV pos e io wall si e was simila ly co ela ed wi h
inc easing S-wa e p og ession in p eco dial leads V5 and V6
(Figu e 5B, ed a ows), as hese a e he only elec odes wi h a
solid angle iew o he LV pos e io wall. Some o he ac i a ion
sequences also p esen ed no ched QRS complexes in p eco dial
and limb leads (Figu e 5AandB, black a ows), as a consequence o
a iabili y in he posi ions o simul aneous LV ea lies epica dial
b eak h oughs. Such QRS ea u es we e also obse ed in heal hy
indi iduals o he PTB Diagnos ic ECG Da abase, as shown in he
wo ECG exce p s p esen ed in Figu e 5C. On he o he hand, ac i-
a ion sequences wi h mo e apical and/o la e al RV si es led o
simila QRS mo phologies compa ed o baseline (Supplemen a y
ma e ial online, Figu e S3). Simul aneous a ia ions o mo e han
one LV/RV coupling si es led o an addi i e e ec o he abo e dis-
cussed con ibu ions.
Knockou o oo nodes ep oduce QRS
pheno ype caused by in a en icula
conduc ion de ec s
Due o hei di ec co espondence wi h he i ascicula ac i a ion
sys em, a iabili y in he numbe o oo nodes can be used o ep o-
duce pa hophysiology o he human conduc ion sys em. This is illus-
a ed in Figu es 6and 7 o LV an e io /pos e io hemiblocks and
comple e LV/RV bundle b anch blocks, espec i ely.
Knocking ou he ac i a ion o he LV an e io wall (Figu e 6A,
black ci cle) esul ed in he la e ac i a ion o he LV basal ee
wall. While he nega i e pole on he ches was almos un a ied,
his la e egion caused he posi i e pole in BSPs o mo e up om
he le hip o he cen e o he back, yielding a la e QRS ec o
mainly o ien ed om on o ea o so su aces and sligh ly le -
wa ds and upwa ds. As a consequence, he econs uc ed QRS
complex exhibi s he clinical mani es a ions o a le an e io asci-
cula block
9
(LAFB; lowe - igh panel o Figu e 6A): QRS wid h
smalle han 120 ms; le -de ia ed la e QRS axis o 50;qRpa -
e n in aVL and RS/Rs pa e ns in leads II, III and aVF, wi h wa e
ampli udes sa is ying RII >RIII and SIII >SII; and delayed in insi-
coid de lec ion ( ime om Q o R peaks) in aVL o 42 ms, close o
he clinical h eshold o 45 ms.
Simila ly, he knockou o LV pos e io wall ac i a ion (Figu e 6B,
black ci cle) was associa ed wi h a delayed ac i a ion o he pos e io
LV basal wall, se e ely a ec ing he ime cou se and posi ion o he
poles in BSPs on he ches . Such a dis ibu ion o BSPs eco e ed he
clinical pheno ype o a le pos e io ascicula block
9
(LPFB; lowe -
igh panel o Figu e 6B): QRS wid h smalle han 120 ms; igh -
de ia ed la e QRS axis o 100; Rs o RS pa e ns in leads I and aVL;
qR pa e ns in leads III and aVF; and S wa es in all p eco dial leads,
wi h in insicoid de lec ion imes in V6 (24 ms) and aVF (34 ms)
g ea e han in aVL (21 ms). The less-symme ic ( apezoidal-like)
mo phology in he las de lec ion o he QRS complex o all limb and
augmen ed limb leads obse ed in he clinical eco dings is also
eco e ed.
Finally, a comple e knockou o bo h LV an e io and pos e io
b anches led o a la e al p og ession o LV ac i a ion, signi ican ly
delaying sep al o basal LV s imula ion (Figu e 7A). This e ie ed he
cha ac e is ic signa u e o le bundle b anch block (LBBB) in p eco -
dial leads: QRS complex o a leas 120ms; ma ked QS pa e n in V1;
and no ched R-wa e in lead V6. Simila ly, he knockou o coupling
si es on he RV led o a la e al sequence o RV ac i a ion (Figu e 7B),
yielding he mani es a ion o igh bundle b anch block (RBBB): QRS
du a ion o a leas 120 ms; e minal R-wa e in lead V1; slu ed
S-wa es in leads I and V6.
Discussion
This s udy p esen s a de ailed in es iga ion on how a iabili y in he
main de e minan s o he human ac i a ion sequence and passi e
myoca dial and body conduc ion p ope ies ansla es in o a iabili y
in clinical QRS bioma ke s. Ou app oach igh ly couples s a e-o -
he-a ana omically-based mul iscale bidomain simula ions o human
en icula elec ophysiology wi h ex i o and in i o human ac i a ion
da a. This allows augmen ing he in o ma ion a ainable om sca ce
and limi ed expe imen al eco dings in human (usually a ailable o
only small numbe s o subjec s, o low- esolu ion, and highly a iable
be ween indi iduals) and gaining no el physiological knowledge on
he in o ma ion enclosed in he human QRS complex.
R-peak ampli udes exhibi he la ges amoun o a iabili y, as
hese a e shown o be simul aneously modula ed by endoca dial
ac i a ion speed, and by myoca dial in acellula and ex acellula
conduc i i ies. Va iabili y in o so and o gans conduc i i ies shows
an impac on QRS magni ude, in ag eemen wi h he li e a u e.
13
On he con a y, QRS wid h was only egula ed by ac i a ion
speed and in acellula myoca dial conduc i i ies, whe eas in in-
sicoid de lec ion (QR in e al) is only a ec ed by he endoca dial
ac i a ion. In pa icula , he e is a high impac in educing he
ac i a ion sequence speed on QRS du a ion. These insigh s can be
Simula ion o he human QRS complex i 11
Downloaded om h ps://academic.oup.com/eu opace/a icle/18/suppl_4/i 4/2737164 by Uni e sidad de Za agoza. Biblio eca Uni e si a ia use on 26 Feb ua y 2021
o clinical ele ance in o de o sepa a e om he ECG he con-
ibu ion o hese h ee ac o s, as well as o make p og ess
owa ds mo e pe sonalized ac i a ion sequences in compu a-
ional s udies.
Va iabili y in he ana omical loca ions o ac i a ion oo si es (co -
esponding o he coupling o he LV ascicula bundles wi h he myo-
ca dium) was associa ed wi h S-wa e p og ession in he limb and
p eco dial leads, and occasional no ched QRS complexes in he
Figu e 5 Va iabili y in he loca ion o LV endoca dial ac i a ion si es a ec s S-wa e p og ession. (A) Mo e apical coupling si es o he LV an e io
ascicula b anch we e associa ed wi h inc easingly mo e nega i e S-wa es in leads II, III and aVF (blue a ows). (B) Simila ends we e p esen o a
mo e basal coupling o he LV pos e io ascicula b anch in p eco dial leads V5 and V6 ( ed a ows). Va iabili y in LV s imula ion also esul ed in
no ched QRS complexes in p eco dial and limb leads (black a ows). (C) QRS complexes o wo heal hy indi iduals in he PTB Diagnos ic ECG
Da abase
8
(subjec s 237 and 117, espec i ely), exhibi ing simila QRS ea u es. Simula ed and clinical ECG g id esolu ions: 40 ms/0.1 mV.
i 12 L. Ca done-Noo e al.
Downloaded om h ps://academic.oup.com/eu opace/a icle/18/suppl_4/i 4/2737164 by Uni e sidad de Za agoza. Biblio eca Uni e si a ia use on 26 Feb ua y 2021