Ad ances in Elec ical and Elec onic Enginee ing
124
NONINVASIVE IDENTIFICATION OF ISCHEMIC LESIONS IN THE HEART
M. Tyšle , M. Tu zo á, J. Š ehlíko á, E. Hebláko á
Ins i u e o Measu emen Science, Slo ak Academy o Sciences, Dúb a ská ces a 9, 841 04, B a isla a, Slo akia
E-mail:ume ysl@sa ba.sk, Phone: +421 2 5910 4550, Fax: +421 2 5477 5943
Abs ak
V š údii je na hnu á me óda na nein azí nu iden i ikáciu oblas í poškodenia s dca so zmenenou epola izáciou
spôsobených lokálnou ischémiou. Me óda bola es o aná na poí ao om modeli a na skupine pacien o . Vyhodnocuje
zmeny QRST in eg álo ých mapách me aných na po chu h udníka so známou geome iou a ypoí a a ek i alen ný
elek ický dipól, k o ý ep ezen uje mies o, ekos a o ien áciu lézie. Tes o anie me ódy na poí ao om modeli ukázalo jej
schopnos lokalizo a malé subendoka diálne a subepika diálne poškodenia s chybou okolo 1 cm. Z 11 pacien o
s ocho ením 1 cie y, k o í boli elek oka diog a icky mapo aní p ed a po pe ku álnej ka diálnej in e encii, bolo u 8
pacien o možné zmeny QRST in eg álo ých mapách ep ezen o a pomocou dipólu. Ischemické poškodenia boli
lokalizo ané sp á ne u 6 pacien o s ocho ením a ej (LAD) a u 1 pacien a s ocho ením p a ej ko oná nej a é ie (RCA), 1
lokalizácia ocho enia RCA nebola sp á na. Výsledky š údie naznaujú, že zmeny in eg álo ých mapách môžu pomôc p i
iden i ikácii malých ischemických oblas í oblas i epika du alebo endoka du na základe u enia pa ame o ek i alen ného
dipólu cha ak e izujúceho poškodenie.
Summa y
A me hod o nonin asi e iden i ica ion o hea lesions wi h changed epola iza ion caused by local ischemia
was p oposed and es ed on a model and on a g oup o pa ien s. I e alua es changes in QRST in eg al maps measu ed on a
ches su ace o known geome y and compu es an equi alen dipole ep esen ing he posi ion, size and o ien a ion o he
lesion. Tes ing on a compu e model indica ed abili y o he me hod o localize small subendoca dial and subepica dial
lesions wi h an e o less abou 1 cm. F om 11 pa ien s wi h single essel s enosis mapped be o e and a e he pe cu aneous
ca diac in e en ion, di e ences in QRST in eg al maps could be ep esen ed by a dipole in 8 pa ien s. 6 LAD and 1 RCA
lesion we e iden i ied success ully, localiza ion o 1 RCA lesion ailed. Resul s o he s udy sugges ha di e ence QRST
in eg al maps can help in iden i ica ion o small ischemic egions on he epica dial o endoca dial su ace by es ima ing
pa ame e s o an equi alen dipole cha ac e izing he lesion.
1. INTRODUCTION
Fo ischemic ca diac cells, sho ening and
dec ease o ac ion po en ials (AP) is ypical. Sub le
a ia ions o AP in luence he o e all epola iza ion
p ocess and a e exp essed mainly in he ST-T
in e al o su ace ECG signals. I was shown, ha
in eg als o po en ials o e he en icula
depola iza ion - epola iza ion pe iod (QRST
in e al in ECG) depend only on he ac ion
po en ials a ia ions and no on he en icula
ac i a ion sequence [1]. Di e ences in QRST
in eg als o e he o so oge he wi h he knowledge
o o so geome y and elec ical p ope ies hus can
be used o a nonin asi e iden i ica ion o ischemic
egions wi h changed epola iza ion. The aim o he
s udy was o analyze he possibili y o a nonin asi e
dipole model-based iden i ica ion o small ischemic
lesions caused by s enosis o a single co ona y
essel.
2. METHOD AND MATERIAL
Di e ences in QRST in eg als due o he
changed epola iza ion can be in e p e ed as being
caused by addi ional sou ces o igina ing om
changed ac ion po en ials in he ischemic egion. I
he egion is ela i ely small, hese sou ces can be
ep esen ed by a single dipole model loca ed a he
cen e o he egion. In ou s udy we applied a dipole
loca ed in one o n p ede ined posi ions on he
epica dial o endoca dial en icula su ace. Dipole
pa ame e s ep esen ing he changes in he QRST
in eg als we e in e sely compu ed using he
o mula:
M
i
= T
i+
 o i=1,2, ... n
whe e  ep esen s di e ences in QRST in eg als
measu ed in mapped su ace poin s, M
i
is an
es ima e o in eg al o he dipole momen o he
dipole loca ed a he i- h posi ion in he myoca dium
and T
i+
is pseudo-in e se o he ans e ma ix
be ween he he i- h dipole and po en ials in mapped
su ace poin s. This ans e ma ix depends only on
he geome y and elec ical p ope ies o he o so.
C i e ion o inding he bes equi alen dipole
ep esen ing he measu ed da a was he minimal
alue o ms de ia ion be ween o iginal di e ence
QRST in eg al map and map p oduced by a dipole
es ima ed a each o he n p ede ined posi ions. A
he same ime, ela i e alue o he de ia ion
indica ed easibili y o he dipole o ep esen he
di e ence in eg al map.
The me hod was es ed on simula ed su ace
po en ials and se e al e o ac o s in luencing he
accu acy o he in e se p ocedu e we e analyzed.
A o wa d model was used o ob ain body
su ace po en ials in no mal case as well as in he
case o abno mal epola iza ion o he en icles. A
ini e elemen model o hea en icles was
employed o simula e ca diac depola iza ion and
epola iza ion [4]. Up o i e laye s wi h di e en
AP cha ac e is ics we e de ined in en icula walls
Nonin asi e iden i ica ion o ischemic lesions in he hea
125
and in he sep um. Realis ic AP shapes as measu ed
in canine le en icula wedge p epa a ion [5] we e
adop ed. In all simula ions, cha ac e o
expe imen ally obse ed ansmu al dis ibu ion o
AP du a ion was p ese ed and i s ansmu al
dispe sion was abou 40 ms.
Local ischemic lesions we e simula ed by
sho ening o AP by 5 % o 20 % om he no mal
alues. Th ee ypical egions o changed AP
in luenced by s enosis o main co ona y essels
we e de ined: an e o-sep al pa o he LV nea apex
(supplied by le an e io descending co ona y
a e y, LAD), pos e o-la e al pa o he LV close o
he hea base (supplied by ci cum lex co ona y
a e y, Cx) and mid pos e o-sep al LV and RV
(supplied by igh co ona y a e y, RCA). In each
egion, smalle subepica dial and subendoca dial
lesions (3 - 8% o he en icula olume) and la ge
ansmu al lesions (10 - 12% o he en icula
olume) we e simula ed.
Mul iple dipole wi h 168 dipoles was used o
ep esen he ca diac elec ic gene a o . Su ace
po en ials we e compu ed in 3 ms s eps in poin s o a
ealis ic o so model wi h lungs and hea ca i ies [2,
3]. ECG signals in posi ions o selec ed lead se s
we e used o compu e su ace QRST in eg al maps.
In e se iden i ica ion o he ischemic egion
was a emp ed by using body su ace po en ials
simula ed in 4 lead se s: 192 leads in 16x12 g id
(G192), 62 leads o he Ams e dam mapping se
(A62), 32 leads o he an e io lead se by Lux (L32)
and 9 leads in posi ions o F ank VCG leads and
bo h a ms (F9). Second and hi d lead se s we e
subse s o he i s one. Bo h, inhomogeneous and
homogeneous o so model we e es ed in he in e se
compu a ions. Accu acy o he in e se p ocedu e
was limi ed by he chosen se o possible dipole
posi ions. Fo es ing on simula ed da a, 298 nodes
on he epi- and endoca dial en icula su ace we e
de ined as possible posi ions o he equi alen dipole
gene a o . Dis ance be ween he nea es possible
posi ion and co ec loca ion o an equi alen dipole
(cen e o simula ed lesion) was om 1.7 o 7.3 mm,
mean 5.5mm.
Measu ed da a om 11 pa ien s a e
myoca dial in a c ion (MI) ha unde wen
success ul pe cu aneous co ona y in e en ion (PCI)
on single essel (8 LAD, 1 Cx, 2 RCA) we e used
o expe imen al e i ica ion o he me hod. QRST
in eg al maps be o e and a e he in e en ion we e
compu ed in a 12x16 g id om 32 ECG leads
measu ed in he L32 lead se . In eg al alues in
maps we e co ec ed o QT in e al leng h i i
a ied mo e han 5% be ween he measu emen s.
Common ealis ic inhomogeneous o so and hea
model geome y we e used in all pa ien s o ind an
equi alen dipole ep esen ing he ischemic egion
wi h changed epola iza ion. Fo eal da a, possible
posi ions o equi alen dipoles we e de ined a he
cen e s o 28 segmen s o a ealis ic hea model.
3. RESULTS
Simula ed da a. In Fig. 1 he e is an example o
simula ed lesions in pos e o-la e al egion o he LV.
Subepica dial lesion (PE), subendoca dial lesion
(P2) and ansmu al lesion (P3) o di e en size
we e c ea ed.
Fig. 1. Simula ed lesions (ligh g ey a eas) in pos e o-
la e al LV. Le : subepica dial (PE), cen e :
subendoca dial (P2), igh : ansmu al (P3).
Simula ed AP changes ep esen ing local
ischemic lesions in h ee selec ed en icula egions
we e p ojec ed o body su ace po en ials and
ypically loca ed di e ences o QRST in eg als we e
clea ly isible in co esponding a eas in body
su ace maps. The di e ences inc eased wi h
inc easing lesion size and deg ee o AP sho ening
excep o ansmu al lesions whe e he di e ences
we e much smalle han in compa able non-
ansmu al lesions.
No mal simula ed QRST in eg al map and
in eg al maps ob ained when AP was sho ened by
20% in lesions PE, P2 and P3 is shown in Fig. 2.
The AP changes we e p ojec ed as inc ease ( o PE
lesion) o dec ease ( o P2, P3 lesions) o he QRST
in eg al mainly on he mid pos e io o so su ace.
Fig. 2. Simula ed QRST in eg al maps o no mal
depola iza ion- epola iza ion (NORM) and o ac i a ions
wi h AP sho ened by 20% in lesions PE, P2 and P3.
Co esponding di e ence QRST in eg al maps a e shown
in he igh column. S ep in maps is 6 mV.ms.
Ad ances in Elec ical and Elec onic Enginee ing
126
Summa y dipola sou ce o he lesion calcula ed
as sum o dipole changes in all model elemen s
wi hin he lesion ( ull line ec o ) and in e sely
es ima ed equi alen dipole (dashed line ec o ) a e
illus a ed in Fig. 3.
Fig. 3. Simula ed ischemic lesions PE, P2, P3 and hei
dipola ep esen a ions (pos e io iew o he en icles).
Rep esen ing dipoles compu ed as sum o simula ed
elemen a y dipola changes in he lesion a e ma ked by
ull lines, equi alen dipoles in e sely calcula ed om
simula ed su ace ECG signals a e ma ked by dashed
lines. Le : PE lesion; cen e : P2 lesion; igh : P3 lesion.
Resul s o he in e se solu ion using 62 ECG
leads and homogeneous o inhomogeneous o so
model wi h lungs and hea ca i ies a e summa ized
in Table 1. Rela i e ms de ia ions be ween o iginal
di e ence QRST in eg al maps and equi alen
dipola maps we e om 9 o 16% and sugges ha
dipole may be an adequa e ep esen a ion o small
ischemic lesions. Fo small subendoca dial and
subepica dial lesions, maximal localiza ion e o
eached 16 mm in inhomogeneous o so and 23 mm
in homogeneous o so. Localiza ion o la ge
ansmu al lesions was less sa is ac o y and maximal
e o eached unaccep able 43 mm. O ien a ion o
he equi alen dipoles ma ched well he simula ed
lesions, howe e , ela i e e o o dipole momen s
subs an ially inc eased o mo e dis ibu ed sou ces,
especially o la ge ansmu al lesions.
Tab. 1. E o s (mean ± s anda d de ia ion) o he in e se
es ima ion o lesion pa ame e s om 62 su ace ECG
leads using inhomogeneous o homogeneous o so model.
Mo e de ailed e alua ion o he localiza ion
e o is shown in Fig. 4. Localiza ion o small
lesions om 192 and 62 leads p o ided simila
esul s, localiza ion om 32 leads was wo se o
la ge lesions and inhomogeneous o so, while
localiza ion om 9 leads was no sa is ac o y. Fo
la ge lesions, in luence o he numbe o leads was
gene ally highe . In mos cases, esul s ob ained
using homogeneous o so model we e less accu a e
han esul s ob ained when inhomogeneous o so
was used.
Fig. 4. Mean alues o he localiza ion e o [mm] o
small subendo- o subepica dial lesions and o la ge
ansmu al lesions when using di e en lead se s and
homogeneous o inhomogeneous o so models.
Measu ed da a. In 8 o 11 s udied MI pa ien s
we ha e ound conside able changes in QRST
in eg al maps a e he PCI ea men ha could be
app oxima ely ep esen ed by a single dipole (wi h
ela i e ms e o less han 50%). In emaining 3
pa ien s he e o was > 60% and hey we e excluded
om u he analysis. In 6 o 8 analyzed pa ien s, he
QT in e al co ec ion was used o compensa e he
changed hea a e be ween he measu emen s.
Fig. 5. Le ( op o bo om): Measu ed QRST in eg al
maps (s ep in maps 8 mV.ms) be o e and a e success ul
PCI on LAD and co esponding di e ence in eg al map in
a 68 yea old male wi h an e io MI (closing a RD 2
b anche). Righ : Localiza ion o an equi alen dipole
sou ce ep esen ing he changed epola iza ion in a
ealis ic myoca dium model.
Pa ame e To so Small
lesions
La ge
lesions
inhomog
9 ± 4 17 ± 14 Localiza ion e o
[mm]
homog 11 ± 8
16 ± 15
inhomog
9 ± 7 14 ± 4
Dipole di ec ion [°]
homog 8 ± 5 17 ± 7
inhomog
51 ± 40
221 ± 206
Dipole momen
[%]
homog 49 ± 33
163 ± 123
inhomog
9 ± 4 16 ± 1
Map el. di e ence
[%]
homog 12 ± 2 16 ± 2
Nonin asi e iden i ica ion o ischemic lesions in he hea
127
Despi e he indi idual pa ien geome y was no
a ailable and single s anda d o so model was used,
in 7 o he 8 analyzed pa ien s he posi ions o
es ima ed equi alen dipoles app oxima ely ma ched
he egion supplied by he ea ed essel o a leas
hey we e co ec ly loca ed a an e io o pos e o-
la e al wall o he LV wi h he dipole di ec ed
owa ds he supposed ischemic egion. Di ec ions o
dipole momen s in se e al cases we e no no mal o
he pa icula hea wall and lesion bo de wha
migh e lec speci ic o m o he a ec ed a ea o o
aniso ophy in eal myoca dium. In 1 pa ien a e
PCI on RCA, he equi alen dipole was loca ed in
mid an e io LV wall wi h a dipole momen di ec ed
ou o he hea olume.
In Fig. 5 he e is an example o measu ed
pa ien da a and success ul loca ion o he equi alen
dipole a e PCI on LAD.
4. DISCUSSION
Ou p e ious s udies on de ec ing small local
ischemic changes by using commonly used
depa u e in eg al maps [6] showed ha he changes
in body su ace po en ials a e small when compa ed
wi h no mal in e -indi idual luc ua ions and can
ha dly be de ec ed by depa u es om mean in eg al
maps compu ed o he no mal popula ion.
In ou simula ions o small ischemic lesions,
ela i e ms di e ences be ween no mal and
changed QRST in eg al maps we e 20 - 45%, and
co ela ions .45 - .99. These da a indica e g ea e
changes han obse ed o al in a-indi idual
a iabili y in maps o heal hy subjec s ( ms
di e ences 5 - 20%, co ela ions >.98) wha , in
p inciple, allows iden i ica ion o he small
ischemic lesions by he p oposed me hod.
In his s udy, ischemic egions we e simula ed
only by AP du a ion changes. Simul aneous changes
o AP ampli udes p esen in eal da a we e also
es ed and hey inc eased he di e ences in he
QRST in eg al maps because o he simila e ec o
bo h, AP sho ening and AP ampli ude dec ease on
he in eg al maps.
Limi a ion o he simula ion s udy was he
simple o wa d model wi h analy ical hea
geome y and use o iso opic myoca dium. Ac ion
po en ial shapes we e de ined a p io i and possible
elec o onic coupling was no simula ed.
A ailable MI pa ien da a measu ed only in 32
mos ly an e io leads and in di e en ime in e als
be o e and a e he PCI we e no ideal o he s udy.
P oposed me hod can e alua e changes o he
esidual ischemia a e MI in luenced by he PCI
ea men . The en icula a ea a ec ed by he
in e en ion may be qui e la ge. Mo eo e , in he
cen e o he in ac ed a ea p obably also pe manen
issue damage exis s ha can make he ea ed a ea
e en mo e agmen ed. This migh be he eason
why single dipole model could no ep esen he
di e ence in eg al maps o 3 ea ed MI pa ien s
wi h accep able accu acy.
5.
CONCLUSIONS
Resul s o ou simula ions showed ha local
epola iza ion changes in di e en hea egions
could be obse ed as changes in body su ace
po en ial maps. Di e ence QRST in eg al maps and
equi alen dipole sou ce model p o ed o be a use ul
ool o assess small ischemic egions and o iden i y
hei p oxima e si e in he myoca dium. Ex en o he
lesion was e lec ed in he dipole momen and
p e alence o subepica dial o subendoca dial
cha ac e o he lesion could be de e mined om he
dipole o ien a ion. Fo accep able localiza ion o
small egions wi h changed epola iza ion, 192, 62
o 32 leads we e su icien e en i homogeneous
o so model was used. Howe e , because o he
limi a ions o he model, alidi y o he ob ained
esul s has o be u he e i ied on addi ional
measu ed da a.
Localiza ion o la ge ansmu al lesions and
es ima ion o hei size was no sa is ac o y. These
lesions a e less clea ly mani es ed in he su ace
po en ials and he dipole model seems no o be
app op ia e o hese cases.
Tes ing o he me hod on a ailable eal da a
sugges s ha he p oposed me hod could be a use ul
ool o nonin asi e assessmen o ischemic egions
wi h changed epola iza ion. Use o indi idual o so
geome y could imp o e pe o mance o he me hod.
Acknowledgemen
This wo k was suppo ed by g an 2/4089/24 om
he VEGA g an agency. The au ho s hank o p o .
Filipo a om he Slo ak Ins i u e o Ca dio ascula
Diseases o o e ing he measu ed pa ien da a and
assis ance in hei e alua ion.
REFERENCES
[1] M.C. T udel e al.: IEEE T ans. on BME, 2004,
51, 1319 (2004).
[2] M. Tyšle , V. Sza hmá y, M. Tu zo á: In . J. o
Bioelec omag., 5, 252 (2003).
[3] V. Sza hmá y, I. Ru kay-Nedecký: S udies in
Heal h Technology and In o ma ics, 90, 88
(2002).
[4] V. Sza hmá y, R. Os ald: Compu Biomed Res
27, 27 (1998).
[5] G.X. Yan, W. Shimizu, Ch. An zele ich:
Ci cula ion 98, 1921 (1998).
[6] S. Filipo á e al.,. Ad ances in Body Su ace
Mapping and High Resolu ion ECG, (Li e
Medicom Co, 1995).