Professional Documents
Culture Documents
tmpB83D TMP
tmpB83D TMP
Chang*,
Y.Z. Tseng*,
*Departments
of Physiology
and Medicine,
College of Medicine,
National Taiwan
Institute
of Electrical Engineering,
University;
%ection
of Biomedical
Engineering,
Research Laboratory, TzuNational Taiwan University, Taipei, Taiwan; *Cardiovascular
Chi Medical Research Center, Hualien, Taiwan
Received 1 September 1993, accepted 4 January
1994
ABSTRACT
This study determines the role of an asymmetric T-tube model as a representation of arterial
model consists of two non-unifrm
mechanical
properties. The
geometric and elastic tapering and each tube terminates in a complex load. Pulsatile pressure andflow
velociQ of the
ascending aorta were measured in 10 closed-chest, anaesthetized dogs. An exponentially tapered transmission line is
used to describe the non-unrfonn properties of the vasculature. The phase constant is a function of position along the
path length due to geometric and elastic tapers. This non-unijonn
the measured
pressure waveform in the ascending aorta. Model parameters could be estimated and used to interpret the physical
properties of the arterial system. The mathematical and experimental model impedance spectra are similar. There is a
close correspondence
derived from
the non-unrfonn
T-tube
computed from measurements on dogs. The results suggest that inclusion of tube tapering improves the mathematical
model so that it closely represents the experimentally &rived arterial impedance
the non-unrform properties of wave-transmission paths may play an important role in governing the behaviour of an
asymmetric T-tube for the description of the arterial system.
Keywords:
model;
Exponentially
tapered
arterial input impedance
transmission
vascular
non-uniformity;
asymmetric
T-tube
September
INTRODUCTION
Vascular impedance is a useful description of the
physical properties of the arterial system, because it
encompasses
both the steady state and pulsatile
components
of the load presented
to the heart.
Many models have been pro osed to analyse the
impedance of the vasculature P-. One of the most
appealing models is the asymmetric
T-tube with
terminal loads. This model consists of two sections of
different lengths. The shorter section represents the
circulation of head, neck, and upper limbs (head or
upper body circulation),
and the longer section
represents the circulation of trunk and lower limbs
(body or lower body circulation). The model was first
suggested by McDonald
in 19683 to explain the
presence of two functionally discrete reflection sites
in the systemic arterial system, and subsequently
championed by ORourke and Avolio.
Recently, Burattini et ~1.~ proposed a modified
asymmetric
T-tube
model to relate the pulsatile
Correspondence and reprint requests to: Kuo-Chu Chang PhD,
Department of Physiology, College of Medicine, National Taiwan
University, No. 1, Sec. l;Jen-Ai Rd, Taipei, Taiwan.
0 1994 Butterworth-Heinemann
1350-4533/94/05370-09
line;
for BES
Med. Eng. Phys. 1994, Vol. 16, September
370
properties
of the transmission
tubes to relate the
pulsatile
pressure
and flow in the ascending
aorta.
An exponentially
tapered transmission
line is used to
describe
the non-uniform
property
of the vasculature ,18. We examine
the ability
of this model
when applied
to data from basal, vasoconstricted,
and vasodilated
states.
MATERIALS
AND METHODS
Experimental preparations
We used 10 mongrel
dogs of either sex weighing
12.3 f 1.2 kg. Each dog was anaesthetized
with pentobarbital
sodium
(25 mg kg- i.v.), intubated,
and
ventilated
with room air supplemented
with oxygen.
To measure
the aortic pulsatile
flow velocity
and
pressure,
a high-fidelity
micromanometer
(Model
SVPC-664D,
Millar
Instruments
Inc., USA)
was
inserted
through
the right carotid
artery and was
advanced
into the ascending
aorta. The left carotid
artery
was kept intact.
The pressure
sensor
and
electromagnetic
fluid velocity sensor were located at
the same position. The velocity sensor was connected
to a flowmeter
(Model
501D,
Carolina
Medical
Electronics,
Inc.,
USA).
In all dog experiments,
measurements
were taken under basal conditions.
In
addition,
one dog was given
a vasoconstrictor
(angiotensin
II, 15 ngkg-l)
and another
dog was
given a vasodilator
(Captopril,
0.5 mg kg-). Each of
these compounds
was administered
via an intravenous drip until the desired pressure effect, hypertension or hypotension,
was achieved,
and then waveforms were measured.
The analogue
waveforms
were
sampled
at 250Hz, digitized
and stored on a digital
computer
for off-line analysis.
Signals (5-10 beats at
steady
state)
were selected
on the basis of the
following
criterialg:
(1) recorded
beats with optimal
velocity profile;
(2) beats with an RR interval
less
than 5% different
from the average
value for all
recorded
beats;
(3) exclusion
of ectopic and postectopic beats. The optimal
flow velocity signal was
characterized
by a steady diastolic
level, maximal
systolic amplitude,
and minimal
late systolic negative how.
The
investigation
was performed
in
accordance
with the Home Office Guidance
on the
operation
of the Animals
(Scientific
Procedures)
Act
1986, published
by Her Majestys
Stationery
Office,
London.
Figure 1
the relation
between input impedance
Zini of the ith
tube,
characteristic
impedance
Zci and reflection
coefficient Fini at the entrance
of the ith tube, phase
constant pi and effective length d; can be described
by the following equations
(i=h,bj
Theoretical formulation
The asymmetric
T-tube
model
and its terminal
complex load are shown in Figure 1. The properties
of
each tube include
a characteristic
impedance
Z, at
the entrance
of the tube and a transmission
time 7.
r is the time for a wave to propagate
from one end of
the tube to the other. Properties
of the load are given
a high-frequency
tubeby the
load
elements:
matching
impedance
element
R,, a load compliance
C,, and a terminal
resistance
Rp. In this system, the
ascending
aorta section is not a part of the model,
and the input impedance
of the model is a parallel
combination
of the two arms of the T-tubea10~20.
From linearly
tapered
wave-transmission
theoryr8,
1994, Vol.
16, September
371
Vascular non-unifnnily
et al
d(ln Zi)
-=dBi
In Z,i
(7)
&i
Again, substitution
of equation
(7) into equation
(4)
with the performance
of straightforward
integration
gives rise to an equation
as follows:
++
I
BOi
sin Figure 2
Tapered
transmission
r.
= 1 ,-jWW2) ln &
In2
2
eOi
2
dTi=
2+dZi-~~
_
Zi+dZi+Zi
1 d
_ E-(In Zi) dz
2 dz
Boj=
where u is a dummy
variable
that measured
the
distance
from the point 2 = 0 toward the load end.
The phase angle of the reflected wave arising at u is
2pi(2U) relative to the forward propagating
wave
at z. Differentiating
I?i(z) with respect to z gives
dri
1 d(ln
dr -2j/3jri---
2)
(2)
dt
Since ,B is a function
introduce
an auxiliary
equation
(2) as follows:
04
= 2wri
(9)
Qhik)
where
Ti =
7i
is the time for a wave to propagate
from the
beginning
to the end of the tube. Substitution
of
equation
(9) into equation
(8), leads to the following
important
equation:
of 2 along
the taper,
we
variable
Bj to integrate
(10)
Equation
(10) describes
the relation
between
the
input
reflection
coefficient,
the terminal
complex
load and the wave transmission
time over the entire
path length. The normalized
terminal
impedance
of
the ith tube could be expressed as follows:
and hence
dri
de_1 -ITi-
1 d(ln
ZJ
57
This is readily
integrated
(3)
to give
(4)
From equations
(l), (10) and ( 1 1), we can readily
calculate
the input
impedance
of the individual
tube. The input impedance
represented
by the Ttube model is the parallel
combination
of Zinh and
Zinb a3 below:
(5)
za(jO)
The exponential
taper-is
one for which 1n Zi varies
linearly,
and hence Zi varies
exponentially,
from
unity
t0
lnZLi.
Since
the phase
constant
is a
function
of the path length,
the relation
between
lnZi and lnZLi could be expressed as follows:
Total
aortic
where
Zinh(j@)
BOj=
2Pi k) dz
In Zi =
Pi(<)
Od,
d<
In ZLi
&I/l@)
(12)
peripheral
pressure,
Rp = $&b&t,
+&lb(@)
_resistance
P, to cardiac
+
- Rpb)= P/Q
(13)
(6)
Pi(z) dz
I
Differengating
In .Zi with respect to z: and substituting equations
(5) into equation
(6) with the performance
of variable
translation
gives
the following
equation:
372
zinb(jw)
,=
R*izci
(14)
R,c+i-Z,i
Wave reflection
is therefore absent
cies.
The model as it is formulated
at high
requires
frequena priori
Vascular
specification
of the peripheral
resistances
in the two
This
requirement
could
be
transmission
paths.
satisfied
when
the ratio
(K&
between
measured
descending
thoracic
aorta
mean
flow (&J
and
From this ratio and
cardiac output (Q) is calculated.
knowledgeof
Rp, the resistances
&, and Rpb can be
calculated.
&
= &I( 1 - &to)
Rpb = &l&t,
Calculation
of &,h and Rpb would reduce the number
of free model parameters
to six. In this study Kdto
could, however,
not be calculated
directly
because
all dogs studied
had closed chests. To conquer
this
deficiency,
we. took assumed
values
of Kdt, from
Campbells
report9 for all dogs under basal, vasoconstricted,
and vasodilated
states. The ratio of Kdl,
was assumed
to be 0.70, 0.62, and 0.76 under these
conditions,
respectively.
The result of the study with
these assumed values are satisfactory.
Parameter estimation method
Given the non uniform
T-tube
model and sampled
values of measured
aortic pressure,
P(t), and flow,
Q(t), the model parameters
could be estimated
and
functional
properties
of the
the
corresponding
arterial
load identilied8~i0.
The
six model
parameters
are the characteristic
impedances
at the
( CLh,
entrance
of tubes, (Z,,, Z,b), load compliances,
cLb), and wave transmission
times over the path
length in the individual
tubes, (rh, 76).
For estimating
parameters
P(t) is taken as output
variable.
variables
while Q(t) is the model input
Harmonics
of ascending
aortic
flow, Q( Jo),
are
calculated
by the fast Fourier
transform.
These
harmonics
are multiplied
by the T-tube
input
the corresponding
impedance,
Z, ( ju), t o calculate
harmonics
of ascending
aortic pressure,
P( jw):
Zz(jw)
(17)
Model-estimated
ascending
aorta
pressure
in the
time domain,
P(t),
is obtained
by applying
the
inverse
Fourier
transform
to harmonics
calculated
from equation
( 17). A normalized
root-mean-square
error, e* is calculated
as
e*=f
Ji
e(i)2
i=l
(18)
where
e(i.
Chang
et al.
(15)
(16)
fb4 = QW)
non-unifomityand
=Pj(jiO)
P
n is the number
of data points, and P(i) and P(i) are
respectively
sample
values
of the measured
and
model-generated
aortic pressure
waveforms.
Parameters of the model are then adjusted
to minimize
e* using
the Nelder-Meade
Simplex
algorithml.
The algorithm
is a direct search method for finding
a functions
minimum.
The parameters
coincident
with the minimum
are taken as the model estimates
of arterial properties.
(i=h,6)
(19)
(20)
where Pi is pressure
and Qi is flow of ith tube. The
subscripts
f and r indicate
forward
and reflected
(backward),
respectively.
Equations
( 19) and (20)
show that pressure
and flow waves at the input of
individual
tubes are the sum of forward and backward waves.
Harmonics
of forward pressure
could be obtained
from
ascending
aortic
pressure
harmonics
after
estimation
of the model parameters,
that is after
calculating
tube
input
impedance
Zini (jo).
The
following
procedure
is used. The relation
between
forward and backward
transmitted
harmonic
waves
is given by the reflection coefficient Fini ( jw):
Pr,(jw)
Qti(jm)
= rini(jw)
Substituting
in
Pfi(jW)
pJ;(j)
= -rini(jm)
=P(jw)l[l
(21)
Qf;Cjw)
equations
(22)
(19) results
(23)
+rini(jw)l
Because
the following
Fini
= [zini(jw) -Z~ill[Zini(jw)
relation
holds:
+ Zil
(24)
by substituting
equation
(24) into equation
(23),
forward pressure
harmonics
in each tube could be
calculated
from P( jw) and Zini(jU).
After
forward
pressure
harmonics
have
been
calculated
and taking into account
that the following relations hold:
JjWVQfi1i(j4
P(i) -P(i)
=P(jw)
Pi(jw)lQti(jo)
= Zi
(25)
= -Z,
(26)
forward
flow harmonics
could be computed
from
equation
(25). Finally, backward
pressure harmonics
could be computed
from equation
(21) and backward
flow harmonics
from equation
(26). Time-dependent
signals are given by the inverse Fourier transform.
After setting
the ascending
aorta
characteristic
impedance
Z, equal to the parallel
combination
of
Z,, and Z&, we could infer forward
(Py, QJ) and
backward
(Pr, Q,) p ressure and flow waves at the
entrance
of the T-tube
model.
At this point,
the
Med. Eng. Phys. 1994, Vol. 16, September
375
Chang et al.
Table 1
Indices
of model
Dog
e*x
1BA
2BA
3BA
4BA
5BA
6BA
7BA
BBA
9BA
1OBA
4.76
8.27
6.71
5.48
5.52
10.13
11.68
5.15
5.48
6.54
Mean
fSD
applied
to experimental
data
Sl0pe
Intercept
rp
1.0049
0.992
-0.6837
0.0734
1.0075
0.9902
1.0083
1.0917
0.9943
1.0067
0.9968
0.9999
-0.8908
1.3621
-1.1272
- 1.6274
0.7795
-0.9149
0.3390
-0.0046
0.9829
0.9563
0.9881
0.9781
0.9734
0.9464
0.9363
0.9785
0.9770
0.9557
6.91
k2.39
1.0092
+o.o29i
-0.2695
+ 0.9354
0.9673
kO.0173
6.49
7.98
0.9949
1.0011
0.7190
-0.1665
0.9688
0.9543
9vc
1OVD
1o-4
fits when
e*, normalized
root-mean-square
error;
slope,
intercept,
and
r,
linear
regression
parameters
of the model output
variable,
P(t), on
measured
variable,
P(t). BA, VC, VD = basal, vasoconstricted,
and
vasodilated
conditions,
respectively
reflection
coefficient
could
be expressed
as follows:
(27)
By analogy
with equations
(2 l-26),
harmonics
of
forward and backward
pressure
and flow could be
computed
as Pf(jw)
= P(jo)l[l
+r,(jw)],
Qf(jw)
= (pf(jw)lZ,
J,(W)
= -LW
signals
form.
QfW4,
are obtained
= r,(_M) p(jw),
and Q,(@)
respectively.
Time-dependent
from the inverse Fourier
trans-
RESULTS
Data fit
Time (msec)
Figure 3
Table 2
indices
Haemodynamic
P
and
estimated
global
parameters
Dog
(mmHg)
e
(ml s-)
Z,*
(dynes
IBA
2BA
3BA
4BA
5BA
6BA
7BA
BBA
9BA
1OBA
135
117
115
142
140
105
140
133
111
134
23
23
25
22
40
20
30
33
35
24
152
159
137
171
154
140
186
100
88
121
Our non-uniform
T-tube model fits the data well in
every dog. A summary
of the measures
indicating
goodness
of tit is given in Table 1 and an example
showing
the similarity
between
computed
and
measured
pressure and flow waveforms
is illustrated
in Figure 3. Figure 3 shows the measured
ascending
of the non-uniform
Z
(dynes
cm-)
T-tube
model
CL
(ml mmHg_)
RP
(dynes
197
156
162
210
115
129
144
84
106
127
0.0490
0.0690
0.1693
0.0637
0.2792
0.1759
0.1351
0.2511
0.4747
0.2339
7689
6769
6059
8552
4657
6927
6286
5409
4265
7452
cmm5)
*SD
128
15
28
+7
141
231
143
f39
0.1901
50.1286
6407
+ 1355
9vc
1OVD
142
102
22
34
100
71
128
93
0.2053
0.2834
8711
4016
Mean
P = mean aortic
pressure;
Q= mean aortic
flow; Z:= characteristic
data points obtained
from the ratio of the corresponding
harmonics
model-estimated
load compliance;
Rp = peripheral
vascular
resistance
374
Med.
Eng.
Phys.
1994, Vol.
16, September
impedance
of pressure
calculated
and flow;
cm-)
by averaging
high-frequency
moduli
of impedance
Z, = model-estimated
characteristic
impedance;
CL =
80 -
Forward
and
backward
pressure
waves
at the
entrance
of head and body transmission
paths are
calculated
from
the measured
ascending
aortic
3 Estimated and calculated
by the non-uniform T-tube model
parameters
of head circulation
RPh
Tube
parameters
Load
R.h
~$mmHg-)
%ynesa&)
(III:
0.0350
0.0190
0.0974
0.0325
0.0784
0.0477
0.0491
0.0738
0.1045
0.0978
407
519
287
421
307
261
249
242
193
266
29.00
23.94
14.63
23.77
18.54
13.56
22.56
15.58
22.66
13.54
320
0.0635
f 104 rtO.0308
315
flO1
19.78
+5.35
293
201
289
198
8.72
15.24
Dog
(dynes cmm5)
IBA
2BA
3BA
4BA
5BA
6BA
7BA
8BA
9BA
IOBA
25 628
22564
20197
28 505
15523
23089
20950
18027
14214
24838
413
531
291
427
313
264
252
245
196
269
Mean
*SD
21354
-t4516
9vc
1OVD
22744
16392
Subscript
h = head
circulation.
Rp = peripheral
resistance;
R. =
high-frequency
tube-matching
impedance;
CL = load compliance;
Z,= characteristic
impedance
at the entrance of the tube; 7=
transmission time
CL,
Dog
(dynes cm-)
IBA
2BA
3BA
4BA
5BA
6BA
7BA
8BA
9BA
IOBA
10988
9 674
8 659
12221
6 655
9 899
8983
7 729
6 094
10649
9 155
1936
Mean
*SD
9VC
1OVD
14120
5321
of body circulation
by
Tube
RPb
0.0568
0.1061
400
Figure
Table 4
Load
300
Time (msec)
Table
200
100
(ml mmHg-)
Z,b
(dynes cm-)
Tb
(ms)
228
388
432
188
262
354
129
245
248
0.0140
0.0500
0.0719
0.03 12
0.2008
0.1283
0.0859
0.1773
0.3710
0.1361
384
222
371
417
183
255
340
127
235
243
84.89
62.42
20.89
69.26
48.15
39.04
48.73
35.95
47.28
19.28
287
f99
0.1266
kO.1053
278
+95
47.60
f20.56
232
181
0.1484
0.2075
229
175
25.67
21.07
Rob
397
315
Va.wuccular
non-unifnni~
pressure
after estimation
of the model parameters.
The time course of head-end
reflected wave (dashed
line) and body-end
reflected
wave (dotted line) is
showed in the lower panel of Figure 4. Since the
characteristic
impedance
at the entrance
of the Ttube is equal to the parallel combination
of Z,, and
and reflected
pressures
in the
Z cb, the forward
ascending
aorta could be computed
(solid lines in
Figure 4).
DISCUSSION
Non-uniform
Impedance patterns
Since the harmonic
amplitudes
are low at frequencies above 20 Hz, they have little effect upon the time
domain waveforms,
but the errors appear to be large.
The aortic input impedance
spectra in the same dog
are shown in Figure 5. The solid lines represent
the
predicted spectra at the entrance
of the non-uniform
T-tube
model.
The
dotted
lines are the spectra
1 i:
: I,
: I,
i
0.8 - i
::
:
:
t>
:
,
O0
,O
10
20
15
2-
T-tube system
Explanation of results
L
0
I
5
10
Frequency
15
20
(Hz)
376
contrast
to the findings of Burattini
and Campbel18,
et al. lo, who found that
Campbell
et al., and Burattini
was significantly
(24%,
21% and 32%,
respec%b
tively) larger than Z,,,. All their studies adopted the
uniform T-tube model to relate the pulsatile pressure
and flow in the ascending
aorta. In comparison
with
the uniform T-tube
model, our non-uniform
T-tube
has slightly better fits to the measured
signals, with
the smaller
magnitude
of e* (6.91 k2.39
US. 8.06
of
* 3.22 x lo+, p < 0.05) and the larger coefficient
determination,
r*, (0.9673 f 0.0173 vs. 0.9472 zb 0.0350,
p < 0.05). The better model fit and more reasonable
estimates
in Z,, and Z,b are the improvement
of our
non-uniform
T-tube
model over the earlier T-tube
More
studies
are
required
to delineate
model.
relative
responsiveness
of the estimated
Z,b and
Z,,, to the uniform and non-uniform
T-tube models.
The parameter
that describes
the distensibility
of
large arteries
is tube compliance.
Since
we have
spatial
tapering
of our tube for which
the phase
constant
is a function
of the path
length,
the
calculation
of tube compliance
from the knowledge
of Zci and 7, seems to be meaningless.
The T-tube
model encourages
investigators
to use
this model to assess the influence of wave reflections
on heart-arterial
coupling.
From equation
(21) with
its inverse Fourier transform,
the backward
pressure
waves from the head end and body end are calculated.
When
these
equations
are applied
to the
measured data from dog 1, the result is the backward
waves shown in Figure 4. In the basal state,
the
backward
waves from head end and body end of the
circulation
are distinct,
each wave arriving
at wellseparated
times
and each wave having
sufficient
amplitude
to make a noticeable
impression
on the
With
the T-tube
measured
pressure
waveform.
model,
there
is the opportunity
to estimate
the
strength
of these head-end
and body-end
reflected
waves respectively,
and their contribution
to left
ventricular
loading.
Such could not be done with a
single-tube
or windkessel
model.
Limitations
of model
Our
non-uniform
T-tube
model
is necessarily
a
reduced
representation
of the
systemic
arterial
system. Many important
aspects of arterial distribution and wave transmission
have not been included in
the model. The most significant
features not represented in the model are wave attenuation,
non-linear
pressure-diameter
relation,
and vessel
branching
beyond main branching
into head and body circulation at the level of the aortic arch. Therefore,
head
and body tubes are assumed
to be loss free and are
loaded with an effective low-pass filter complex load
rather than with branching
elements.
Non-linearities
arise from the elastic properties
of
arterial
wall* as well as from
the existence
of
convective
acceleration
which is mainly
caused by
the arterial diameter
tapering.
The convective
acceleration
is negligible
compared
with the transient
acceleration
if a characteristic
velocity of the distribthe wave propagation
uted flow, Q, is less than
is most valid for the
speed, cp,,. This assumption
higher harmonics,
wherein
q becomes
quite small
compared
to $h. Even at the lowest harmonics
f/$h is
never greater than about 0.1, as can be inferred from
studies
of the harmonic
spectra
of blood
flow
contours.
The usefulness
of the concept
of impedance
rests
on the assumption
that the arterial system is linear
with
respect
to pressure
and
flow.
The
linear
model assumes that the variables
at each harmonic
frequency
are independent
of all others,
whereas
non-linearity
implies
an
interaction
between
different harmonics.
Linearity
guarantees
that impedance is an indication
of the physical
state of the
blood vessel, not of the particular
pulsations
that
exist at the moment*.
Although
perfect linearity
is
not to be expected in any biological system, there are
many
findings31-35
suggesting
that
the
arterial
system
is approximately
linear
with
respect
to
impedance,
at least within the accuracy
of present
techniques
for measuring
the
input
impedance
spectrum
in viuo. Since the validity of linear models
depends on the needs of a particular
application,
in
the present study the relationship
between pressure
and flow is assumed to be linear.
Nevertheless,
our non-uniform
T-tube
model is an
appealing
modification
with respect to the uniform
T-tube
system.
Information
related
to path length
and wave velocity
is contained
within
the single
parameter
r. To estimate
path length,
an extra
assessment
of wave velocity
would be required.
It
deserves further quantitative
analysis.
ACKNOWLEDGEMENTS
This
study
was
National
Science
634).
supported
Council
by grants
from
the
(NSC
81-0412-B-002-
REFERENCES
1. Noordergraaf A, Boom HBK, Verdouw PD. The use of an
analog computer in a circulation model. Prog Cardiovasc Dis
1963; 5: 419-39.
2. Taylor
MG. The input impedance
of an assembly of
randomly branching elastic tube. Biophys J 1966; 6: 29-51.
3. McDonald DA. Hemodynamics.
Annu Rev Physiol 1968; 30:
525-56.
4. Westerhof N, Bossmann F, De Vries CJ, Noordergraaf
A.
Analog studies of the human systemic arterial
tree. J
Biomech 1969; 2: 121-43.
5. Avolio AP. Multi-branched
model of the human arterial
system. Med Biol Eng Comput 1980; 18: 709-18.
6. ORourke MF, Avolio AP. Pulsatile flow and pressure in
human systemic arteries: studies in man and in a multibranched model of human systemic arterial tree. Czrc Res
1980; 46: 363-72.
Toy S, Melbin J, Noordergraaf
A. Reduced models of the
circulation. IEEE Trans Biomed Eng 1985; 32: 174-6.
Burattini R, Campbell KB. Modified asymmetric
T-tube
model to infer arterial wave reflection at the aortic root.
IEEE Trans Biomed Eng 1989; 36: 805- 14
Campbell
KB, Burattini
R, Bell DL, Kirkpatrick
RD,
Knowlen GG. Time-domain
formulation
of asymmetric
T-tube model of arterial system. Am J Physiol 1990; 258:
H1761-74.
R, Knowlen GG, Campbell KB. Two arterial
10. Burattini
effective reflecting sites may appear as one to the heart.
Circ Res 1991; 68: 85-99
DA. Blood Flow in Arteries. London: Arnold,
11. McDonald
377
10: 539-50.
through an assembly of
15. Taylor MG. Wave transmission
randomly branching elastic tubes. Biophys J 1966; 6: 697716.
A. Arterial viscoelasticity:
a
16. Westerhof N, Noordergraaf
generalized model. Effects on input impedance and wave
travel in the systematic tree. J Biomech 1970; 3: 357-79.
M. Exponentially
tapered
17. Einav S, Aharoni S, Manoach
transmission line model of the arterial system. IEEE Tran.r
Biomed Eng 1988; 35: 333-9.
18. Collin RE. Foundations for Microwave Engineering. New York:
McGraw Hill, 1966.
19. Chang KC, Hsieh KS, Kuo TS, Chen HI. Effects of
nifedipine on systemic hydraulic vascular load in patients
with hypertension.
Cardiovasc Res 1990; 24: 719-26.
20. Chang KC, Kuo TS, Chen HI. Prediction of arterial input
impedance
spectrum
by an asymmetric
T-tube
model
terminated with an opportune complex load. J Chin Inst
Eng 1992; 5: 533-47.
21. Dennis JE, Woods DJ. In: Wouk A, ed. New Computing
Environments: Microcomputers in Large-Scale Computing, 1987.
des arteriellen pulses. 2 Biol
22. Frank 0. Die grundform
1899; 37: 483-526.
JJr, Schipke J. Assessment
of
23. Burkhoff D, Alexander
Am J
windkessel as a model of aortic input impedance.
Physiol 1988; 255: H742-53.
378
D, Schipke J, Sagawa
K.
JJr, Burkhoff
24. Alexander
Influence
of mean pressure on aortic impedance
and
reflections in the systemic arterial system. Am J Physiol
1989; 257: H969-78.
MG. The input impedance
of an assembly of
25. Taylor
randomly
branching
elastic tubes. Biophys J 1966; 6:
29-51.
26. Westerhof N, Sipkema N, Van Den Bos GC, Elizinga G.
Forward
and backward
waves in the arterial system.
Cardiovasc Res 1972; 6: 648-56.
27. Latson TW, Hunter WC, Katoh N, Sawaga K. Effects
of nitroglycerin on aortic impedance, diameter, and pulse
wave velocity. Circ Res 1988; 62: 884-90.
28. Cox RH, Pace JB. Pressure-flow relations in the vessels of
the canine aortic arch. Am J Physiol 1975; 228: l-10.
M, Histand
MB,
Ogden
E. Dispersion
and
29. Anliker
attenuation
of small artificial
pressure waves in the
canine aorta. Circ Res 1968; 23: 539-51.
30. Pate1 DJ, Greenfield JCJr,
Fry DL. Hydraulic
input
impedance to aorta and pulmonary artery in dogs. J Appl
Physiol 1964; 18: 134-40.
31. Bergel DH, Milnor WR. Pulmonary vascular impedance
in the dog. Circ Res 1965; 16: 401-15.
DE, Kendrick
JE,
Matson
GL,
Rideout
VC.
32. Dick
Measurement
of nonlinearity in the arterial system of the
dog by a new method. Circ Res 1968; 22: 101-l 1.
33. Wemple
RR, Mockros
LF. Pressure and flow in the
systemic arterial system. J Biomech 1972; 5: 629-41.
34. ORourke
MF, Taylor MG. Vascular
impedance
of the
femoral bed. Circ Res 1966; 18: 126-39.
35. Noble MIM, Gable IT, Guz A. Blood pressure and flow in
the ascending aorta of conscious dogs. Cardiovasc Res 1967;
1: 9-20.