Biomedical Engineering Reference
In-Depth Information
structures, we may suspect that this is due to the lumped characterization in model
Extended as opposed to Mead or RLCES, where part of this resistance's effects are
transmitted to other model parameters.
In DuBois, tissue compliance correlated well with physiologically expected val-
ues and pathology. Bronchial compliance
C
b
correlated well between models Mead
and RLCES for asthma and COPD cases, but not in healthy subjects. We suspect this
may be due to inaccurate partitioning of wall and lung compliance. Nevertheless, the
values for
C
b
in asthma are very close to the reported values by Van Noord, where
C
b
=
0
.
005 l/cmH
2
O and although the values for
C
e
are somewhat lower than ex-
pected, they are similar to what other authors report [
29
,
39
,
93
]. Airway inertance
correlated well between model structures DuBois, Mead, Extended, RLCES, as well
as in relation to expected pathology.
From the point of view of reported total errors, we may add that for all subject
groups the best performance was given by Mead's model and highest errors for
viscoelastic model. We may conclude that Mead's model is still the structure with
least errors in parametric estimations. However, the newly proposed integer-order
model RLCES gives similar total errors with less number of model parameters. This
is indeed an advantage when a specific characterization is not intended, but merely a
clear-cut within subject population for preliminary diagnosis. It is clear that if more
specific information is required, the Mead model must be employed. Nonetheless,
the model structure in Mead may not necessarily be optimal, for it over-estimates
lung compliance (in healthy and asthmatic).
It is also noticeable that the total errors given by the CP5 model are compara-
ble to the ones given by the Mead model. The main advantage of the CP5 model
structure over the Mead model structure is its reduced number of parameters to be
identified. The reason for giving such good estimations is that the fractional or-
der captures in a more accurate and flexible way the frequency dependence of the
complex impedance. It also seems that the model from (
3.10
) gives most accurate
estimates for the COPD case. For healthy persons, the airway resistance is very low.
The reason for such low values is that part of dissipation properties are captured by
the fractional orders [
118
]. This observation suggests eliminating the term
R
r
from
(
3.10
), but this issue will be revisited later in this topic.
3.4 Summary
This chapter introduced the basic principles of estimating the respiratory impedance
from measurements performed with the forced oscillation lung function test. The
non-parametric identification of the respiratory impedance has been derived and
presented by means of spectral analysis. Next, a comparison of most representative
parametric models from literature for assessing respiratory input impedance shows
that FO models are more efficient than integer-order models in capturing frequency-
dependent impedance values variations. This motivates the development of the next
chapters, where theoretical models will be derived and it is shown that convergence
of these models for many airways will lead to the appearance of FO terms in the
lumped model structure.
