19 Complexity-Based Analysis of Microvascular Blood …
299
19.4.3 Application of Complexity-Based Analysis of Laser
Doppler BF Signals
Non-linear methods such as entropy and complexity analysis have been increasingly used to explore the regularity and the randomness of blood flux signals derived
from microvascular networks as a possible predictor of functional dysregulation,
both in human cohorts and in animal models of human disease. One of the first
of these studies was by Tigno et al. [85] who investigated the complexity of LDF
signals from nondiabetic, prediabetic and diabetic primates. They report an LZC
that decreased with progression of diabetes [85]. Similarly, Chipperfield et al. [18]
showed that in humans with and without type 2 diabetes mellitus LZC of the LDF
signal reduced as CVD risk increased. Together, these findings suggest that the impact
of diabetes on spontaneous oscillations in the bio-signal derived from microvascular
blood flow may be reflected in the complexity of the low frequency flowmotion
activity rather than, or as well as, in their oscillatory power content. Contrary to
the studies described above using LZC, Hsiu et al. [37] using approximate entropy
to compare beat-to-beat cardiac rhythms in the LDF signal in nondiabetic, prediabetic and diabetic humans have shown an increased signal complexity in people with
diabetes. However, the increase in complexity reported by Hsiu et al. [37] relates
to complexity between consecutive high frequency heart beats rather than to low
frequency (flowmotion) activity. To what extent the impact of diabetes on the microcirculation involves a decrease in complexity of the low frequency oscillations and
an increase in complexity of conducted signals such as that of the cardiac rhythm
requires further investigation.
Age-related changes in microvascular flows have been investigated by Humeau
et al. [39] using wavelet-based representations, with Hölder exponents to measure the
regularity of the LDF signal and sample entropy to assess its complexity. They showed
that endothelium-related activity decreased with age while microvascular perfusion
became more regular and less complex, although not significantly. Liao et al. [52]
using a modified SampEn algorithm similarly found that the LDF signal showed a
higher degree of regularity during thermal warming as compared to the baseline in
both young (mean age 27 years) and older (mean age 72 years) individuals. The LDF
signal also showed a higher degree of regularity in the older group as compared to
the young group, attributed in part to enhanced cardiac oscillations.
SampEn of LDF signals has also been used to discriminate between normal and
abnormal skin microvascular function in diseases affecting digits, such as Raynaud’s
phenomenon and systemic sclerosis [27]. These authors found that baseline entropy
was significantly increased in patients with systemic sclerosis compared to those with
Raynaud’s phenomenon and with controls, on the finger pad but not on the forearm;
consistent with the pathophysiology of the disease, which predominantly affects the
digital microcirculation.
Overall, these studies appear consistent with the premise that a greater complexity
(variability) of the blood flux signal may indicate a more effective microvascular
system, whereas a lower variability in microvascular activity corresponds to a loss of
299
19.4.3 Application of Complexity-Based Analysis of Laser
Doppler BF Signals
Non-linear methods such as entropy and complexity analysis have been increasingly used to explore the regularity and the randomness of blood flux signals derived
from microvascular networks as a possible predictor of functional dysregulation,
both in human cohorts and in animal models of human disease. One of the first
of these studies was by Tigno et al. [85] who investigated the complexity of LDF
signals from nondiabetic, prediabetic and diabetic primates. They report an LZC
that decreased with progression of diabetes [85]. Similarly, Chipperfield et al. [18]
showed that in humans with and without type 2 diabetes mellitus LZC of the LDF
signal reduced as CVD risk increased. Together, these findings suggest that the impact
of diabetes on spontaneous oscillations in the bio-signal derived from microvascular
blood flow may be reflected in the complexity of the low frequency flowmotion
activity rather than, or as well as, in their oscillatory power content. Contrary to
the studies described above using LZC, Hsiu et al. [37] using approximate entropy
to compare beat-to-beat cardiac rhythms in the LDF signal in nondiabetic, prediabetic and diabetic humans have shown an increased signal complexity in people with
diabetes. However, the increase in complexity reported by Hsiu et al. [37] relates
to complexity between consecutive high frequency heart beats rather than to low
frequency (flowmotion) activity. To what extent the impact of diabetes on the microcirculation involves a decrease in complexity of the low frequency oscillations and
an increase in complexity of conducted signals such as that of the cardiac rhythm
requires further investigation.
Age-related changes in microvascular flows have been investigated by Humeau
et al. [39] using wavelet-based representations, with Hölder exponents to measure the
regularity of the LDF signal and sample entropy to assess its complexity. They showed
that endothelium-related activity decreased with age while microvascular perfusion
became more regular and less complex, although not significantly. Liao et al. [52]
using a modified SampEn algorithm similarly found that the LDF signal showed a
higher degree of regularity during thermal warming as compared to the baseline in
both young (mean age 27 years) and older (mean age 72 years) individuals. The LDF
signal also showed a higher degree of regularity in the older group as compared to
the young group, attributed in part to enhanced cardiac oscillations.
SampEn of LDF signals has also been used to discriminate between normal and
abnormal skin microvascular function in diseases affecting digits, such as Raynaud’s
phenomenon and systemic sclerosis [27]. These authors found that baseline entropy
was significantly increased in patients with systemic sclerosis compared to those with
Raynaud’s phenomenon and with controls, on the finger pad but not on the forearm;
consistent with the pathophysiology of the disease, which predominantly affects the
digital microcirculation.
Overall, these studies appear consistent with the premise that a greater complexity
(variability) of the blood flux signal may indicate a more effective microvascular
system, whereas a lower variability in microvascular activity corresponds to a loss of