Quantitative analysis of microvascular oscillations and their alterations

The microcirculation is regulated by the complex, and inter-dependent activity of neurogenic, myogenic and endothelial control mechanisms, which dynamically adapt the spatial distribution of the cardiac output in order to meet the local metabolic and thermal demands of the organism. Owing to these physiological mechanisms, and the characteristic waves arising centrally from the cardiac cycle and the rhythmic respiratory activity, the basal microvascular blood flow exhibits separate spontaneous non-stationary components within established frequency ranges. The cutaneous microcirculation provides a unique window on the functional state of the systemic microcirculation and, thanks to its accessibility, permits the non-invasive monitoring of the physiological processes underlying these cardiovascular oscillations. Furthermore, the skin represents a suitable target for testing sympathetic and endothelial vasomotor pathways by means of acute thermal or biochemical stimuli. This has been possible since the introduction of optical techniques such as laser Doppler flowmetry (LDF), able to record data of tissue perfusion at high temporal resolution. The work outlined in the present thesis focuses on the development and application of biosignal processing methods for the analysis of LDF signals of microvascular flow, that might support the early identication of a decline in vascular function and, thus, might have clinical potential in the non-invasive diagnosis and monitoring of subjects with metabolic disease and/or at risk of cardiovascular complications. Hence, the research activities relied on the analysis of LDF perfusion signals, recorded from healthy control subjects, and patients affected by diabetes mellitus. To this end, a wavelet-based time-frequency analytical framework has been adopted for disentangling the physiological components of the microvascular perfusion, since it enables the parallel characterization of the cardiac dynamics and the slower non-stationary fluctuations produced by the local vasomotion. Moreover, a model-based approach has been implemented with the aim to describe the composite hyperaemic response of the cutaneous microcirculation to local thermal stimulation, and to assess the potential impairment of the neurovascular and endothelial mechanisms involved in this thermoregulatory feedback loop. Furthermore, peripheral pulse waveforms are intimately related to the physical properties of the arterial tree within which they propagate; thus, their contour might carry information for the non-invasive diagnosis of vascular alterations. In this regard, a new pulse decomposition algorithm has been developed for deriving an accurate reconstruction of the cardiac pulsatility of LDF signals, via a linear combination of Gaussian bases. This technique paves the way for the robust extraction of physiologically meaningful features: in particular, two biomarkers of large artery stiffness and peripheral wave reflection have been investigated. The present strategy has been successfully applied to the automatic detection of vascular ageing and, moreover, to the automatic assessment of respiratory-vascular couplings and, particularly, the transient neurogenic vasoconstrictive pathway, triggered by a rapid inspiratory manoeuvre.