An Improved Model for the Assessment of Cutaneous Microcirculation in Type 1 Diabetes

Patients with Type I Diabetes may develop some complications during their lifetime as a result of impaired metabolic control; among others, some of these common complications concerns microvascular function. The possibility of evaluating the cutaneous microcirculation is considered an important translational model because it could offer a important opportunity to non-invasive evaluation of diabetes-related vascular disorders. In our previous works we demonstrated the alteration occurring in the vascular system reflect themselves in the shape of the peripheral waveform. In particular, we demonstrated the decomposition of the waveform as a sum of Gaussian curves provides a set of parameters that can be fed into a classifier that successfully discriminate between younger and older subjects. In the present work, we investigated the application of a different decomposition model, based on a sum of exponential pulses, for the analysis of microvascular alterations correlated with the presence of Type I Diabetes. The processing pipeline follow the basic architecture of our previous work: a pre-processing step segments each pulsation of the input signal and the removes the long-term trend. Afterward, a Least Squares fitting algorithm determines the set of optimal model parameters that best approximate each single cardiac cycle. Each vector of model parameters constitutes a compact representation of the pulse waveform; thus, any structural difference in the pulse waveform between controls and patient produces a corresponding alteration of model parameters. Finally, an example classifier is used to explore the feasibility of discriminating pathological and normal subjects by looking only at the peripheral waveform.