Implementation of advanced algorithms for non-invasive Pulse Wave Velocity extraction

Cardiovascular diseases globally are the leading cause of death. For this reason, it is necessary to recognize and prevent them. A possible way to accomplish this purpose is by analyzing the circulatory system; recently, experts discovered how an increase in arterial stiffness is an expression of vascular damage, regardless of the contribution of classic cardiovascular risk factors. Arterial stiffness refers to the capability of arteries to expand and contract during a heart cycle: it depends on aging, genetic factors, hypertension. When arteries become stiffer, the heart loses efficiency, and for this reason, the likelihood of stroke or heart attack increases. The Pulse Wave Velocity (PWV) is a parameter that allows acquiring information about arterial stiffness, arterial pressure, and vasomotor tone. It is the velocity at which the arterial pulse propagates between the cardiovascular system. The Pulse Wave Velocity can be computed through the ratio between two parameters: the distance between two sensors in correspondence of two arteries and the time needed for the pulse to propagate among them (Pulse Transit Time). In the non-invasive clinical analysis, PWV can be acquired with different methods: the most used is the applanation tonometry, followed by ultrasound and optical sensors. Over the years, the Polytechnic of Turin has led the development of a new system, based on applanation tonometry, called ATHOS. A couple of years ago, a partnership between the University of Strasbourg and the Polytechnic was born to improve PWV extraction. The goal is to replace the tonometers with a new type of graphene-based sensors. The ATHOS device is composed of an acquisition system based on two tonometers, a dedicated Printed Circuit Board (PCB) and a graphical interface that allows real-time qualitative visualization of the two pulses. In order to validate the feasibility of the graphene-based system, ten healthy subjects were enrolled in this study. They were asked to take part in data collection sessions in which the PWV was estimated through SphygmoCor (AtCor Medical, Sydney, Australia), considered as the reference device in the clinical environment, and ATHOS, equipped firstly, with tonometers, and then with graphene sensors. \\ From the analysis of the collected data, emerged that the algorithm used on tonometers, called Intersecting Tangent Point, gives critical issues with the graphene sensors. This thesis project aims to implement an algorithm that ensures PWV extraction from signals obtained from graphene sensors applied to the ATHOS system. The algorithm is based on signal windowing, such that after filtering the output signal, different statistical approaches were applied to extract the PWV. For this purpose, several techniques such as cross-correlation, least-squares difference and statistical phase offset were used. Although the three presented techniques give a good response, the best and most feasible results are obtained with the cross-correlation technique. Once the best performing technique was detected, two further improvements were implemented: the first one was based on the Hilbert envelope, while the second one was on interpolation. For the two latter cases, the results obtained compared to those with cross-correlation showed to be less performing. For this reason, it was taken into consideration the working algorithm based only on the cross-correlation technique.