A region-based cross-correlation approach for tonometric carotid–femoral Pulse Wave Velocity Assessment

Abstract

This work introduces a novel region-based algorithm to determine carotid–femoral Pulse Wave Velocity (cfPWV), decoupling the assessment of this biomarker from point-to-point feature extraction. A dataset of 75 healthy participants, previously recruited to compare the performances of a new instrument, Athos, with the gold standard for non-invasive PWV (SphygmoCor), was used to set up a new algorithm for determining the clinical cfPWV. The proposed approach locates and processes a specific window on the carotid and femoral signals. Cross-correlation is employed to compute the pulse transit time within the pulses. Finally, the cfPWV is assessed. A set of indicators has been defined to quantify the stability and reliability of the window used by the algorithm. The results obtained through the proposed algorithm, the Intersecting Tangent method, and the direct application of the cross-correlation technique on signal epochs have been systematically compared with outcomes derived from the SphygmoCor device. The retrieved results show a Pearson coefficient r = 0.96, P< 0.001, with a mean difference of 0.16m/s for the first case, r = 0.965, P< 0.001, with a mean difference of −0.07m/s for the second, and r = 0.82, P< 0.001, with a mean difference of −1.04m/s for the third. Despite the intersecting tangent method showing slightly better agreement when comparing cfPWV values with those derived from the SphygmoCor, the proposed approach yields significantly lower errors (P< 0.05) in the presence of different levels of added noise, demonstrating greater robustness in this context. This outcome, combined with the excellent accuracy, sustains the suitability of the proposed method for clinical applications.