Comparison of RootMUSIC and Discrete Wavelet Transform Analysis of Doppler Ultrasound Blood Flow Waveforms to Detect Microvascular Abnormalities in Type I Diabetes

Abstract

The earliest signs of cardiovascular disease occur in microcirculations. Changes to mechanical and structural properties of these small resistive vessels alter the impedance to flow, subsequent reflected waves, and consequently, flow waveform morphology. In this paper, we compare two frequency analysis techniques: 1) rootMUSIC and 2) the discrete wavelet transform (DWT) to extract features of flow velocity waveform morphology captured using Doppler ultrasound from the ophthalmic artery (OA) in 30 controls and 38 age and sex matched Type I diabetics. Conventional techniques for characterizing Doppler velocity waveforms, such as mean velocity, resistive index, and pulsatility index, revealed no significant differences between the groups. However, rootMUSIC and the DWT provided highly correlated results with the spectral content in bands 2-7 (30-0.8 Hz) significantly elevated in the diabetic group (p < 0.05). The spectral distinction between the groups may be attributable to manifestations of underlying pathophysiological processes in vascular impedance and consequent wave reflections, with bands 5 and 7 related to age. Spectral descriptors of OA blood velocity waveforms are better indicators of preclinical microvascular abnormalities in Type I diabetes than conventional measures. Although highly correlated DWT proved slightly more discriminatory than rootMUSIC and has the advantage of extending to subheart rate frequencies, which may be of interest.