A low-pass differentiation filter based on the 2nd-order B-spline wavelet for calculating augmentation index

• Published in: Medical engineering & physics Vol. 36; no. 6; pp. 786 - 792
• Main Authors: He, Zijun, Zhang, Yongliang, Ma, Zuchang, Hu, Fusong, Sun, Yining
• Format: Journal Article
• Language: English
• Published: England Elsevier Ltd 01.06.2014
• Subjects: Adult; Aged; Algorithms; Arterial stiffness; Augmentation; B-spline wavelets; Carotid Arteries - physiology; Computer Simulation; Convolution; Diseases; Female; Fourier Analysis; Humans; Male; Mathematical analysis; Mathematical models; Middle Aged; Numerical differentiation; Pulse - methods; Pulse wave analysis; Radial Artery - physiology; Radiology; Shoulder point; Shoulders; Signal-To-Noise Ratio; Time; Vascular Stiffness - physiology; Waveforms; Wavelet; Wavelet Analysis; Young Adult; Pulse wave analysis; Convolution; B-spline wavelets; Arterial stiffness; Shoulder point
• ISSN: 1350-4533; 1873-4030
• DOI: 10.1016/j.medengphy.2014.02.007

Abstract The key point to calculate augmentation index (AIx) related to cardiovascular diseases is the precise identification of the shoulder point. The commonly used method for extracting the shoulder point is to calculate the fourth derivative of the pulse waveform by numerical differentiation. However, this method has a poor anti-noise capability and is computationally intensive. The aims of this study were to develop a new method based on the 2nd-order B-spline wavelet for calculating AIx, and to compare it with numerical differentiation and Savitzky–Golay digital differentiator (SGDD). All the three methods were applied to pulse waveforms derived from 60 healthy subjects. There was a significantly high correlation between the proposed method and numerical differentiation ( r = 0.998 for carotid pulses, and r = 0.997 for radial pulses), as well as between the proposed method and the SGDD ( r = 0.995 for carotid pulses, and r = 0.993 for radial pulses). In addition, the anti-noise capability of the proposed method was evaluated by adding simulated noise (>10 Hz) on pulse waveforms. The results showed that the proposed method was advantageous in noise tolerance than the other two methods. These findings indicate that the proposed method can quickly and accurately calculate AIx with a good anti-noise capability.