Improved Ultrasonic Local Pulse Wave Velocity Estimation Via Optimal Time Fiduciary Point Combination and Time Delays...

Local pulse wave velocity (PWV) is key for quantifying arteriosclerosis progression, vital for early cerebrovascular diseases diagnosis and prevention. The ultrasonic transit time (TT) method calculates PWV by fitting time delays (TDs) of pulse wave (PW) propagation with the known distances. However, reflected waves and noise can bias the time fiduciary point (TFP) positioning, reducing the fitting performance and estimation accuracy. This work proposes a PGC (PSO-GA and Cook's distance) method using optimal multi-TFP (MTFP) combination and TDs fitness via PSO-GA for enhancing the TT-based local PWV estimation. First, PWs are calculated, and then a MTFP combination is randomly selected to calculate the TDs, whose outliers are detected by Cook's distance with a randomly selected threshold. Thereafter, the reciprocal of coefficient of determination for fitting is used as the fitness function to optimally update the MTFP combination and Cook's distance threshold via PSO-GA. Finally, the optimal MTFP and Cook's distance threshold are obtained to compute the improved PWV. The proposed PGC method is compared with the traditional method without optimal processing. The results show that the PGC-based mean and standard deviation of the normalized root mean squared errors between the estimated and preset PWVs decreased from 10.28 ± 2.51 to 5.96 ± 1.41% in simulation experiments, and the coefficient of variation for the measured PWVs decreased from 11.87 to 8.53% in in vivo experiments. In conclusion, the PGC method demonstrated improved accuracy in simulations and enhanced repeatability with reduced variability in vivo, offering the potential to provide more accurate diagnostic information for cerebrovascular diseases.