不同采样频率对脉搏波形参数计算的准确性研究

Accuracy study of using different sampling frequency for the calculation of pulse contour parameter

目的比较不同采样频率获取的脉搏波形参数的差异,研究低采样频率对脉搏波形参数计算结果的影响,以探讨低采样频率的穿戴式心血管功能测试仪是否影响心血管参数的准确性。方法利用平面张力法获取100例颈动脉脉搏数据(采样频率1000 Hz),通过MATLAB重采样至500 Hz、250 Hz和128 Hz,分别利用波形分析技术提取包括中心动脉收缩压(carotid systolic blood pressure,c SBP)、增长指数(augmentation index,AIx)和心内膜下心肌活力率(subendocardial viability ratio,SEVR)等参数。以1000 Hz的采样频率作为参考,对比不同采样频率计算的脉搏波形参数差异。结果 500 Hz、250 Hz和128 Hz颈动脉数据计算的波形参数与其参考值之间具有非常高的相关性(r≥0.993,P<0.001),BlandAltman分析显示95%的差异点位于一致界限范围之内,表明具有很好的一致性。配对t检验结果显示500 Hz、250 Hz和128 Hz采样频率计算的颈动脉波形参数与参考值之间都存在显著差异(P<0.05),但平均差异很小,不具有临床诊断意义。结论对于获取脉搏波形参数,低采样频率可以达到与高采样频率一样准确的结果。但低采样频率降低了对硬件的要求,更能满足穿戴式心血管功能监测设备的要求。

Objective This paper evaluates the difference of pulse contour parameter obtained from different sampling frequency and studies the effect of low sampling frequency for the calculation of pulse contour parameter in order to explore whether low sampling frequency of wearable cardiovascular function monitoring equipment affects the accuracy of the cardiovascular parameter. Methods A total of 100 subjects were studied. Carotid waveforms were measured by applanation tonometry with a sampling frequency of 1000 Hz. All the carotid waveform data were converted offline to 500 Hz,250 Hz and 128 Hz by MATLAB. Waveform analysis technology was utilized to extract carotid systolic blood pressure（ c SBP）, augmentation index（ AIx）,subendocardial viability ratio（ SEVR）, and other parameters. The parameters obtained from 500 Hz,250 Hz and 128 Hz were compared with the parameter obtained from 1000 Hz. Results There was a significantly high correlation between each parameter and its referencestandard（ r ≥ 0. 993,P〈0. 001）. Bland-Altman plots showed a good consistency,since 95% of the difference points were located in the limits of agreement. Paired t-test results indicated that each parameter obtained from 500 Hz,250 Hz and 128 Hz had significant difference with its reference standard（ P〈0. 05）.However,the average difference was small so that there was no clinical diagnosis significance. Conclusions For the calculation of pulse contour parameter,low sampling frequency can achieve the same accurate results as high sampling frequency. And low sampling frequency could reduce the requirement for hardware to better meet the requirement of wearable cardiovascular function monitoring equipment.