Session P34.5

Novel Method of Automatic Auscultation for Blood Pressure Measurement Using Pulses in Cuff Pressure and Korotkoff Sound

DK Park*, HS Oh, JH Kang, S Kim,
IY Kim, YJ Chee, JS Lee

Hanyang University
Seoul, Korea

The gold standard of Non-Invasive Blood Pressure (NIBP) measurement is the auscultation method using Korotkoff sound (K-sound). But it requires trained observer and there is difference in reading values between observers. There are some automatic algorithms to overcome these problems. In automatic algorithm, we have to select the proper epoch of K-sound for each heart beat to discriminate noise by motion artifacts or other reason. One of the solutions is using electrocardiogram (ECG) but it needs additional measurement apparatus other than microphone. In this study, we propose new technique to select heart beat epoch, and locate the timing of appearance and disappearance of K-sound.
We measured the cuff pressure and the microphone signal while deflating 2~3 mmHg/s. Using the band pass filter and morphological analysis of cuff pressure, we can locate the epoch of systolic period. The sound for these epochs is converted into frequency domain and we can get the total power. After plotting the power in each heart beat against cuff pressure, we decided systolic blood pressure (SBP) point when there was abrupt jump from the baseline. Diastolic Blood Pressure (DBP) was decided when there was the biggest drop in power spectrum ration of successive beats. For the validation of our suggested method, well trained staff assessed SBP/DBP according to EHS (European Hypertension Society) recommendation.
The BLAND ALTMAN plot was shown to review overall performance of our suggested algorithm for ten subjects with 10 times of measurements under different conditions for each subject. The mean differences between the automatic method and traditional auscultation method were 2.0 mmHg for SBP and 3.1 mmHg for DBP. We also took the paired t-test with the two results, and there was no statistically significant difference between these two methods.

(Abstract Control Number: 241)