Elliptical method for time-domain quantification of the arterial baroreflex

Abstract

The quantification of the cardiovagal baroreflex sensitivity BRS over postural changes has been regarded as a valuable prognostic indicator for evaluation of the current state of autonomic functions. The usual time domain assessment is based on the linear regression of both systolic blood pressure values PS and cardiac period values RR during segments of increasing or decreasing trend. These fluctuations can occur spontaneously or be induced, e.g. by vasoactive drugs.This work proposes a novel method using image processing of the area defined by rise and fall segments. The best fit ellipse to this area corresponds to the bivariate normal distribution of the pixel area and characterizes the spontaneous BRS. These segments are analysed both individually and as cycles, which are composed of consecutive sequences of opposing trend. The cycle analysis allows the quantification and characterization of changes in the RR-PS relation related to the direction of pressure change (hysteresis) behaviour as change in working point and steepness.The study was performed on the EuroBaVar dataset (21 subjects in supine and standing positions during 10 to 12 minutes). The data was composed of the blood pressure signal in the finger artery and the 2 lead electrocardiogram. Additionally, a Graphical User Interface including both regression and ellipse method as well as spectral analysis was implemented, allowing the cardiovagal baroreflex computation for any dataset.Ellipse method results as BRSe were compared with results from the regression method BRSr. BRSe mean value accurately reflects the sensitivity decrease due to postural change (14.9 and 7.2 ms/mmHg, p < 0.005) verifying the results obtained for BRSr (respectively, 14.8 and 6.8 mmHg, p < 0.005). BRS estimates from the two methods were correlated with r > 0.98 in both positions. The characteristic ellipse median inclination with respect to the PS axis was steeper in supine position (86.4°) compared to standing (81.9°) and the median elongation was higher in supine by about 46%. The median magnitude of overall hysteresis was 48% superior in supine position.With the proposed method, hysteresis can be characterized using spontaneous (i.e. not pharmacologically induced) changes in PS, providing a potential tool for future baroreflex characterization. On the other hand, this characterization heavily relies on the number of cycles found (only 4 subjects in this dataset showed a significant number of cycles) and further approval with other datasets is necessary.