Computer-Aided Detection of Fiducial Points in Seismocardiography through Dynamic Time Warping

Abstract

Accelerometer-based devices have been employed in seismocardiography fiducial point detection with the aid of quasi-synchronous alignment between echocardiography images and seismocardiogram signals. However, signal misalignments have been observed, due to the heartbeat cycle length variation. This paper not only analyzes the misalignments and detection errors but also proposes to mitigate the issues by introducing reference signals and adynamic time warping (DTW) algorithm. Two diagnostic parameters, the ratio of pre-ejection period to left ventricular ejection time (PEP/LVET) and the Tei index, were examined with two statistical verification approaches: (1) the coefficient of determination (R2) of the parameters versus the left ventricular ejection fraction (LVEF) assessments, and (2) the receiver operating characteristic (ROC) classification to distinguish the heart failure patients with reduced ejection fraction (HFrEF). Favorable R2 values were obtained, R2 = 0.768 for PEP/LVET versus LVEF and R2 = 0.86 for Tei index versus LVEF. The areas under the ROC curve indicate the parameters that are good predictors to identify HFrEF patients, with an accuracy of more than 92%. The proof-of-concept experiments exhibited the effectiveness of the DTW-based quasi-synchronous alignment in seismocardiography fiducial point detection. The proposed approach may enable the standardization of the fiducial point detection and the signal template generation. Meanwhile, the program-generated annotation data may serve as the labeled training set for the supervised machine learning.

Comparison between Chest-Worn Accelerometer and Gyroscope Performance for Heart Rate and Respiratory Rate Monitoring

“…Eleven healthy volunteers were enrolled in this study. To each volunteer, a single IMU sensor (Xsens DOT, by Xsens), embedding a triaxial ACC (full scale ± 16 g) and a triaxial GYR (full scale ± 2000°/s), was fixed at the fifth-left intercostal space in the midclavicular line of the mitral valve [ 27 ]. The small size (36 × 30 × 11 mm) and mass (11.2 g) of this device gave the sensor optimal adherence to the subject’s chest without introducing discomfort to the user and ensuring a good coupling effect between the sensor and the body.…”

A dynamic time warping-based algorithm for characteristic peak group localization from ballistocardiograms

“…The DTW algorithm can effectively align two time series with different lengths or phase shifts [10] . Suppose there are two time series, R and T, with lengths n and m. First, we need to construct a distance matrix W of size n×m, where each element w i,j is the Euclidean distance between point r i in the time serie R and point t j in the time serie T:…”

Cardiac Time Intervals Derived from Electrocardiography and Seismocardiography in Different Patient Groups

“…DTW has been previously applied to SCG signals for detection of fiducial points in smaller sample sizes [6] [10]. The method used to group the cardiac cycles into similar and dissimilar beats is based off prior work in [8].…”