The cardiac systolic mechanical axis: Optimizing multi-axial cardiac vibrations by projecting along a physiological reference frame

• Published in: Biomedical signal processing and control Vol. 59; p. 101933
• Main Authors: Cordero, Rafael, Feuerstein, Delphine, Joubert, Pierre-Yves
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01-05-2020; Elsevier
• Subjects: Accelerometer; Engineering Sciences; Hemodynamic; Reference frame; Seismocardiogram; Subcutaneous; Accelerometer; Reference frame; Seismocardiogram; Subcutaneous; Hemodynamic
• ISSN: 1746-8094; 1746-8108
• DOI: 10.1016/j.bspc.2020.101933

[Display omitted]
•Cardiac vibrations recorded from multi-axial accelerometers demonstrate significant anisotropy.•Few groups have accounted for or studied this anisotropy.•Projection along the transverse plane axis maximizing the first heart sound amplitude optimizes signal quality.•This axis is physiological and specific to each patient.•The anisotropy of cardiac vibrations likely represents an understudied indicator of hemodynamic performance.
Multi-axial accelerometers are invaluable for the analysis of anisotropic cardiac vibrations during the first-heart-sound (S1). However, most studies simply consider the multi-axial accelerometer components independently along the different anatomical axes and rarely properly account for this anisotropy. The objective of this study was to evaluate a simple method, intended for chronically implanted medical devices, to identify a physiological reference frame maximizing S1 vibrations in the transverse plane. We attached a tri-axial accelerometer on the tip of a cardiac pacing lead to the chest of 10 human volunteers and implanted it subcutaneously in one pig. The volunteers and pig underwent separate experimental protocols capturing a wide range of conditions. The acceleration components of the tri-axial accelerometer were projected along the identified reference frame and a frame perpendicular to this. These were denoted the cardiac systolic mechanical axis and perpendicular axis, respectively. Results showed that projection along the mechanical axis tended to maximize signal-to-noise ratio and minimize cycle-to-cycle morphological variability. Significant differences were also observed between the mechanical and perpendicular axes projections, demonstrating sensibility to orientation. Along the mechanical axis, signal-to-noise-ratio was up to 21.72 dB greater, and cycle-to-cycle variability was up to 7.47 dB smaller, than along the perpendicular axis. The transverse plane angle defining the reference frame was relatively similar throughout the volunteers (28.5–44.0° from the medial-lateral axis), demonstrating inter-subject homogeneity. Also, it varied little across the different conditions (15.6° mean standard deviation across the volunteers), demonstrating stability. In conclusion, the proposed algorithm optimized signal quality at low computational cost.