Aortic Pressure Estimation Using Blind Identification Approach on Single Input Multiple Output Nonlinear Wiener Systems

• Published in: IEEE transactions on biomedical engineering 2018-06, Vol.65 (6), p.1193-1200
• Main Authors: Patel, Amit M., Li, John K-J., Finegan, Barry, McMurtry, Michael S.
• Format: Article
• Language: English
• Subjects: Aorta; Aortic pressure; arterial system model; Arteries; Biomedical measurement; Cardiovascular diseases; Finite impulse response filters; FIR filters; Heart diseases; Hemodynamics; Kernel; Mathematical model; Maximum likelihood detection; Nonlinear systems; Pressure; Pressure measurement; System dynamics; System identification; transfer function; Viscoelasticity; Waveforms
• ISSN: 0018-9294; 1558-2531
• DOI: 10.1109/TBME.2017.2688425

Aortic pressure (P a ) is important for diagnosis of cardiovascular diseases, but it cannot be directly measured by noninvasive means. We present a method for its estimation by modeling arterial system as multichannel Weiner system with linear finite impulse response filter accounting for larger arteries transmission channel and nonlinear memoryless function block accounting for all nonlinearities due to narrowing of arteries, branching and visco-elastic forces. With this structure when pressure waveforms are measured from two distinct peripheral locations, multichannel blind system identification (MBSI) technique can be used to estimate common input pressure signal or P a . Nonlinear MBSI method was employed on previously acquired human hemodynamic measurements (seven datasets); results show P a can be accurately derived. This method by nature is self-calibrating to account for any interpersonal, along with intrapersonal, vascular dynamics inconstancy. Besides Pa estimation, the proposed MBSI method also allows extraction of system dynamics for vascular channels.