Optimization Framework for Patient-Specific Cardiac Modeling

• Published in: Cardiovascular engineering and technology Vol. 10; no. 4; pp. 553 - 567
• Main Authors: Mineroff, Joshua, McCulloch, Andrew D., Krummen, David, Ganapathysubramanian, Baskar, Krishnamurthy, Adarsh
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.12.2019; Springer Nature B.V
• Subjects: Aged; Algorithms; Animals; Biomechanical Phenomena; Biomechanics; Biomedical Engineering and Bioengineering; Biomedicine; Cardiac Resynchronization Therapy; Cardiology; Data acquisition; Diffusion Tensor Imaging; Dogs; Engineering; Finite Element Analysis; Heart Failure - diagnostic imaging; Heart Failure - physiopathology; Heart Failure - therapy; Hemodynamics; Humans; Image Interpretation, Computer-Assisted; Magnetic Resonance Imaging, Cine; Male; Middle Aged; Models, Cardiovascular; Numerical Analysis, Computer-Assisted; Optimization; Parameters; Patient-Specific Modeling; Pattern search; Predictive Value of Tests; Reproducibility of Results; Species Specificity; Tuning; Ventricular Function, Left; Cardiac biomechanics; Lumped-parameter circulation model; Patient-specific modeling; Optimization
• ISSN: 1869-408X; 1869-4098; 1869-4098
• DOI: 10.1007/s13239-019-00428-z

Purpose Patient-specific models of the heart can be used to improve the diagnosis of cardiac diseases, but practical application of these models can be impeded by the computational costs and numerical uncertainties of fitting mechanistic models to clinical measurements from individual patients. Reliable and efficient tuning of these models within clinically appropriate error bounds is a requirement for practical deployment in the time-constrained environment of the clinic. Methods We developed an optimization framework to tune parameters of patient-specific mechanistic models using routinely-acquired non-invasive patient data more efficiently than manual methods. We employ a hybrid particle swarm and pattern search optimization algorithm, but the framework can be readily adapted to use other optimization algorithms. Results We apply the proposed framework to tune full-cycle lumped parameter circulatory models using clinical data. We show that our framework can be easily adapted to optimize cross-species models by tuning the parameters of the same circulation model to four canine subjects. Conclusions This work will facilitate the use of biomechanics and circulatory cardiac models in both clinical and research environments by ameliorating the tedious process of manually fitting the parameters.