Towards model-assisted clinical decision making in cardiac disease

Diagnosis of cardiac disease in the clinic remains difficult due to limited possibilities to extract data from a patient and problems in translating clinical data into the underlying pathological tissue properties. Consequently, diagnosis is based on combining limited patient information with general knowledge on pathophysiology, past experience with similar patients, and clinical guidelines. This approach is also known as evidence-based medicine and the outcome represents the best diagnosis and treatment selection for the average patient with similar symptoms, but not necessarily represents the best choice for the individual patient at hand. Information on myocardial deformation, acquired through ultrasound or magnetic resonance imaging, may contribute to more patient-specific decision making.

This presentation will address the possibilities to assess pathology from deformation patterns, either directly from the deformation fields, or indirectly by interpreting these fields using mathematical models of cardiac function. The last option involves the transformation of generic models of cardiac function into patient-specific models in an inverse analysis, by iteratively tuning model properties for maximal agreement between clinically observed and model predicted deformation. Since the clinical measurements are subject to measurement uncertainties, the outcome of the inverse analysis will be uncertain as well, and might even be non-unique. We envision that the inverse analysis may benefit from the use of models of cardiac growth and remodelling, that mimic the normal physiological processes through which the tissue adapts to changes in mechanical loading conditions. These few-parameter models might automatically generate spatial maps of model parameter values and thus substantially reduce the number of model parameter values that needs to be estimated separately.