Aims: Patient-specific cardiac models can help improving understanding, diagnosis and therapy. This approach has been quite successful over the last decade in academic studies. However, there are still only very few clinical applications using such models. In this presentation, I will survey some academic results of image-based and electrophysiology models. Then discuss the potential explanations for their limited spread. Finally, suggestions for wider acceptance will be proposed.
Methods: Medical imaging and computer modelling have seen enormous progress over the last decades. This enabled the development of 3D electromechanical models of the heart. Several strategies for patient-specific parameter estimation were proposed but still challenging to deploy in clinical applications. In the recent years, machine learning has seen a huge increase of interest and such methods could help alleviating the current difficulties of personalised electromechanical models.
Results: Quantitative results on the use of personalised electromechanical models for different clinical applications will be presented. Several ways to leverage on machine learning will be detailed.
Conclusion: Patient-specific cardiac models have now enough maturity to enter the field of clinical applications. This is demonstrated by the important effort in this field by major healthcare companies. However, there is still a critical clinical validation to achieve as well as a seamless integration in clinical workflows in order to reach acceptance from clinicians.