From bits to bedside: entering the age of digital twins in cardiac electrophysiology

Figures

• 

  Figure 1. Calibration of a digital twin. Generalized workflow for the generation of a digital twin of cardiac electrophysiology. Clinical data acquired from a given patient including imaging (MRI, CT), sensor data (ECG, electro-anatomical mapping, body surface potential maps) and health data are used to build a virtual patient. Twinning workflows typically comprise two stages: (1) The anatomical twinning stage involves multi-label segmentation of clinical imaging data to generate computational meshes representing relevant anatomy, typically the heart or parts thereof, embedded in a torso. (2) The functional twinning stage involves functionalization, i.e. providing a reference frame for defining the space-varying parameter fields to be identified, the evaluation of a physiological forward model to compute predictions of the observed clinical data such as the ECG, and the measurement of the discrepancy between prediction and real physical observation. Parameter fields are then adjusted iteratively to reduce the gap between observations in the virtual and the physical space until the gap is deemed small enough such that the digital twin can be considered to behave sufficiently similar to the…

• 

  Figure 2. Clinical application of computational modelling for CRT. Top panel illustrates the different concepts and mechanisms of resynchronization therapy. Middle panel demonstrates computational modelling derived simulations of electrical activity and mechanical/hemodynamic response during resynchronization therapy. The lower panel presents the key insights that can be gained when applying computational modelling.

• 

  Figure 3. Clinical application of computational modelling in atrial fibrillation. Top panel illustrates clinical treatment strategies for managing atrial fibrillation. It includes catheter ablation, anti-arrhythmic drug therapy, and measures for stroke mitigation such as anti-coagulation. The middle panel demonstrates how computational modelling can inform AF treatment by showing that models can aid in understanding the arrhythmogenic substrate, planning model-based ablation strategies, and stratifying patients’ stroke risk. Lower panel presents the insights that can be derived from computational modelling in AF management, including improved understanding of AF mechanisms, optimized ablation targets, and better risk assessment for stroke.

• 

  Figure 4. Clinical application of computational modelling for ventricular tachycardia. Top Panel illustrates model-guided VT ablation with the left section illustrating the process of direct VT circuit simulation, where induction pacing sites are used to identify the ablation target automatically. The simulation provides a visual map of the induced VT activation time, highlighting critical regions such as the exit site and target areas for ablation. The right section depicts in silico pace mapping, which involves the use of computational models to simulate pacing an identify regions of high correlation to pinpoint VT sites of origin, guiding precise ablation targeting.