Electrophysiology (EP) study is an invasive and time-consuming procedure that maps the heart activity and evaluates the inducibility of ventricular tachycardia (VT). Unfortunately, this diagnostic procedure often fails to identify patients at risk of potentially lethal VT in scar-related patients. Advances in computer modelling have enabled virtual patient-specific simulations of VT using cardiac imaging to delineate heart geometry and its structural changes (e.g., fibrosis). In this study, we build upon existing work wherein in-silico induction of VT was obtained through an EP model personalised from CT-defined wall thickness (WT). We propose an automated framework for multi-modal parameterisation of the EP model by combining information derived from CT scans and 12-lead ECGs, aiming to fine-tune the model parameters based on electrical features extracted from recorded ECGs. Myocardial WT was used for baseline parameterisation, then the electrical wave propagation through the heart and the surface 12-lead ECG signals were simulated. We compared the simulated and measured ECGs recorded in sinus rhythm and optimised the model configuration in two stages: i) early onset approximation and ii) EP model parameter adjustment, using the Covariant Matrix Adaptation Evolution Strategy (CMA-ES). The optimisation framework was tested on a small patient dataset (n = 6). Compared with the recorded ECG, the optimised model achieved up to \(89.81\%\) accuracy in QRS peaks detection, with an average QRS duration error of 5.93 ms—an improvement of 31.49 ms over the initial baseline parameters. We also tested the optimised parameters on two patients, where electro-anatomical maps of VT were available. The results show that the updated parameters are able to correct the VT cycle length discrepancy reported in the reference study and successfully induced virtual VT with cycle length and pattern closer to those recorded, as compared to the baseline parameters.