The 12 th edition of the international Functional Imaging and Modeling of the Heart (FIMH, https://fimh2023.sciencesconf.org/) conference was held in Lyon, France, on June 19-22 2023. FIMH is a biennial scientific event that aims to integrate state-of-the-art research and novel development efforts in the fields of cardiovascular imaging, image analysis, and heart modeling, both from the methodological and clinical sides. With over 130 registered participants, 20 oral presentations and 52 posters published in the Lecture Notes in Computer Science (LNCS -FIMH-23, volume 13958, Springer) (https://link.spri nger.com/book/10.1007/978-3-031-35302-4), this edition was an exciting return to Lyon 20 years after the second edition of the conference in 2003. The program also featured 3 associated workshops and 1 data challenge. This special issue of the Medical Image Analysis journal introduces 5 papers that were collectively selected among the presentations during the conference by the scientific and organizing committees. Romitti et al., introduce a Cellular Automaton (CA) model tailored to replicate atrial electrophysiology in different stages of Atrial Fibrillation (AF), including persistent AF. Their CA model allowed a significant decrease in computing time compared to the biophysical solver while presenting very promising performance in predicting AF inducibility. Banduc et al. propose a classification method to predict AF inducibility in patient-specific cardiac models without additional simulations. The approach introduces the fibrotic kernel signature, a set of features faster to compute than a single AF simulation, which, when paired with machine learning classifiers, can predict AF inducibility in the entire domain within clinical timelines. Still on the AF management, Qureshi et al. leverage patient-specific MRI-based computational fluid dynamic modeling of left atrial flow coupled with kinetics equations for blood constituent clotting proteins, leading to an innovative risk stratification score for thrombus formation in AF patients. Khaledian et al. propose a subject-specific fluid-structure simulation framework based on the immersed boundary method to produce accurate simulations of mitral valve closure and maps of contact reflecting the quality of the closure. They extracted valve geometries from 3D CT data of porcine hearts, and investigated both isotropic and anisotropic valve models. The latter were shown to represent the physiological characteristics of the valve tissue more accurately. Finally, in the context of congenitally corrected transposition of the great arteries (ccTGA), and its treatment through a double switch operation (DSO), Gusseva et al. propose a computational modeling framework to reconstruct left (LV) and right-ventricular (RV) pressurevolume loops and gather model-derived mechanical indicators of