AI-assisted Imaging and Prediction of Cardiac Arrhythmia Origins using 4D Ultrasound

点击翻译按钮获取中文摘要

PROJECT SUMMARY / ABSTRACT Cardiovascular disease is the major cause of mortality and morbidity worldwide. Despite significant progress in biomedical imaging, the imaging of heart rhythm disorders remains a major technological and scientific challenge. Consequently, the origins and mechanisms for the onset and progression of cardiac arrhythmias remain largely insufficiently understood. Patients suffering from cardiac arrhythmias have high recurrence rates and often require repeated therapeutic interventions, in part because adequate imaging of the processes underlying heart rhythm disorders has yet to be developed. The state-of-the-art for the diagnosis of heart rhythm disorders, such as atrial fibrillation or ventricular tachycardia, is catheter-based electro-anatomic contact mapping. However, catheter mapping is time-consuming and invasive, involving the insertion of electrodes into the heart’s chambers, where abnormal electrical activity triggering the heart’s irregular contractions is recorded on its surface. Because the measurements are superficial, they do not adequately capture the full, three- dimensional electrophysiological dynamics, which evolve underneath the surface and often have their origin inside the heart muscle. In this project, the applicant aims to develop a novel and radically different approach for the in-depth transmural imaging of heart rhythm disorders based on high-resolution 4D (time-resolved 3D) ultrasound and artificial intelligence (AI). Instead of imaging the heart’s electrical activity, the applicant will image the heart’s 4D deformation and use AI to predict the electrical phenomena from the deformation with the precision of high-resolution measurements. To achieve this ground-breaking goal, the applicant will generate an extensive high-resolution dataset, capturing the 4D electrical and mechanical dynamics of arrhythmic hearts, and train an AI to learn the complex relationship between the heart’s deformations and the electrophysiological wave phenomena that cause these deformations. The AI will become highly specialized in recognizing cardiac deformation mechanics and associating them with the corresponding underlying electrical arrhythmia morphology. The data will be generated in beyond-state-of-the-art voltage-sensitive ex vivo fluorescence imaging experiments with intact, isolated hearts, as well as during clinical imaging and in computer simulations. The high- risk approach, which preliminary data suggests is achievable, will be enabled by the applicant’s unique expertise in ex vivo imaging, which, combined with recent advancements in AI, could lead to a major breakthrough. Ultrasound-based imaging providing transmural 4D visualizations of cardiac arrhythmias in real-time would be transformative in cardiac electrophysiology and provide novel insights into many of the yet unseen processes underlying heart rhythm disorders. If successful, the entirely non-invasive imaging technique could greatly advance the diagnosis of heart rhythm disorders and be used to guide therapeutics, such as catheter ablation, more reliably and effectively.