Resumen de Inverse estimation of the cardiac purkinje system from electroanatomical maps
Cardiovascular disease is the number one cause of mortality in the world, accounting for 17.7 million deaths each year, an estimated 31% of all deaths worldwide (World Health Organization (WHO) 2018). Ventricular arrhythmias are a major cause of sudden death, which accounts for approximately half of cardiac mortality. Some of those arrhythmias are attributed to the Purkinje network (PKN), which under certain conditions can generate both automatic and triggered focal rhythms, and its network configuration can sustain re‑entrant circuits. Focal Purkinje triggers can serve as initial points of ventricular fibrillation in a wide spectrum of patients.
The management of cardiac electrical diseases is an expanding clinical activity. New non-invasive imaging and mapping technologies, allow to acquire high resolution clinical images (MRI, CT) that can be used to localize and characterize pathological cardiac tissue. Furthermore, electroanatomical navigating (EAM) systems, can aid electrophysiologist to find the sources of arrhythmogenic activity or circuits maintaining arrhythmia, and eliminate them by radio-frequency ablation (RFA).
Despite all the technical advances, overall clinical outcome for those diseases is still perceived as suboptimal, with long-term treatment success rates in the range of 60 to 65%. Therefore, there is a compelling need to improve clinical outcomes for the benefit of the patients and the healthcare system.
The area of computational biophysical modeling has already started to penetrate in clinical environments in a few technologically advanced research oriented hospitals in the world. The main objective of these techniques is the development of realistic 3D models of different organs, such as the heart, that include, with a high degree of detail, genetic characteristics of the ionic currents, their mutations, the electrophysiological characteristics of the different cardiac cell types, the anatomical structure of cardiac tissues, and in general of the human body. Following, the models are used to simulate the heart function, e.g., electrophysiology, to try to stratify patients or improve therapy planning and delivery.
Computer-based approaches are still facing several challenges that prevent their complete penetration into clinical environments. Arguably, one of the most important obstacles is the time and expertise required to build a patient-specific model of the heart, even if all necessary clinical data are available. In that sense, one of the model components that has remained largely elusive to modelers has been the PKN, which is key for cardiac electrophysiology. The main reason is that due to its small dimensions there is no clinical technique with enough resolution to allow its visualization in vivo.
The main purpose of this thesis is to develop a methodology able to inversely estimate a reduced PKN of patient from his EAM. That involves, first, finding in the EAM the sources of electrical activation, so called Purkinje-myocardial junctions (PMJs), and, following, finding the structure that interconnects those PMJs and reproduces the patient sequence of activation. In summary, the main contributions of this thesis are: - Methodology to estimate the PMJs, or the sources of electrical activity, from a 3D representation of the ventricular endocardium provided by an EAM. The method developed can process directly the data acquired by an electrophysiologist in the Cathlab, re-annotate the time samples, and obtain the PMJ locations and activation times, explicitly considering noise in the samples.
- Methodology to estimate the patient PKN from the estimated PMJs, that is able to reproduce the patient's sequence of electrical activation with a minimal error. The method has been validated on synthetic EAMs as well as in 28 real EAMs, showing errors of a few milliseconds. In addition, an estimated PKN has been used to simulate the virtual ECG of a patient, showing a good match with the clinical one.
In conclusion, I have developed and validated a methodology that permits the estimation of a patient's PKN with small errors in the sequence of activation, that can be used to personalize biophysical simulations of the heart or aid electrophysiologist in the planning of RFA interventions.
