2025 Theses Doctoral

Full and Multi-Cycle Clinical Electromechanical Wave Imaging for Arrhythmogenic Tissue Characterization and Arrhythmia Treatment Monitoring

Tourni, Melina

Cardiovascular disease remains the leading cause of death worldwide, contributing significantly to morbidity, hospitalizations, and reduced quality of life. The complexity of cardiac function makes effective diagnosis and treatment particularly challenging. The heart comprises four interdependent systems—electrical, mechanical, valvular, and vascular—each essential to overall function. Disorder in any one of these systems can impair the others, underscoring the need for comprehensive, integrative approaches to cardiac assessment and therapy.

Robust treatment of atrial arrhythmias such as atrial fibrillation (AF), atrial flutter (AFL), and ventricular arrhythmias such as premature ventricular contractions (PVCs) and Wolf-Parkinsons-White (WPW) syndrome represent major clinical challenges. AF alone affects over 33 million people worldwide and contributes significantly to stroke, heart failure, and sudden cardiac death.

Moreover, valvular dysfunction, such as Mitral Valve (MV) disease, may yield electromechanical deficits that trigger arrhythmogenesis. Despite this burden, current diagnostic and treatment guidance tools remain invasive, spatially limited, or insufficiently precise. There is, therefore, a critical need for non-invasive, high-resolution imaging techniques that can dynamically characterize electromechanical dysfunction, automatically identify arrhythmogenic substrates, and inform real-time, clinical decision-making at the bedside.

This dissertation addresses that need through the advancement, validation, and clinical translation of two ultrasound-based imaging modalities—Electromechanical Cycle Length Mapping (ECLM) and Electromechanical Wave Imaging (EWI)— enabling full- and multi-cycle, real-time mapping of activation and recovery electromechanical functionality of the heart. Moreover, EWI is applied beyond the current high-frame-rate, clinically restricting operational setup, to being implemented on a portable, FDA-approved ultrasound platform for the first time.

ECLM is first validated as a non-invasive method for mapping atrial electromechanical cycle lengths (CLs) and detecting arrhythmic substrates in patients undergoing direct current cardioversion (DCCV). In a cohort of 45 subjects, including AF (n=21), AFL (n=9), and healthy controls (n=15), ECLM-derived atrial CLs showed strong agreement with P-wave intervals from surface electrocardiography (ECG) (?2=0.96). Two novel electromechanical metrics were introduced:%ACL, the percentage of atrial myocardium with arrhythmic CLs (≤333 ms), and CL dispersion, the standard deviation of CLs across the atria. These measures significantly distinguished arrhythmic versus post-cardioversion sinus rhythm states and were found to independently predict short-term (1-day) and long-term (1-month) DCCV outcomes (p<0.01). In two cases, ECLM is employed longitudinally across three-time points of consecutive AF treatment attempts to track AF progression and recurrence, demonstrating its potential for monitoring atrial arrhythmia burden and individualizing treatment over time.

ECLM is further applied in catheter ablation therapy to localize AF triggers and AFL circuits and stratify AF ablation responders. In patients with typical AFL (n=7), pre-ablation ECLM successfully identified the cavotricuspid isthmus as the primary reentrant circuit, with electromechanical CLs correlating strongly with both ECG-derived atrial CLs (?2=0.94) and intracardiac electrograms from electroanatomical mapping (EAM) (?2=0.87). In AF patients (n=15), 3D-rendered ECLM risk maps were developed, classifying local atrial regions between arrhythmic-variable, arrhythmic-synchronous, sinus-variable, and sinus-synchronous zones, capturing spatial heterogeneity and organization of electromechanical activity. Pre-ablation non-responders exhibited significantly higher %ACL (19.20± 13.27) and CLD (158.40± 57.12 ms) compared to responders (15.09± 13.56, p<0.05; 105.60± 42.37 ms, p<0.01, respectively) at baseline. An exhaustive regression model combining history of vascular disease, ECLM CLD and %of arrhythmic-variable regions of myocardium —achieved an AUC of 1.00, compared to 0.69 for the CHA2DS2-VASc score, demonstrating the superior prognostic power of ECLM-derived risk quantification.

Next, EWI is employed beyond arrhythmia mapping to assess electromechanical activation and recovery across the entire cardiac cycle in patients with MV disease. In an open-chest canine model, EWI confirmed, for the first time, the propagation of electromechanical waves through the MV leaflets. Clinically, EWI diastolic recovery mapping is developed, and full-cycle EWI is performed in patients with mitral valve prolapse (MVP, n=7), mitral regurgitation without prolapse (MR-only, n=5), and healthy controls (n=4), revealing significant electromechanical alterations in diseased subjects. In MVP patients, left ventricular (LV) activation is found delayed (76.0±12.5 ms) relative to controls (47.6±2.6 ms, p=0.0013), particularly at the papillary muscle insertions. Both MVP and MR-only groups exhibited prolonged recovery intervals compared to controls (MR-only: 277.0±27.6 ms, Controls: 248.7±10.4 ms, p=0.0395). In one clinical arrhythmogenic MVP case, EWI mapped the onset of a PVC beat on the region of delayed sinus activation and extended recovery, co-localizing with the myocardial segment affected by the prolapse, thus confirming the method’s potential to non-invasively characterize arrhythmogenic MVP.

To support clinical integration, EWI is adapted to operate at standard clinical framerates. A machine learning (ML) pipeline is developed to automate zero-crossing (ZC) detection in low-framerate strain data, using a Random Forest Classifier trained on high-framerate reference maps. In six (n=6) WPW syndrome patients imaged post-ablation, high-framerate raw ultrasound data were decimated to simulate low-frame-rate EWI datasets. The ML-based method achieved robust ZC selection at lower framerates, showing strong agreement with high FR ground truth data. Following, the EWI algorithm is adapted to perform on lower framerate clinical datasets, and feasibility is explored with the FDA-approved Terason uSmart® 3200T NexGen system, marking the first clinical deployment of EWI on a portable ultrasound scanner. Clinical-EWI (cEWI) successfully and accurately characterized the expected electromechanical activation wave propagation in one pre-clinical pacing animal model (n=1), one healthy volunteer (n=1), and two WPW patients (n=2). In both WPW cases, cEWI correctly localized the accessory pathway with higher accuracy than the 12-lead ECG alone, in an automated and fast manner.

Altogether, this dissertation establishes ECLM and EWI as reliable and clinically translatable tools for non-invasive mapping of cardiac electromechanical activity. By translating these methods onto already clinically approved ultrasound platforms and automating key processing steps, this work lays the foundation for real-time, patient-specific imaging of arrhythmogenic tissue in both pre-procedural planning and long-term care. Laptop-based cEWI has the potential to transform non-invasive electrophysiology characterization, making arrhythmia diagnosis and therapy guidance more accessible, precise, and personalized.