Cardiac conduction abnormalities can often lead to heart failure, stroke and sudden cardiac death. Heart disease stands as the leading cause of mortality and morbidity in the United States, accounting for 30% of all deaths. Early detection of malfunctions such as arrhythmias and systolic heart failure, the two heart conditions studied in this dissertation, would definitely help reduce the burden cardiovascular diseases have on public health and overcome the current clinical challenges. The imaging techniques currently available to doctors for cardiac activation sequence mapping are invasive, ionizing, time-consuming and costly. Thus, there is an undeniable urgent need for a non-invasive and reliable imaging tool, which could play a crucial role in the early diagnosis of conduction diseases and allow physicians to choose the best course of action.
The 12-lead electrocardiogram (ECG) is the current non-invasive clinical tool routinely used to diagnose and localize cardiac arrhythmias prior to intracardiac catheter ablation. However, it has limited accuracy and can be subject to operator bias. Besides, QRS complex narrowing on the clinical ECG after pacing device implantation is also used for response assessment in patients undergoing Cardiac Resynchronization Therapy (CRT). The latter is an established treatment for systolic heart failure patients who have Left Bundle Branch Block as well as a reduced ejection fraction and prolonged QRS duration. Yet, it is still not well understood why 30 to 40 % of CRT recipients do not respond.
Echocardiography, due to its portability and ease-of-use, is the most frequently used imaging modality in clinical cardiology. In this dissertation, we assess the clinical performance of Electromechanical Wave Imaging (EWI) as a high frame rate ultrasound-based functional modality that can non-invasively map the electromechanical activation of the heart, i.e., the transient deformations immediately following the electrical activation. The objective of this dissertation is to demonstrate the potential clinical value of EWI for both arrhythmia detection and CRT characterization applications.
The first step in translating EWI to the clinic was ensuring that the technique could reli- ably and reproducibly measure the electromechanical activation sequence independently of the probe angle and imaging view in healthy human volunteers (n=7). This dissertation then demonstrated the accuracy of EWI for localizing a variety of ventricular and atrial arrhythmias (accessory pathways in Wolff-Parkinson-White (WPW) syndrome, premature ventricular contractions, focal atrial tachycardia and macro-reentrant atrial flutter) in pediatric (n=14) and adult (n=55) patients prior to catheter ablation more accurately than 12-lead ECG predictions, as validated against electroanatomical mapping.
Additionally, 3D-rendered EWI isochrones were illustrated to be capable of significantly distinguishing different biventricular pacing conditions (pā¤0.05) with the RWAT and LWAT metrics, assessing the ventricular dyssynchrony change in heart failure patients (n=16) undergoing CRT, and visualizing it in 3D. EWI also provided quantification of %šššš in CRT patients (n=38): the amount of left-ventricular resynchronized myocardium, which was found to be a reliable response predictor at 3-, 6-, or 9-month clinical follow-up through its post-CRT values by significantly identifying super-responders from non-responders within 24 hours of implantation (pā¤0.05). Furthermore, 3D-rendered isochrones successfully characterized the ventricular activation resulting from His Bundle pacing for the first time (n=4), which was undistinguishable from true physiological activation in sinus rhythm healthy volunteers with the EWI-based activation time distribution dispersion metric. The dispersion was, however, reported to significantly discriminate novel His pacing from other more conventional biventricular pacing schemes (pā¤0.01).
Finally, we developed and optimized a fully automated zero-crossing algorithm towards a faster, more robust and less observer dependent EWI isochrone generation process. The support vector machine (SVM) and Random Forest machine learning models were both shown capable of successfully identifying the accessory pathway in WPW patients and the pacing electrode location in paced canines. Nevertheless, the best performing algorithm was hereby proven to be the Random Forest classifier with n=200 trees with a precision rising to 97%, and a predictivity that was not impacted by the type of testing dataset it was applied to (human or canine).
Overall, in this dissertation, we established the clinical potential of EWI as a viable assisting visual feedback tool, that could not only be used for diagnosis and treatment planning prior to surgical procedures, but also for monitoring during, and assessing long-term resolution of arrhythmia after catheter ablation or heart failure after a CRT implant.
| Identifer | oai:union.ndltd.org:columbia.edu/oai:academiccommons.columbia.edu:10.7916/d8-nxy6-ks03 |
| Date | January 2020 |
| Creators | Melki, Lea |
| Source Sets | Columbia University |
| Language | English |
| Detected Language | English |
| Type | Theses |
Page generated in 0.0025 seconds