Maria Inês Francisco Gândara.
Inês joined our lab as an international graduate student from NOVA school of science & technology, Lisbon, Portugal to pursue her M.S. thesis. I had the good fortune of being her thesis advisor. I am excited and happy to report that Inês successfully defended her dissertation and is now a proud graduate of NOVA University.
Effects of spatial resolution on arrhythmia driver detection and localization
Arrhythmia is a cardiac rhythm disorder that can be fatal. Its treatment includes ab- lation of the cardiac tissue and/or defibrillation. Advances are being made for both treatment options to localize the culprit region and apply therapy directly where it is needed. However, success rates have been inconsistent, with frequent arrhythmia recurrence. A likely reason is the limited current resolution of mapping devices, that averages 4 mm. Higher resolution may improve localization of arrhythmia drivers, termed rotors, and consequently improve efficacy of treatment.
This study evaluates the effects of spa- tial resolution on arrhythmia dynamics, rotor tracking, and rotor localization. Optical data from ex vivo human hearts was used, being clinically relevant and with ultra-high spatial resolution. To simulate different resolutions, original data was downsampled by multiple factors and upsampled back to full resolution. Rotors were tracked for each sub-resolution and compared to the rotors in the original data. Further comparisons were made according to arrhythmia type, sex, anatomical region, and mapped surface. Accuracy profiles were created for both rotor detection and localization, describing how accuracy changed with spatial resolution and spatial accuracy.
Rotor detection accuracy for currently used mapping devices was found to be 57±4%. Localization accuracy is 61±7%. Detection accuracy was above 80% only for a resolution of 1.4 mm. Moreover, the detection and localization accuracies were affected by arrhythmia type, and rotor incidence was found to be higher in the endocardium. Therefore, current clinical rotor detection and localization accuracies can be expected to fall within a confidence interl,gttval of 47-67% and 46-75%, respectively. This means that a higher spatial resolution is needed in cardiac mapping devices than what is currently available.
For high accuracy, a resolution of at least 1.4 mm is required. The accuracy profiles provided in this thesis may serve as a guideline for future mapping device development.