Date of Award

3-27-2018

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Bioinformatics

First Advisor

Hari Eswaran

Abstract

In-utero monitoring and assessment of fetal heart are crucial to early diagnose and treat the electro-physiological problems. Fetal magnetocardiography (fMCG) has become an attractive technique for in-utero assessment of cardiac electrophysiology. The measured fMCG signals are affected by the characteristics of the fetal heart and volume conductor which is formed by fetal and maternal tissues, and the position of the fetus with respect to sensors. Therefore, the signal morphology of fMCG shows differences for each measurement which prevents their comparison across subjects. These confounding factors can be removed by applying source localization techniques, which estimate the source parameters of the fetal heart. The main problem of source localization for fetuses is that the characteristics of the volume conductor are completely unknown and the forward model needs to take into account not only the cardiac source parameters but also the volume conductor. Moreover, even with a complete knowledge of the electromagnetic field, the inverse problem cannot be uniquely solved because the same field distribution can be produced by many configurations of the current sources. We have developed a novel hybrid method that integrates the magnetic multipole expansion with the equivalent current dipole (ECD) model. Our method first identifies the best-fitted multipole components and uses their spatial information to define a region of interest (ROI). The hybrid method then can estimate the center of conducting volume using the ROI. Finally, the parameters of the ECD are estimated using the identified center position. Identifying the center of the conducting volume reduces the complexity of solving the inverse problem. The results have shown that the hybrid method is capable of to locate source positions with a localization error as low as 2-4 mm for most of the cases. Also, the estimated ECDs explained the recorded data with a goodness of fit (GOF) value of higher than 97%. The estimation of the source parameters was highly consistent within the same recordings as well as across the different measurements of the same subject. Estimated parameters are used to define metrics to use in the assessment of fetal cardiac activity and development.

Download

Share

COinS