No description available.

published_or_final_version / abstract / Diagnostic Radiology / Master / Master of Philosophy

No description available.

Thesis (M.Sc.)--York University, 2005. Graduate Programme in Computer Science. / Typescript. Includes bibliographical references (leaves 186-196). Also available on the Internet. MODE OF ACCESS via web browser by entering the following URL: http://gateway.proquest.com/openurl?url%5Fver=Z39.88-2004&res%5Fdat=xri:pqdiss &rft_val_fmt=info:ofi/fmt:kev:mtx:dissertation&rft_dat=xri:pqdiss:MR11837

Non-invasive electrical imaging of the heart aims to quantitatively reconstruct information about the electrical activity of the heart from multiple thoracic ECG signals. The computational framework required to produce such electrical images of the heart from non-invasive torso surface signals is presented. It is shown reliable electrical images of the heart can be obtained under a controlled environment. This has been demonstrated using an anatomically realistic boundary element porcine torso model. The procedures required to create a subject specific model using a small number of control points and to create a specific heart model from three-dimensional ultrasound images using a linear fitting procedure are presented. From discrete ECG electrodes a continuous representation of the potential field over the entire torso surface can also be produced using this linear fitting procedure. The construction of the transfer matrices for the two predominant electrocardiographic sources (epicardial potentials and myocardial activation times) are described in detail. The transfer matrices are used to compute activation times within the heart and epicardial potentials on the heart surface. Myocardial activation times are computed using an algorithm based on the Critical Point Theorem while epicardial potentials are computed using standard Tikhonov and Truncated SVD spatially regularised methods as well as Greensite's spatial and temporal regularisation method. The regularisation parameters for the epicardial potentials are determined using a variety of methods (e.g., CRESO criterion, L-curve, zero-crossing). The potential and activation based formulations are compared in a comprehensive inverse simulation study. To try and capture the dynamic and variable nature of cardiac electrical activity, the study is performed with three different types of cardiac sources with a realistic porcine model. These simulations investigate the effect on the computed solutions of individual and combinations of modelling errors. These errors include corruption in the torso surface signals, changes in material properties and geometric distortion. In general, the activation based formulation is preferred over the epicardial potential formulations, with Greensite's method found to be the best method for reconstructing epicardial potentials. Under optimal conditions, the activation approach could reconstruct the activation times to within RMS. Both potential and activation based formulations were found to be relatively insensitive to changes in material properties such as lung conductivities and activation function shapes. When examining individual errors, the geometry and positions of the torso and heart had the greatest effects on the inverse solutions. The relative heart position needed to be determined to within to obtain results within of the solutions obtained under control conditions. When the modelling errors are combined to produce errors which can be expected in a clinical or experimental situation the activation based solutions were consistently more accurate than potential based solutions. The next necessary step in this project is the detailed validation of the results against in-vivo data. This step is necessary before such algorithms can be reliably used to aid in the assessment of heart function in a clinical environment.

Non-invasive electrical imaging of the heart aims to quantitatively reconstruct information about the electrical activity of the heart from multiple thoracic ECG signals. The computational framework required to produce such electrical images of the heart from non-invasive torso surface signals is presented. It is shown reliable electrical images of the heart can be obtained under a controlled environment. This has been demonstrated using an anatomically realistic boundary element porcine torso model. The procedures required to create a subject specific model using a small number of control points and to create a specific heart model from three-dimensional ultrasound images using a linear fitting procedure are presented. From discrete ECG electrodes a continuous representation of the potential field over the entire torso surface can also be produced using this linear fitting procedure. The construction of the transfer matrices for the two predominant electrocardiographic sources (epicardial potentials and myocardial activation times) are described in detail. The transfer matrices are used to compute activation times within the heart and epicardial potentials on the heart surface. Myocardial activation times are computed using an algorithm based on the Critical Point Theorem while epicardial potentials are computed using standard Tikhonov and Truncated SVD spatially regularised methods as well as Greensite's spatial and temporal regularisation method. The regularisation parameters for the epicardial potentials are determined using a variety of methods (e.g., CRESO criterion, L-curve, zero-crossing). The potential and activation based formulations are compared in a comprehensive inverse simulation study. To try and capture the dynamic and variable nature of cardiac electrical activity, the study is performed with three different types of cardiac sources with a realistic porcine model. These simulations investigate the effect on the computed solutions of individual and combinations of modelling errors. These errors include corruption in the torso surface signals, changes in material properties and geometric distortion. In general, the activation based formulation is preferred over the epicardial potential formulations, with Greensite's method found to be the best method for reconstructing epicardial potentials. Under optimal conditions, the activation approach could reconstruct the activation times to within RMS. Both potential and activation based formulations were found to be relatively insensitive to changes in material properties such as lung conductivities and activation function shapes. When examining individual errors, the geometry and positions of the torso and heart had the greatest effects on the inverse solutions. The relative heart position needed to be determined to within to obtain results within of the solutions obtained under control conditions. When the modelling errors are combined to produce errors which can be expected in a clinical or experimental situation the activation based solutions were consistently more accurate than potential based solutions. The next necessary step in this project is the detailed validation of the results against in-vivo data. This step is necessary before such algorithms can be reliably used to aid in the assessment of heart function in a clinical environment.

Non-invasive electrical imaging of the heart aims to quantitatively reconstruct information about the electrical activity of the heart from multiple thoracic ECG signals. The computational framework required to produce such electrical images of the heart from non-invasive torso surface signals is presented. It is shown reliable electrical images of the heart can be obtained under a controlled environment. This has been demonstrated using an anatomically realistic boundary element porcine torso model. The procedures required to create a subject specific model using a small number of control points and to create a specific heart model from three-dimensional ultrasound images using a linear fitting procedure are presented. From discrete ECG electrodes a continuous representation of the potential field over the entire torso surface can also be produced using this linear fitting procedure. The construction of the transfer matrices for the two predominant electrocardiographic sources (epicardial potentials and myocardial activation times) are described in detail. The transfer matrices are used to compute activation times within the heart and epicardial potentials on the heart surface. Myocardial activation times are computed using an algorithm based on the Critical Point Theorem while epicardial potentials are computed using standard Tikhonov and Truncated SVD spatially regularised methods as well as Greensite's spatial and temporal regularisation method. The regularisation parameters for the epicardial potentials are determined using a variety of methods (e.g., CRESO criterion, L-curve, zero-crossing). The potential and activation based formulations are compared in a comprehensive inverse simulation study. To try and capture the dynamic and variable nature of cardiac electrical activity, the study is performed with three different types of cardiac sources with a realistic porcine model. These simulations investigate the effect on the computed solutions of individual and combinations of modelling errors. These errors include corruption in the torso surface signals, changes in material properties and geometric distortion. In general, the activation based formulation is preferred over the epicardial potential formulations, with Greensite's method found to be the best method for reconstructing epicardial potentials. Under optimal conditions, the activation approach could reconstruct the activation times to within RMS. Both potential and activation based formulations were found to be relatively insensitive to changes in material properties such as lung conductivities and activation function shapes. When examining individual errors, the geometry and positions of the torso and heart had the greatest effects on the inverse solutions. The relative heart position needed to be determined to within to obtain results within of the solutions obtained under control conditions. When the modelling errors are combined to produce errors which can be expected in a clinical or experimental situation the activation based solutions were consistently more accurate than potential based solutions. The next necessary step in this project is the detailed validation of the results against in-vivo data. This step is necessary before such algorithms can be reliably used to aid in the assessment of heart function in a clinical environment.

Non-invasive electrical imaging of the heart aims to quantitatively reconstruct information about the electrical activity of the heart from multiple thoracic ECG signals. The computational framework required to produce such electrical images of the heart from non-invasive torso surface signals is presented. It is shown reliable electrical images of the heart can be obtained under a controlled environment. This has been demonstrated using an anatomically realistic boundary element porcine torso model. The procedures required to create a subject specific model using a small number of control points and to create a specific heart model from three-dimensional ultrasound images using a linear fitting procedure are presented. From discrete ECG electrodes a continuous representation of the potential field over the entire torso surface can also be produced using this linear fitting procedure. The construction of the transfer matrices for the two predominant electrocardiographic sources (epicardial potentials and myocardial activation times) are described in detail. The transfer matrices are used to compute activation times within the heart and epicardial potentials on the heart surface. Myocardial activation times are computed using an algorithm based on the Critical Point Theorem while epicardial potentials are computed using standard Tikhonov and Truncated SVD spatially regularised methods as well as Greensite's spatial and temporal regularisation method. The regularisation parameters for the epicardial potentials are determined using a variety of methods (e.g., CRESO criterion, L-curve, zero-crossing). The potential and activation based formulations are compared in a comprehensive inverse simulation study. To try and capture the dynamic and variable nature of cardiac electrical activity, the study is performed with three different types of cardiac sources with a realistic porcine model. These simulations investigate the effect on the computed solutions of individual and combinations of modelling errors. These errors include corruption in the torso surface signals, changes in material properties and geometric distortion. In general, the activation based formulation is preferred over the epicardial potential formulations, with Greensite's method found to be the best method for reconstructing epicardial potentials. Under optimal conditions, the activation approach could reconstruct the activation times to within RMS. Both potential and activation based formulations were found to be relatively insensitive to changes in material properties such as lung conductivities and activation function shapes. When examining individual errors, the geometry and positions of the torso and heart had the greatest effects on the inverse solutions. The relative heart position needed to be determined to within to obtain results within of the solutions obtained under control conditions. When the modelling errors are combined to produce errors which can be expected in a clinical or experimental situation the activation based solutions were consistently more accurate than potential based solutions. The next necessary step in this project is the detailed validation of the results against in-vivo data. This step is necessary before such algorithms can be reliably used to aid in the assessment of heart function in a clinical environment.

Diss. (sammanfattning) Linköping : Univ., 2005.

Non-invasive electrical imaging of the heart aims to quantitatively reconstruct information about the electrical activity of the heart from multiple thoracic ECG signals. The computational framework required to produce such electrical images of the heart from non-invasive torso surface signals is presented. It is shown reliable electrical images of the heart can be obtained under a controlled environment. This has been demonstrated using an anatomically realistic boundary element porcine torso model. The procedures required to create a subject specific model using a small number of control points and to create a specific heart model from three-dimensional ultrasound images using a linear fitting procedure are presented. From discrete ECG electrodes a continuous representation of the potential field over the entire torso surface can also be produced using this linear fitting procedure. The construction of the transfer matrices for the two predominant electrocardiographic sources (epicardial potentials and myocardial activation times) are described in detail. The transfer matrices are used to compute activation times within the heart and epicardial potentials on the heart surface. Myocardial activation times are computed using an algorithm based on the Critical Point Theorem while epicardial potentials are computed using standard Tikhonov and Truncated SVD spatially regularised methods as well as Greensite's spatial and temporal regularisation method. The regularisation parameters for the epicardial potentials are determined using a variety of methods (e.g., CRESO criterion, L-curve, zero-crossing). The potential and activation based formulations are compared in a comprehensive inverse simulation study. To try and capture the dynamic and variable nature of cardiac electrical activity, the study is performed with three different types of cardiac sources with a realistic porcine model. These simulations investigate the effect on the computed solutions of individual and combinations of modelling errors. These errors include corruption in the torso surface signals, changes in material properties and geometric distortion. In general, the activation based formulation is preferred over the epicardial potential formulations, with Greensite's method found to be the best method for reconstructing epicardial potentials. Under optimal conditions, the activation approach could reconstruct the activation times to within RMS. Both potential and activation based formulations were found to be relatively insensitive to changes in material properties such as lung conductivities and activation function shapes. When examining individual errors, the geometry and positions of the torso and heart had the greatest effects on the inverse solutions. The relative heart position needed to be determined to within to obtain results within of the solutions obtained under control conditions. When the modelling errors are combined to produce errors which can be expected in a clinical or experimental situation the activation based solutions were consistently more accurate than potential based solutions. The next necessary step in this project is the detailed validation of the results against in-vivo data. This step is necessary before such algorithms can be reliably used to aid in the assessment of heart function in a clinical environment.