Feasibility of Atrial Activation Time Imaging F Hanser1 2 B Tilg1 R Modre1 G Fischer1 B Messnarz1 2 F Hintringer3 FX Roithinger3 Institute for Medical Signal Processing and Imaging University for Health Informatics and Technology Tyrol Innsbruck Austria 2 Institute of Biomedical Engineering Graz University of Technology Austria 3 Clinical Department of Cardiology University Hospital Innsbruck Austria 1 Abstract Atrial activation time imaging constitutes a more sophisticated problem with respect to ventricular activation time imaging The complex geometry of the atria orifices of the pulmonary veins orifices of superior and inferior vena cava tricuspid and mitral annuli and right and left appendages makes it more difficult to generate a geometrical model that qualifies for a boundary element formulation The poor contrast in the MR images makes the segmentation process quite delicate In addition only the endocardial boundaries can be seen in the atrial MR images Epicardial boundaries have therefore to be constructed artificially A signal to noise ratio of about 20dB more than 40dB in ventricular activation time imaging and the significantly smaller effective rank of the ECG data matrix impose additional challenges on the stability of the inverse algorithm In this work we investigate the feasibility of atrial activation time imaging based on data sets of four patients who underwent an electrophysiologic study The feasibility of atrial activation time imaging is investigated based on data sets of four patients who underwent an electrophysiologic study Several pacing protocols with pacing sites at the right atrial appendage coronary sinus and high right atrium were part of the study and were employed to reconstruct the associated atrial activation time patterns The localization error was estimated to be between 8 and 14 mm 1 Introduction Atrial and ventricular surface activation time imaging from body surface ECG mapping data is supposed to become a diagnostically powerful clinical tool for assessing cardiac arrhythmias 1 This cardiac source imaging technique aims at providing noninvasively information about electrical excitation in order to assist the cardiologist in developing strategies for the treatment of cardiac arrhythmias Common cardiac arrhythmias such as the Wolf Parkinson White syndrome atrioventricular nodal reentrant tachycardia or atrial fibrillation can at least in many cases be traced back to accessory pathways atrial or ventricular foci e g from the pulmonary veins 2 3 and reentrant circuits Identifying the site of origin of the ectopic focus or the location of an accessory pathway provides the essential information for treatment strategies such as catheter ablation 4 Activation time imaging from 3D time anatomical and body surface ECG mapping data enables noninvasively the imaging of the electrical function in the heart 5 The method yields solutions to the electrocardiographic inverse problem and is based on an electrodynamic model of the patient s volume conductor and heart The model of the heart includes a model of both the atrium and ventricle A separate model for the atrium and ventricle has been inevitable because whole heart models still resist a computational and technical implementation for providing solutions to the electrocardiographic inverse problem 0276 6547 02 17 00 2002 IEEE 2 Data acquisition and modeling Body surface ECG data were acquired under clinical conditions with a 62 lead ECG mapping system The Mark 8 body surface potential mapping system Biosemi V O F The Netherlands is an on line portable computer acquisition system with data transmission via optical fiber A radiotransparent carbon electrode array was utilized to record unipolar ECG signals from 62 torso sites anterior 41 posterior 21 while simultaneously enabling X ray examination The Wilson Central Terminal defined as usual the reference potential 6 Electrode signals were amplified and high pass and low pass filtered at edge frequencies of 0 3Hz and 400Hz with a first and fourth order analog Bessel filter respectively Analog to digital conversion was realized by a 16 bit AD converter at a sampling rate of 2048Hz per channel and a quantization resolution of 500nV bit No additional digital filtering was applied The torso was imaged with a 1 5Tesla MR scanner 601 Computers in Cardiology 2002 29 601 604 Siemens Vision Plus The myocardium was additionally imaged in an ECG gated breath hold oblique imaging mode in order to model the heart s surface Vitamin E markers were utilized to determine 7 reference positions from the axial MR scans to be able to couple MRI with ECG data The 62 electrode and 7 reference positions were acquired with the FASTRAK system Polhemus Inc USA The entire volume conductors including the blood masses were modeled with about 4100 triangles 7 The atrial surfaces were represented by about 1050 triangular elements The different individual compartments comprised the torso the lungs and the blood mass The associated conductivities were assumed to be 0 2 0 08 and 0 6Sm 1 respectively Figure 1 shows a geometrical representation of a patient s volume conductor together with electrode positions of the 62 lead ECG mapping system Mathematically speaking the relation between activation time on the atrial surface and the body surface ECG is represented by a nonlinear inverse ill posed problem It can be formulated in the following form F D 1 where F is a nonlinear operator which maps the activation time onto the body surface ECG data D Assuming k k 0 to be a series of approximations of the true solution and linearizing Eq 1 in the point k yields F k Fk k D 2 where Fk is an abbreviation for F k and F represents the Frechet derivative of the operator F Equation 1 can be written in a technically more useful form Fk Dk 3 with Dk D Fk k F k Equation 3 is in general again ill posed In order to find a regularized approximation for a regularization method for linear illposed problems can be employed Applying second order Tikhonov regularization with the Laplacian operator and the regularization parameter k yields the following regularized approximation 1 Fk Fk 2k Fk Dk 4 where the asterisk marks the adjoint operator Repeating this process yields an iteration process k 1 k k with the incremental activation time 1 k Fk Fk 2k Fk D F k 2k k 6 As the index k increases the iteration process in Eq 5 converges to a regularized approximation of the activation time pattern A more detailed
View Full Document
Unlocking...