Entity

Time filter

Source Type

Artigues-près-Bordeaux, France

Pop M.,University of Toronto | Sermesant M.,French Institute for Research in Computer Science and Automation | Flor R.,Sunnybrook Research Institute | Pierre C.,University of Pau and Pays de lAdour | And 8 more authors.
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) | Year: 2013

Sudden cardiac death is a major cause of death in industrialized world; in particular, patients with prior infarction can develop lethal arrhythmia. Our aim is to understand the transmural propagation of electrical wave and to accurately predict activation times under different stimulation conditions (sinus rhythm and paced) using MRI-based computer models of normal or structurally diseased hearts. Parameterization of such models is a prerequisite step prior integration into clinical platforms. In this work, we first evaluated the errors associated with the registration process between contact EP data and MRI-based models, using in vivo CARTO maps recorded in three swine hearts (two healthy and one infarcted) and the corresponding heart meshes obtained from high-resolution ex vivo diffusion weighted DW-MRI (voxel size < 1mm 3). We used the open-source software Vurtigo to align, register and project the CARTO depolarization maps (from LV-endocardium and epicardium) onto the MR-derived meshes, with an acceptable registration error of < 5mm in all maps. We then compared simulation results obtained with the macroscopic monodomain formalism (i.e., the two-variable Aliev-Panfilov model), the simple Eikonal model, and the complex bidomain model (TNNP model) under different stimulation conditions. We found small errors between the measured and the predicted activation times, as well as between the depolarization times using these three models (e.g., with a mean error of 3.4 ms between the A-P and TNNP model), suggesting that simple mathematical formalisms might be a good choice for integration of fast, predictive models into clinical platforms. © 2013 Springer-Verlag.


Pop M.,Sunnybrook Research Institute | Sermesant M.,French Institute for Research in Computer Science and Automation | Mansi T.,Siemens AG | Crystal E.,Sunnybrook Research Institute | And 10 more authors.
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) | Year: 2012

Our broad aim is to integrate experimental measurements (electro-cardiographic and MR) and cardiac computer models, for a better understanding of transmural wave propagation in individual hearts. In this paper, we first describe the acquisition and processing of the data provided to the EP simulation challenge organized at STACOM'11. The measurements were obtained in two swine hearts (i.e., one healthy and one with chronic infarction) and comprise in-vivo electro-anatomical CARTO maps (e.g., surfacic endo-/epicardial depolarization maps and bipolar voltage maps recorded in sinus rhythm), and high-resolution ex-vivo diffusion-weighted DW-MR images (voxel size < 1mm 3). We briefly detail how we built anisotropic 3D MRI-based models for these two hearts, with fiber directions obtained using DW-MRI methods (which also allowed for infarct identification). We then focus on applications in cardiac modelling concerning propagation of depolarization wave, by employing forward mathematical approaches. Specifically, we present simulation results for the depolarization wave using a fast, macroscopic monodomain formalism (i.e., the two-variable Aliev-Panfilov model) and comparisons with measured depolarization times. We also include simulations obtained using the healthy heart and a simple Eikonal model, as well as a complex bidomain model. The results demonstrate small differences between computed isochrones using these computer models; specifically, we calculated a mean error ± S.D. of 2.8 ± 1.67 ms between Aliev-Panfilov and Eikonal models, and 6.1 ± 3.9 ms between Alie-Panfilov and bidomain models, respectively. © 2012 Springer-Verlag.

Discover hidden collaborations