Progress Towards Model - Based Estimation of the Cardia

Ele trome hani al A tivity from ECG Signals and 4D Images M. Sermesant1, Y. Coudière2, H. Delingette1, N. Aya he1, J. Sainte-Marie3, D. Chapelle3, F. Clément4, M. Sorine4 1: Epidaure Resear h Proje t, INRIA Sophia-Antipolis, BP 93, 06902 Sophia-Antipolis, Fran e 2: Mathemati s Laboratory, Nantes University, Fran e 3: Ma s Resear h Proje t, INRIA Ro quen ourt, Fran e 4: Sosso Resear h Proje t, INRIA Ro quen ourt, Fran e 7th May 2002 Abstra t We present re ent advan es on a 3D numeri al modeling of the myo ardium whi h ouples ele tri al and biome hani al models. The long-term obje tive is to simulate a realisti ontra tion of the heart from both ele tri al measurements (typi ally the ECG) and geometri al measurements (typi ally provided by medi al imaging). This realisti ontra tion should provide useful quantitative parameters for the diagnosis and also for guiding some new forms of therapy. Our modeling is based on a multi-s ale analysis ranging from mi ros opi to ma ros opi s ales, and integrates a priori information on the overall geometry and on the ber dire tions extra ted from spe i medi al imaging te hniques (e.g. MRI). The FitzHugh-Nagumo equations are solved along with a onstitutive law based on the Hill-Maxwell Rheologi al model, on whi h a data assimilation analysis is done. In medi al image analysis, we believe that this new generation of physi s-based deformable models will be useful to provide a more robust quantitative interpretation of temporal series of ardia images. 1 Introdu tion The knowledge of the heart fun tion has greatly improved on the nanos opi , mi ros opi and mesos opi s ales during the last de ades, thus a global integrative work of this organ be omes on eivable [1℄. It is the obje tive of our multidis iplinary proje t ICEMA1 to build a generi dynami model of the beating heart and a pro edure to automati ally adjust the parameters to any spe i patient from relatively easy-to-a ess measurements: ECGs (ele tro ardiograms) and time sequen es of volumetri medi al images [2℄. On e the generi model is adapted to a spe i patient, it be omes possible to derive a set of quantitative and obje tive parameters useful for lini ians and physiologists.