|
|
 |
 |
 |
This work deals with the creation of a multi media application for the medical education of the electrophysiology of the human heart. The visualized electrical fields are based on simulations of the propagation of electrical excitation within the human heart. The distributions of transmembrane potential, source current density, and electrical potential were calculated and visualized for sinus rhythm as well as for different pathologies. The visualizations consist of animations and 3D models providing information about the spatiotemporal field distribution.
In this work, the Visible Man dataset ([3]) of the National Library of Medicine, Bethesda, Maryland (USA) is used to derive a detailed anatomical model of the human body. Therefore different techniques of digital image processing ([4]) are applied to the cryosections (see Figure 1) and CT scans of the Visible Man dataset. Initially, these images are preprocessed correcting alignment deficiencies by means of 2D correlation techniques ([5]). Unusable or missing images of the dataset are interpolated applying an image morphing method based on radial basis transformations ([5]) (see Figure 2). In a final preprocessing step the images are adjoined forming a multi-modal 3D dataset. To distinguish the different types of tissue in this 3D dataset different methods of 3D digital image processing are applied to segment and classify the dataset. Besides different filters, using 3D region growing (see Figure 3) supported by interactively deformable surface meshes allowed the distinction of about 40 types of tissue. In Figure 4, two three dimensional views of the classified datasets are displayed.
In order to include the anisotropic properties of the skeletal and cardiac muscles, the orientation of the muscle bundles are determined by means of a texture analysis ([6]). Therefore, for a set of user defined locations in the heart, the principle components are calculated and interactively corrected by means of a 3D editor. Thus, the muscle fiber orientation of each volume element of this set is represented by the solid angle of the first principle component. To obtain the fiber orientation of the remaining volume elements, an iterative interpolation algorithm is applied to the cardiac and skeletal muscle domains. The cardiac fiber orientation is displayed in Figure 5.
The temporal and spatial propagation of the electrical excitation is strongly affected by the specialized cardiac conduction system. It consists of specialized muscle fibers building an arboreal structure connecting the atrio-ventricular node with the Purkinje fiber endings through the bundle of His and the left and right bundle branches. Due to the small size of these specialized fibers and due to the similarity of its colors to the colors of the adjacent tissue it is hardly possible to segment the conduction system based on the Visible Man data. Therefore, in this work, it is built semi-automatically. It is represented by a tree connecting nodes by edges. For the majority of the nodes the location is determined automatically. Only few nodes are determined by user interaction, e.g. the atrio-ventricular node and few nodes of the left and right bundle branches. The connection between the nodes is calculated automatically applying a variant of Prim's algorithm ([7]) to find the shortest subspanning tree within all nodes and edges. The specialized cardiac conduction system is shown in Figure 6.
In a further step the source current density is calculated outgoing from the simulated
transmembrane potential distribution
using a bidomain model ([11]) which averages the extra cellular and
intracellular space into
two domains of the same location separated at each point by the cell membrane.
The electrical sources 
are calculated from the
transmembrane potential 
with regard to the effective intracellular conductivity tensor
as follows ([11]):
 , |
is the impressed current density.
Finally, this source current distribution is used to calculate the electrical potential distribution in the entire body. Therefore the generalized Poisson equation
 , |
is the extracellular potential and
is the conductivity tensor,
is solved by means of a finite difference multi-grid solver
regarding the heterogeneous anisotropic tissue distribution ([12]).
In addition, the standard electrocardiograms (ECG) are calculated by
evaluating the potential distribution on the body surface.
Changing the anatomical or electrophysiologic properties of the tissue facilitates the simulation of pathologies. E.g. by cutting the specialized cardiac conduction system at the right bundle branch it is possible to simulate a right bundle branch block. By changing the shape of the transmembrane potential course and therefore the excitation velocity and refractory periods it is possible to simulate flutter and fibrilation as well as an infarction. Adding pathways from the atrial to the ventricular myocardium enables the simulation of the Wolff-Parkinson-White syndrome.
Some exemplary animations are listed here: The first animation shows the electrical excitation propagation of a sinus rhythm. The transmembrane potential distribution is shown in Movie 1. In Movie 2 the source current distribution of an transmural infarction in the left ventricular free wall is visualized with a semi-transparent heart surface. Movie 3 shows the body surface potential map of a right bundle branch block. Figure 8 shows an electrocardiogram (ECG) of the right bundle branch block in comparison to the sinus rhythm ECG.
Future work will focus on cardiac motion and the simulation of the elasto-mechanical contraction of the heart. The simulation of surgical interventions may be included in future work as well.
|
|
|