Remodeling of the Left Heart Ventricle

A art attack often leaves parts of the muscular tissue of the ventricular wall afunctional. This tissue is called scar tissue or avital heart muscle tissue. To compensate for this a functional scar tissue, the ventricle enlarges up to an aneurysmatic enlargement, thus creating a hemodynamic adverse left ventricle geometry. In order to improve the hemodynamics of the left ventricle (and the mitral valve), ventricular reconstructions are performed to improve the ejection fraction. This ventricular reconstruction – or remodeling – aims at an ellipsoid shape, which is the optimal geometry for the ventricle. Currently, standardized elliptical balloons are positioned in the ventricle to provide a template for the remodeled ventricle. However, the reconstructed ventricles often still have an adverse hemodynamic and do not achieve the optimal medical shape. To allow a better reconstruction quality, patient-individual ventricle templates may be used, which in turn require more sophisticated planning to establish a better template shape and size. This project focuses on the planning for these patient-individual ventricle templates.

Based on Cardio-CT datasets of 20 patients, we segmented the contrast enhanced enddiastolic blood pool of the left ventricles using threshold-based 3D region growing. These segmentations result in voxel-accurate representations of the left ventricles, which furthermore allows a precise volume measurement of the enlarged ventricles. Late enhancement MRI data is used to differentiate the avital scar tissue from the vital muscle tissue to provide information which parts of the ventricular walls should be remodeled. Afterwards, the template is modeled based on the segmentation and the scar tissue localization (Figure 1), taking into account the optimal size (152 ± 33ml) and geometry (ellipsoid). For consistent modeling, the previous (already processed) slice of the template is projected on top of the current slice.

Overall, the templates of 20 patient datasets were modeled with an average volume of 144 ml, while the original volume of the enlarged ventricles was on average 343 ml. All templates maintained an ellipsoid shape and were thus well suited as a ventricular shape template.
Future work will focus on addressing technical shortcomings and particularly on the building of patient-specific ventricle templates, which are then assessed in surgical use.