Cardia3D (CARdiac Image Analysis) is an automated method for heart MR images analysis.

Modern MR scanners generate huge amounts of data over complete heart cycles. The manual analysis of this data can be tedious. The goal of Cardia3D is to ease this task by using a biomechanical model of the heart to automatically retrieve the heart's contours in the image. Once the contour extraction process is done, functional parameters ranging from ejection fraction to motion parameters can be extracted. Cardia3D has been applied to both human and small animal MR images.
Cardia3D's biomechanical model is called the Deformable Elastic Template. It is composed of a n ap priori template coupled with a non linear hyperelastic deformation model. The problem of adapting the model to a specific image is treated as an energy minimization problem, between external (image) and internal (elasticity) energies. The internal energy is used as a balance between image information and the model's a priori shape information. The resulting physics problem is solved using the Finite Element Method. Unfortunately, this results in heavy computational needs.
The grid can help Cardia3D in two ways:

• In interactive mode, the clinician is asking for the extracted data. The computations must be performed really fast. The computational problem can be broken up in smaller parts and distributed on several nodes for faster results.
• In batch mode, the grid can process a great amount of examinations. The computation results can then be associated with the images in a medical database, for future research.

Video

References

J Schaerer, P Clarysse, B Hiba, P Croisille, IE Magnin, Simultaneous Segmentation of the Left and Right Heart Ventricles in 3D Cine MR Images of Small Animals, Computers In Cardiology, Lyon, 2005.