Segmentation of the left ventricle in myocardial perfusion SPECT using variational level set formulation

Myocardial scintigraphy is an imaging modality which provides functional information. We used myocardium SPECT images performed at stress and rest with Tc-99m MIBI. The myocardial perfusion is estimated by comparing images at two different instant rest and stress and it is indicated when ischemia is from the normal case suspected. The diagnosis is obtained by comparing the topology of myocardium blood flow at these different instants. Shape instabilities have been an important problem, particularly in 3D. Since then, different approaches to the solution of this problem have been developed, we proposed an variational level set based on Mumford-Shah segmentation techniques and the variational level set method. Its strategy is first searching the object in the dataset. For tracing its exact contour, we can detect objects whose boundaries are not necessarily defined by gradient or with very smooth boundaries, for which the classical active contour models are not applicable. This segmentation method was fast and robust on noisy, low-resolution data. Rendered surface of the endocardium and epicardium wall at stress and rest the contours of heart are very well detected and preserved. The method could be used to extract quantitative functional parameters.