Dynamical segmentation of the left ventricle in echocardiographic image sequences

Presents a new methodology for the analysis of echocardiographic image sequences. The echocardiographic medical examination is a frequent practice in the cardiology clinic, and generally there is bi-dimensional echocardiographic equipment. We have developed a. new processing chain that will allow a medical diagnosis issued from a 3D reconstruction of the ultrasound images. The project is divided into several 3D modules: acquisition, filtering, segmentation, reconstruction, and visualization. Inside the processing chain, the segmentation module of ultrasound images has always been a problem. In this paper, we propose two different methods for the 3D segmentation: a model of 3D snakes and a front propagation model that allow an accurate segmentation and reconstruction of the internal wall of the ventricle. From these segmentation methods, some important medical parameters are computed: ejection fraction, cardiac index, ventricle volume, etc.