In this article we introduce a novel variational model for two-phase segmentation tasks, which is related to the popular ChanVese(CV) segmentation model[4]. In particular, the proposed model is based on a discriminant analysis of the given data. Level set methods are used to realize this approach is numerically. We start by applying CV model on medical ultrasound data perturbed by multiplicative speckle noise. The respective observations are noted i.e. Section1. Subsequently, discriminant analysis based segmentation model is applied i.e. in Section2.Finally, we validate the methods on real patient data from echocardiographic examinations in Section3.