Abstract: The extraction of the vascular envelope from intravascular ultrasound(IVUS) images is helpful to the diagnosis of the coronary diseases.In this paper,a novel automatic envelope extraction method is proposed using multiple categories of information including gray scale parameters,temporal features of the sequential images and the prior knowledge.The temporal features and prior knowledge are firstly used to reduce the noise and the artifacts,and extract an initial envelope of the first frame.To obtain its final envelope,a Bspline GVF snake combining with the gradient,variance and mean information of the gray image is then adopted to deform the initial envelope.Finally,temporal features of sequential images are utilized to extract the envelopes of the following frames.It is shown that this approach using multiple categories of information is effective for vascular envelope extraction and superior to traditional methods.