Abstract:The intima and adventitia edge detection of vascular wall is a key step in Intravascular Ultrasound(IVUS) processing. An accurate selection of initial contour would be significant for edge extraction based on active contour model. This paper puts forward a new method for extracting the edge of vascular wall. At first, adaptive threshold segmentation is used to reduce the complexity of the IVUS original images,then find the initial contour line of the vascular wall with Marr edge detection. Finally, an improved GVF-snake model is used to make the initial contour line converge to the true edge of vascular wall accurately. The experimental results confirm the validity of the new method.