Segmentation of Ultrasonic Images by Application of Active Contour Models

Segmentation of ultrasonic images (US) enables the measurement of the area or in §D US the volume of objects. Due to speckle noise which is immanent to this modern, non-invasive, low-cost acquisition technique, segmentation is a skillful task. Here, we apply an active contour model (ACM). By using a global energy minimization approach in combination with a region oriented energy term our scheme offers great robustness and enables the interpolation of contours at image regions of low contrast. The large number of parameters of the ACM are adapted by evaluation of a training region sketched by the user.