AbstractID: 9101 Title: Automatic Segmentation of Tumor and Organ Contours for Adaptive Treatment using the Active Contours (Snakes) Technique Tumor volume and critical structure image segmentation is an essential part of an online adaptive radiation therapy process. Real-time treatment planning and delivery (image guided radiotherapy) requires both fast and accurate online image analysis (automatic segmentation). The method proposed in this study is to position contours carefully drawn offline using a planning study within the current online image set, then deform each contour using the active contours technique (snakes) so that conformity improves. Given an online image set, an edge matrix is derived using a combination of Gaussian function and gradient operator. Predefined contours are moved around the edge matrix. A goodness of match parameter is calculated that sums the squares from all the points of the contour(s). The value of this parameter determines the final position of the contour(s). The contour(s) are then deformed to conform to the current image by applying the gradient vector flow (GVF) force fields of the snake technique. The GVF field is defined as the vector field that minimizes the energy function that consists of partial derivatives of the image matrix and other variables. This approach is applied to sets of images of the prostate where it is especially suitable due to the limitation of current imaging techniques with soft tissue structure discernibility. Results show good conformity of the contour(s) to the image with little human intervention. It is feasible to apply these techniques for automatic segmentation of online images obtained through in-room CT or volume image devices for adaptive radiation therapy.