Cell segmentation using a charged parametric active contour

In image processing, deformable models known as active contours have been used extensively to locate object boundaries in images. In this thesis, we combine concepts from two active contour methods, energy-minimizing snake active contours and charged active contours based on electrostatics to segment cells in images of human epithelial cell cultures. The first method uses energy minimization and forces computed from the gradient of an image to deform a contour so that it conforms to the boundary of an object the contour is placed in proximity to. The second method accomplishes the same task by using forces based on charged particle dynamics. We will show that by using the energy minimization model from the snake active contour and forces based on charged particle dynamics that we are able to segment the nucleus and cell wall of human epithelial cells.