Automated detection of optic disk boundary by a new deformable model technique

A new algorithm to detect the exact optic disk boundary, based on deformable model, is proposed in this paper. The energy function for deformation is defined as a combination of local energies. The proposed method modified the original snake technique in two aspects: clustering the contour points into edge-point group or uncertain-point group after each deformation, and updating the contour by variable updating-sample numbers. The modifications not only directly solve the blood vessel problem, but also make this algorithm more accurate and robust to noises, weak edges and fuzzy contour shapes. Based on 100 randomly-selected testing images, the success rates are 94% for the proposed method, as compared to 12% for the GVF-snake and 82% for the modified ASM algorithm, which show the effectiveness of the proposed method.