A quantum mechanics-based algorithm for vessel segmentation in retinal images

摘要: Blood vessel segmentation is an important step in retinal image analysis. It one of the steps required for computer-aided detection ophthalmic diseases. In this paper, a novel quantum mechanics-based algorithm presented. The consists three major steps. first preprocessing images to prepare further processing. second feature extraction where set four features generated at each pixel. These are then combined using nonlinear transformation dimensionality reduction. final applying recently proposed framework step, pixels mapped systems that allowed evolve from initial state governed by Schrodinger's equation. evolution controlled Hamiltonian operator which function extracted A measurement consequently performed determine whether pixel belongs or non-vessel classes. Many functional forms proposed, and best performing form was selected. tested on publicly available DRIVE database. average results sensitivity, specificity, accuracy 80.29, 97.34, 95.83 %, respectively. compared some published techniques showing superior performance method. Finally, implementation computer challenges facing introduced.