Automated registration and fusion of the multi-modality retinal images

Biomedical image fusion is generally scene dependent, which requires intensive computational effort. A novel approach of feature-based registration and area-based heuristic optimization fusion of multi-modality retinal images is proposed in this paper. The new algorithm, which is reliable, robust, and time-efficient, has an automatic adaptation from frame to frame with few tunable threshold parameters. The registration approach is based on the retinal vasculature extraction using Canny Edge Detector, and control point identification at the vessel bifurcations using the adaptive exploratory algorithm. The shape similarity criteria are employed to fit the control points. The new fusion approach implements the Mutual-Pixel-Count (MPC) maximization based heuristic optimization procedure, which adjusts the control points at the sub-pixel level. MPC is the new measurement criteria for fusion accuracy being proposed. This study achieved a global maxima equivalent result by calculating MPC local maxima with an efficient computation cost. The new method is a promising step towards useful clinical tools for retinopathy diagnosis, and thus forms a good foundation for further development.