Retinal stroke prediction using logistic-based fusion of multiscale fractal analysis

Medical images enhanced at different scales have shown to give good results in mammogram enhancement, tumor classification, and lung nodule detection. However, the studies of multiscale analysis on retinal vasculature have been primarily limited to vessel segmentation. We propose using the Fourier Fractal dimension (FFD) to extract the complexity of the retinal vasculature enhanced at different wavelet scales and a model was formulated for stroke prediction. Sensitivity and specificity were found to be 72.52% and 69.67% respectively.