Hand radiograph image segmentation using a coarse-to-fine strategy

Image segmentation techniques have been widely applied in diagnosis systems with medical image support. Information about metaphyseal and epiphyseal regions is crucial in bone age assessment. In this study, hand radiograph images have been segmented using a coarse-to-fine strategy. Watershed transform is first done to get metaphyseal regions at a coarse level. Some image processing algorithms, such as noise removal, labeling, and ellipse region fitting, are performed to find the epiphyseal regions of interest (ROIs). The epiphyseal regions are extracted using an active contour model approach based on GVF (gradient vector flow) at a fine level. Some hand radiograph images are processed to show the validity of the proposed approach.