An automatic image segmentation scheme for phalanx and epiphyseal/metaphyseal region of interests

Abstract

Bone age assessment is usually to diagnose the growth disorders, growth hormone deficiency, and the effect of growth therapy of children in clinic. The crucial role of Bone age assessment is that it is dependent on the accurate bony segmentation from soft tissues on hand radiogram, especially in the separation of phalanx and epiphyseal/metaphyseal region of interests from the background and soft tissue. In this study, our proposed method was to locate and crop the phalanx region of interests for the first step. Next, the integration of a difference in sub-neighborhood technique and average threshold of image processing was applied to segment the epiphyseal/metaphyseal region of interests. The experiments were shown to be successful in the segmentation on different parts of knuckles, i.e. distal, middle, and proximal. In conclusion, the proposed procedure can be implemented in the segmentation of epiphyseal/metaphyseal region of interests automatically and efficiently.