Hand Segmentation and Joint Localization in Fluorescence Optical Imaging

Abstract

An essential step in the automated analysis of fluorescence optical imaging (FOI) sequence data for rheumatic diseases of the hands lies in the precise detection of the hands and joint positions. We demonstrate the application and derivation of a hierarchical algorithm that enables a precise segmentation of each patient’s hands, relying on geometrical constraints and highly-adaptive thresholding-like approaches. The improvements made compared to reference solutions are demonstrated. In particular, it is shown that—based on the reliable segmentation of the hand—one can robustly detect the joint positions in the hand by morphological constraints based on biological principles. Ways to further improve on our findings are suggested, and the applicability of current state-of-the-art instrumental machinery is demonstrated.