A Segmentation Method for Bone Marrow Cavity Imaging Using Graph Cuts

Abstract

The improvement of bioimaging technologies enables the observation of cellular dynamics invivo. Some new bioimaging technologies are expected to contribute to the discovery of new drugs and mechanisms of disease. To improve the contributions of bioimaging, it is required to extract a particular region or to detect a particular cell's motion within bioimages. Moreover, automatic extraction and detection with image processing is also required because the accurate and uniformed processing of a massive number of images manually is unrealistic. To help automate this process, we introduce a bone marrow cavity segmentation method for two-photon excitation microscopy images. Specialists of cellular dynamics define regions of bone marrow cavity by considering several criteria, including characteristics of intensity and blood flow. We take those criteria into our method as the energy function of graph cuts. Results of evaluations and comparison with normal graph cuts show that our proposed method that does not use hard constraints achieved a performance better than normal graph cuts with hard constraints.