Figure 2

Segmentation of muscles in 3D time-lapse Images. Muscle cells are visualized in vivo using the UAS-Gal4 system with the Mef2-Gal4 driver and the fluorescent reporter proteins Grasp65-GFP (green) and Histone-mKO (red). The segmentation of 3D time-lapse image data consists of three major steps. First, after the user places spherical seeds in selected muscles in frame k, Bayesian level-set based active surface evolution detects the cytoplasmic regions (green). Second, the surfaces of detected objects are projected into the next frame k+1 to initialize and perform surface evolution of the same muscles in the next time point. Third, an automated 3D multi-level-set based nuclear segmentation method is applied to detect syncytial nuclei (red) within the confines of muscles cells.