Skip to main content
Fig. 4 | BMC Biology

Fig. 4

From: Extracting multiple surfaces from 3D microscopy images in complex biological tissues with the Zellige software tool

Fig. 4

Cochlea specimen. A Volume rendering of a 3D confocal swept-field image of the mouse cochlear embryo on embryonic day E14.5. The dataset (of dimensions 1024 × 1024 × 45 voxels) shows a portion of the sensory epithelium (at the topmost sections of the stack) and the underlying non-cellular layer of mesenchyme on which the organ develops. Both structures are stained with phalloidin to reveal F-actin. The surface of interest is the epithelium surface, harboring the sensory and supporting cells under differentiation. The mesenchyme layer is not (strictly speaking) assimilable to a surface, but it produces a strong background signal nearby the surface of interest, hampering its extraction. B 3D representations of the height map extracted by Zellige (in green) and the GT height map (in blue), of the epithelium surface. C Color-coded error map of the reconstructed height map, which shows subpixel accuracy (errors <1) over a large majority (83%) of pixels, while being close to 1 pixel on average (RMSE ~ 1.1). D Projections localized to the GT height map of the epithelium surface (left most panel), and to the height maps extracted with the four different algorithms: FastSME, LocalZProjector, PreMosa, and Zellige. Only Zellige correctly extracts the surface of the epithelium in this example. Scale bar 40 μm

Back to article page