Fig 1.
Preprocess: Acquisition of a Z-stack. 3D cell segmentation is performed to obtain a 3D labelled image. Prerequisite: Create single cell files for faster analysis. Possibility to create 3D object file for 3D rendering. CellMet analysis: Analysis of the shape of cells and neighbour relationship. Export: Results are stored in .csv files
Fig 2.
3D and 2D cell visualisation of metrics measured in CellMet for cell (top left), cell plane (top right), edge (bottom left) and face (bottom right).
Fig 3.
Application to Drosophila gastrulation.
A. Multi orthoview of PH-mCherry embryo during mesoderm invagination from Gracia [57] (left) and cell identity (right). B. Colour coded cells show volume (left) and neighbour number (right) in the embryo. C. Scheme showing tissue invagination (left) and the corresponding connectivity graph from A (right). The top row is before gastrulation and bottom row is during gastrulation. Node colour represents the number of neighbours and edge colour represents the link between two cells. D. Violin plot of the number of neighbours from C. Measures made in 2D at the apical cell surface (brown) and in 3D (cyan). cells from one embryo.
Fig 4.
A. Colour coded face in prism (left) and scutoid (right) cells. Edges are in black. B. Change in number of neighbours and area along cells. Blue and orange represent a prism and scutoid cell respectively. C. 3D morphology measures for different cell shapes extracted from CellMet.
Fig 5.
Application of CellMet to different tissues.
A. (Left) Schematic of the Drosophila embryo from a dorsal view, with anterior up. The red dashed region corresponds to the images to the right. (Centre) Heart cells in a stage 16 Drosophila embryo, just as the heart lumen forms with segmentation from CellPose. (Right) Connectivity of the heart cells, demonstrating the cell alignment. Numbers correspond to cell labels ID. B. (Left) schematic of the zebrafish embryo, highlighting the myotome region with red box. (Centre) Myotome in 48hpf zebrafish embryo, with segmentation from CellPose. Green dashed line shows the chevron form of the myotome. Bottom shows cross-sectional view. (Right) 3D visualisation of muscle fibres showing the different cell morphologies. For the twisted cell, we can trace the rotation along the cell long axis. Vertex colours are the same as edges showed in 3D. C. (Left) Schematic of organoid culture constrained to triangular domain. (Centre) Organoid culture at 48 hours, with CellPose segmentation. Views shown in x-y and x-z axes. (Right) Comparison of cell sphericity with cell connectivity. Scale bar: A: ; B and C:
.
Fig 6.
Execution time on synthetic data using a computer with ncore = 10, and RAM = 64 GB. A. Relationship between the size of the image (i.e. number of pixels) and the duration of execution. B. Cumulative execution time according to the number of cells in the image. Each subprocess is colour coded (legend), to show the specific time demands.
Table 1.
Comparison of software packages for analysing segmented data.
Table 2.
Comparison of software packages aimed at extracting metrics from segmented data.