Reconstruction of 3D cellular models from acquired image stacks

3D cellular models were reconstructed from acquired z-stack images of the cellular membrane by using MATLAB (MathWorks, Massachusetts, USA). As a preparation for the 3D model reconstruction, the cells to be analyzed were cropped from z-stack images, and surrounding cells were cleared. A median filter was applied to reduce noise, and brightness and contrast were adjusted manually. In the case of confocal images obtained with the 40× dry lens, the intensity of all slices in the z stack was measured for each cell to choose an image slice with the highest intensity as a middle slice (fig. S6). Then, five slice images each above and below the middle slice image, in total 11 slice images, were used for the image projection in a single image and the reconstruction of a 3D cellular model (fig. S6). In the case of the 40× water immersion lens, a total of 31 slice images were used for the 3D model reconstruction because of a higher z resolution. The selected slices were loaded on MATLAB to find boundary points describing the contours of cells and create a 3D volume using the alphaShape function. To acquire a 3D cellular model with triangular meshes, a matrix representing the boundary of the alpha shape was acquired using the boundaryFacets function, and an STL file was created and visualized using the stlwrite function and Partial Differential Equation Toolbox (MathWorks, Massachusetts, USA).