Virtual tomography of a polymer gel's structure from a single SEM image

Area of Science:

• Materials Science
• Computational Imaging
• Biotechnology

Background:

• Electron scanning microscopy (ESM) provides high-resolution 2D images of materials.
• Reconstructing 3D structures from 2D data is crucial for understanding complex architectures.
• Existing methods may struggle with invisible interconnections and require specialized software.

Purpose of the Study:

• To develop a computational framework for 3D reconstruction from 2D ESM images.
• To enable visualization of invisible interconnections within porous materials.
• To create a DICOM-compliant dataset for analysis using standard medical imaging software.

Main Methods:

• Depth estimation from perspective- and depth-of-field-corrected grayscale levels.
• Skeletonization of images at equal depth intervals.
• Gaussian convolution transform for adjusting skeletons to visible counterparts.
• Assembly of image slices into a DICOM-compliant stack.

Main Results:

• Successful conversion of 2D ESM images to 3D objects with reconstructed interconnections.
• Generation of DICOM-compliant datasets for 3D visualization.
• Application to porous organogel structures demonstrating chemomechanical properties.

Conclusions:

• The developed framework enables accurate 3D reconstruction of complex porous materials from 2D ESM data.
• The DICOM output facilitates advanced spatial analysis using existing medical imaging tools.
• This approach enhances the study of materials with intricate internal architectures, such as organogels.