Credits: Russell Garwood, David Smith, Yongle Sun, Martin Turner (Manchester), Jon Hill (York), Susheel Varma (Sheffield), Gowsihan Poologasundarampillai (Birmingham & DLS), Alberto Gambaruto (Bristol).

Disclaimer: The list and descriptions are unavoidably subjective. Please get in touch if you notice any inaccuracies.

Summary: MeshLab, Gmsh, Simpleware are good GUI tools for simple-to-medium mesh-generation; for high-throughput tasks, use script-based Gmsh, Netgen, TetGen, CGAL, iso2mesh (all are highly flexible), and also ANSYS ICEM.

Tips: test the final mesh quality by plotting the distribution of dihedral angles. See also an overview of meshing tips for COMSOL: Part I, Part II.

Further Tools and Resources: Quantification and Segmentation Tools for Micro-Tomography, Software Catalogue for Medical Imaging (e.g. DICOM); NiftyWeb & Nifty (online MRI/CT segementation; NiftyNet is ML-based); A comparison of CT-segmentation tools; MIRTK and Deformetrica (MRI-oriented multi-modal image registration and transformation). Meshing tools: Mesh Repair Sofware (e.g. see ReMESH), SnappyHexMesh from the OpenFOAM suite.

Credits: Kathy Tinoco, Cornelia Vacar (FEI), Philip Pearce (MIT), James Carr, Alexander Erlich, Sebastian Gilbert and David Smith (Manchester)

Manual Skeletonisation as an alternative to Auto Skeleton module:

1. Compute a distance map from your label field, using Distance Map module. Set the Chamfer Weights to float
2. Convert the result to 8-bit data via Convert Image Type module
3. Generate a centerline tree (spatial graph) with Centerline Tree module (this step may take a while)
4. Centerline Tree module has Tube Parameters (slope and zeroVal) that help to reduce the number of false side branches. E.g., for a thin noisy object, large zeroValue will account for the noise when tracing a branch, and for a thick object with irregular surface, large slope will allow for larger local thickness variations.

Segmentation Tips for complex / noisy data:

1. Weka is a 2D/3D machine-learning Fiji plugin, useful for pre-processing
2. Filter Sandbox and Marker-based Watershed Mask are semi-automatic segmentation modules in Amira
3. SuRVoS is a powerful combination of 3D super-voxel segmentation and trainable classifiers (random forest, etc.)
4. MIB is another tool for 2D super-pixel segmentation and has some machine learning functionality (X-ray ZEISS .txm needs to be exported to .mrc/.tiff stacks first)