1. Start with the source
Medical images contain both image values and information about how those values are positioned in space. Before modelling, check that the intended study and series have been selected. Keep the source available so later decisions can be reviewed.
2. Identify the structure
Segmentation means identifying the region you want to represent. A boundary may be drawn, adjusted or proposed by an algorithm. The result still needs review for the particular task and data. An automatic result is not automatically an accurate one.
3. Create a surface
A reconstruction step converts the selected region into a surface representation. The settings can affect shape, detail and connectivity. Keep a version before editing so comparisons have a clear starting point.
4. Review changes
Look at the model alongside the source and inspect areas that are thin, close together or difficult to interpret. Record warnings and unsuccessful results. A good-looking surface is not sufficient evidence of anatomical accuracy.
5. Prepare an export
Check units, coordinate conventions and the receiving application before sharing a model. Some file formats carry less context than a saved project. Keep a separate record of the source and processing steps when the export cannot contain it.
Try it as a learning exercise
Use a synthetic shape or a dataset you are explicitly allowed to use. Write down the input, the settings and one question you want to answer. This guide explains the workflow; it is not a tutorial for an available Wellbeing clinical product.