Seg3D

Seg3D is a free, open-source volume segmentation and image processing tool from the University of Utah. It is aimed at researchers, students and technically minded clinicians who want to label structures in CT, CBCT or MRI volumes with a mix of manual painting and filter-based tools, then export label maps or 3D surfaces. For dental work that means isolating a mandible, teeth or a lesion on a CBCT scan and exporting an STL surface. If you searched for “Seg3D download”, read the status note below first: the program still works, but it has not had a new release since 2021.

Seg3D was developed by the NIH Center for Integrative Biomedical Computing at the Scientific Computing and Imaging (SCI) Institute, University of Utah, and is released under the MIT licence. The old seg3d.org address now redirects to the project’s GitHub repository. The last release is 2.5.1 from May 2021, the last code change on the main branch dates from April 2023, and in August 2026 a volunteer opened a public tracking issue to revive the build. Until that effort produces a new release, treat Seg3D as stable but unmaintained software.

Seg3D at a glance

What it does

• Imports volumes from a single file or an image series, including DICOM series, with a dialog that asks whether the data is an image or a label map.
• Layer-based workflow: each mask is its own layer, grouped with the volume it belongs to.
• Manual tools: paint brush, polyline and speedline for tracing structures.
• Filter-based segmentation from the Insight Toolkit: threshold, neighborhood connected, confidence connected, level set, Canny edge, and the grow cut tool added in 2.5.
• Mask operations: Boolean AND/OR/XOR, connected components, fill holes, plus data filters such as Gaussian blur, median and resampling.
• Computes isosurfaces from masks and exports them as STL (binary or ASCII), OBJ or VTK meshes.
• Exports segmentations as NRRD, MATLAB .mat, TIFF, PNG, BMP or DICOM.
• Volume rendering with transfer functions, clipping planes, provenance tracking of every step and a Python console for scripting.

How dental teams use it

Segment a mandible from CBCT and export STL

1. Start a new project, then use File > Import Layer From Image Series and select the CBCT DICOM files; choose “data” in the import dialog.
2. Apply a median or Gaussian filter to reduce CBCT noise.
3. Run the threshold tool to create a bone mask, then use connected components to keep only the mandible.
4. Clean contacts with the maxilla using the paint brush in erase mode, and fill holes.
5. Compute the isosurface for the mask and export it as binary STL for your mesh editor or printer.

Measure a lesion volume for a research case

1. Place a seed inside the lesion and use confidence connected or level set segmentation to grow the region.
2. Correct the borders with the paint brush on individual slices.
3. Check the mask in the 3D view and export it as NRRD for analysis in other software.

Teach image processing with a real scan

1. Load an anonymised CBCT and let students apply filters step by step.
2. Use the provenance window to show how each mask was produced.

Install and first run

• Windows: run the 2.5.1 64-bit installer. The documentation lists Windows 7-10; on Windows 11 test it with a sample scan first.
• macOS: the .pkg installer is not signed or notarised. Right-click (Control-click) the package and choose Open to install it. The package dates from 2021 and may not run on current macOS versions.
• Linux: there are no binaries. You must compile from source, and the tracking issue opened in 2026 reports that the build no longer configures with current CMake without changes.
• Hardware: the documentation recommends at least 4 GB RAM (8 GB or more preferred) and an OpenGL 2.0 capable graphics card.
• First run: Seg3D works in projects. Create one before importing data, and save it often.

Limitations

• No release since May 2021 and no ongoing funded maintenance; a revival effort only started in 2026.
• Installers are old: Windows 7-10 era builds, an unsigned macOS package and no Linux binaries.
• No dental-specific or AI-based automatic segmentation.
• The interface assumes some image-processing knowledge; it is less approachable than InVesalius.
• No implant planning, guide design or clinical reporting.
• It is a research tool; the developers do not present it as a medical device.

Seg3D vs alternatives

For new projects, most dental users will be better served by actively maintained tools. ITK-SNAP covers similar semi-automatic segmentation and received a new release in 2025. 3D Slicer adds automatic tooth and jaw segmentation through the DentalSegmentator extension. InVesalius is the quickest route when you only need a threshold-based STL from a CBCT. Seg3D still makes sense if you already have Seg3D projects or rely on its filter toolbox and provenance tracking.

FAQ

Is Seg3D free to download?

Yes. Seg3D is free and open source under the MIT licence. Installers for Windows and macOS are attached to the releases on the project’s GitHub page.

What is the latest Seg3D version?

The latest release is Seg3D 2.5.1 from May 2021. There has been no newer release since then.

Is Seg3D still maintained?

Not actively. The last substantive code work was in 2022, and a volunteer revival effort was announced on GitHub in August 2026, but no new release has come out of it yet.

What is Seg3D 2?

Seg3D 2 is the second generation of the program; all current releases, including 2.5.1, belong to this line.

Where is the Seg3D GitHub repository?

The source code, releases and documentation are at github.com/SCIInstitute/Seg3D, maintained under the SCI Institute at the University of Utah.

Does Seg3D run on a Mac?

There is a macOS installer for 2.5.1, but it is unsigned and dates from 2021, so it may not work on current macOS versions.

Can Seg3D export STL files?

Yes. Compute an isosurface for a mask and export it as STL, OBJ or VTK.

Last checked against official sources: 26 September 2026 (developer website: github.com). Versions and licence terms change — confirm on the developer's site before deploying in a clinic.