Repository navigation
Releases: Ajaxels/MIB3
Releases · Ajaxels/MIB3
Release list
MIB3 v2026.10 compiled MIB3 for academic research
Microscopy Image Browser 3, version 2026.10
MIB 3 is the new generation of Microscopy Image Browser: a free, open-source MATLAB application for
processing, segmentation, analysis and visualization of multidimensional (2D-4D) microscopy datasets.
It is a complete rewrite of MIB 2 around a new ribbon-based user interface.
Highlights
- New graphical user interface: ribbon with tabs (Home, Dataset, Image, Model, Mask, Selection, Tools,
Plugins), dockable tabbed image documents, a Datasets panel for up to 10 open datasets per set, light and
dark themes - Stitching: assembles 2D and 3D image tiles into a single mosaic; tiles are placed from a grid, a
position file (TrakEM2, Fibics Atlas, SerialEM), a filename pattern or the stage coordinates, then their
overlaps are measured and a globally consistent position is found for every tile - Instance segmentation: 2D instance segmentation with the SOLOv2 network in DeepMIB, where each object
receives its own label so that touching objects are separated; the 2D instances can be linked into 3D
objects - BigData: segmentation of datasets far larger than the computer memory; pyramidal images (OME-Zarr v3
or whole-slide formats such as SVS, NDPI and CZI) are read on demand and the model is kept on disk
Also new
- Zarr v2 and v3 datasets (local and remote, images and models) are read and written with the included
zarr-matlab engine; Python is no longer required (zarr-python remains available as an option) - Whole-slide images can be opened with the OpenSlide reader (MATLAB version; requires the Medical
Imaging Toolbox Interface for Whole Slide Imaging File Reader support package) - External directories for Fiji, Imaris, BM3D, Python and Java are set in one place:
Home -> Preferences -> External directories - Java-free MATLAB: MIB 3 runs with MATLAB R2026b and newer, which no longer include Java; Bio-Formats,
Fiji and Imaris need Java, and MIB guides you through setting it up
(how to enable Java) - New user guide and API reference, also shipped with the standalone versions so that the Help
buttons work offline - Standalone version for macOS on Apple silicon
Downloads
| Distribution | File | Notes |
|---|---|---|
| MIB for MATLAB users version | MIB3_Matlab.zip from mib.helsinki.fi |
Requires MATLAB R2025a or newer (R2026a recommended) - a trimmed version for users; for development, use the full release from github |
| Windows, installer | MIB3_Win_2026-10_R2026a.zip |
Compiled with MATLAB R2026a; the MATLAB Runtime is installed automatically |
| Linux, installer | MIB3_Linux_2026-10_R2026a_installer.zip |
Compiled on Rocky Linux 9.8 with MATLAB R2026a; the MATLAB Runtime is installed automatically |
| Linux, files | MIB3_Linux_2026-10_R2026a_files.zip |
Compiled files without the installer; install the MATLAB Runtime R2026a separately |
| macOS (Apple silicon), installer | MIB3_Mac_2026-10_R2026a.zip |
Compiled with MATLAB R2026a for Macs with Apple silicon (M1, M2, ...); Intel Macs are not supported |
The standalone versions are compiled with an academic MATLAB license and are intended for
non-commercial academic research only, see the license.
- Installation instructions: https://mib.helsinki.fi/downloads_installation.html
- System requirements: https://mib.helsinki.fi/downloads_systemreq.html
Requirements for the MATLAB version
- MATLAB R2025a or newer (R2026a is recommended)
- Image Processing Toolbox (required)
- Deep Learning Toolbox, Computer Vision Toolbox and Parallel Computing Toolbox (required for DeepMIB)
- Add-ons: Computer Vision Toolbox Model for SOLOv2 Instance Segmentation (instance segmentation),
Deep Learning Toolbox Converter for ONNX Model Format and Deep Learning Toolbox Converter for
TensorFlow Models (export of trained networks)
Known limitations
- The connection to an OMERO server is not implemented in MIB 3 yet; use MIB 2 for OMERO
- The connection to Imaris was converted from MIB 2 but has not been tested with MIB 3 yet
- The BM3D/BM4D filters are available only in the MATLAB version (BM3D MATLAB 4.0.3 or newer, installed
separately); they are not included in the compiled versions - macOS: if macOS reports that the installer or MIB is damaged and cannot be opened, remove the quarantine
attribute:xattr -dr com.apple.quarantine [path/to/installer.app]
MIB 2
MIB 3 is a separate application, MIB 2 is not updated to MIB 3 automatically. MIB 2 remains available
for MATLAB releases older than R2025a: github.com/Ajaxels/MIB2 and
mib.helsinki.fi.