Sep OsiriX Segmentation plugin
student: Brian Jensen
Advisor: Prof. Dr. Nassir Navab
Supervision by: Ralph Bundschuh
is an open source DICOM image viewer for Mac OS X offering advanced image visualization capabilities. The goal of this project was to create a plugin for the OsiriX
DICOM viewer in order to add support for semi automatic segmentation of tumors in fused PET / CT images. After accepting the selection from a user the plugin offers the ability to perform several different kinds of region growing algorithms. The plugin can perform the segmentation on the image itself, or optionally on a registered image, such as a PET image in CT/ PET fused image setup, after which the segmented volume is then mapped back onto the main image. The plugin also provides methods simple integration with other image processing frameworks.
This project was conducted as a cooperation between the Chair for Computer Aided Medical Procedures and Augmented Reality at the TU München, and Department of Nuclear Medicine at the Univserity Hospital Rechts der Isar, TU München.
is a native Mac OS X application and as such is mainly written in objective-c. The main image processing framework used in the plugin is the Insight Segmentation and Registration Toolkit (ITK) which is written using C++. Because of these constraints most of the plugin was written in objective-c++ (the interface and OsiriX
interaction), and the segmentation code was written in pure C++. In addition a shell script was written for the automatic refactoring of the ITK namespace.
The user interface has support for three standard ITK region growing algorithms: connected threshold, neighborhood connected threshold and confidence connected. A fourth algorithm was developed using the ITK framework, gradient threshold. This custom threshold algorithm operates on image gradient magnitudes in the image, segmenting voxels whose gradient magnitude falls below a certain ratio starting with from a seed point.
Latest release: 14.07.09
Here are some screenshots of the OsiriX
plugin NMSegmentation in action.
This screenshot shows the results after performing a connected threshold segmentation on the region of interest
This screenshot shows the results using the gradient magnitude threshold segmentation algorithm on the same region
This screenshot shows the results of the gradient threshold segmentation algorithm on a different lesion type