Medical image segmentation and visualization

Project reference: 1514

Project will focus on rapid prototyping and testing of different algorithms for image segmentation as well as visualization of 3D data. In a first stage implementation and testing of suitable image segmentation techniques for medical image segmentation will be done. As an input data for image segmentation the consecutive series of CT or MRI medical images will be used. After the segmentation visualization of the obtained 3D data will be performed. In the visualization stage focus will be on suitable pre-processing of the data for further 3D visualization

For rapid prototyping Matlab software will be used. Standard existing functions for image segmentation will be used. However there is possibility that new special functions will be developed and optimized using Matlab’s Parallel Computing Toolbox.

Project mentor: Petr Strakoš

Site Co-ordinator: Karina Pesatova

Learning Outcomes

Student will gain experience in:

•  Image segmentation techniques
• 3D visualisation
• Matlab
• C, C++ programming
• using visualization software

Student Prerequisites (compulsory)

Programming skills in C, C++ and Matlab.

Student Prerequisites (desirable)

Image processing and segmentation, 3D data visualization, parallel processing.

Training Materials

Matlab webinars: http://www.mathworks.com/company/events/webinars/index.html?recordedwebinars

 Segmentation & registration toolkit: http://www.itk.org

 Visualization toolkit: http://www.vtk.org

 Open source image segmentation software: http://www.itksnap.org/pmwiki/pmwiki.php, http://www.slicer.org/pages/Introduction, http://www.osirix-viewer.com/AboutOsiriX.html

 Medical image databases: http://www.osirix-viewer.com/datasets/

 Image segmentation algorithms: http://en.wikipedia.org/wiki/Region_growing, http://en.wikipedia.org/wiki/K-means_clustering, http://en.wikipedia.org/wiki/Flood_fill, http://en.wikipedia.org/wiki/Otsu%27s_method

Workplan

• Week 1:          Training
• Week 2 :         Work plan setting
• Week 3 – 6:    Implementation and testing of image segmentation algorithms in Matlab
• Week 7 :         Visualization of the results
• Week 8 :         Final report completion and final presentation preparation

Final Product Description

Resulting visualizations and animations of the 3D models of internal organs can be used to demonstrate HPC capabilities to the public.

Adapting the Project – Increasing the Difficulty

By implementing more difficult image segmentation techniques and by developing the GUI.

Adapting the Project – Decreasing the Difficulty

By using already implemented functions in the image segmentation stage.

Resources

Software: Matlab, visualisation utilities (ParaView, Visit, Ensight, etc.), C, C++ programming environment

Hardware: high memory system, visualization server, anselm cluster

Access to the appropriate software and hardware will be provided by the IT4Innovations National Supercomputing Center.

Organization