plastimatch

Google Summer of Code 2012: Project Ideas

Plastimatch is an open source medical image computation project, focusing on high-performance algorithms for image registration, image segmentation, and image reconstruction.

This page describes a list of project ideas for Google Summer of Code 2012. The projects have been designed to be (1) interesting, (2) useful, and (3) achievable within the 3 month time limit. Other projects that are not on this list are also welcome – please contact us using email list. If you are interested in participating, please familiarize yourself with our software infrastructure:

3D-Slicer plugin for DRR generation

Description: Plastimatch currently runs as a plugin for 3D Slicer. There are already existing plugins for B-spline registration, DICOM-RT interchange, and dose comparison tools, however there is no plugin for DRR generation. In this project, you would create a new command line plugin to allow users to create DRRs in 3D-Slicer. The project could be expanded to allow interactive selection of source and target positions using Slicer’s 3D viewer.

Difficulty: Basic to intermediate

Skills: C++

Mentors: Greg Sharp

OpenCL-accelerated FDK tomographic reconstruction

Description: The FDK algorithm is the most popular method for creating 3D volumetric images from cone-beam projections. Although a basic OpenCL implementation exists, it is not very fast because it doesn’t take advantage of hardware interpolators. The goal of this project is to impove the OpenCL FDK implementation to use hardware interpolation. The project can be expanded to applying hardware interpolation to DRR generation.

Difficulty: Basic to intermediate

Skills: C++

Mentors: James Shackleford

Running registration with python

Description: Python is a becoming a useful and popular language for scientific computing. This project is to design a framework to use python for imaging pre-processing with plastimatch for registration. Ideally, the images can be loaded and manipulated in python, and then registered with plastimatch directly in shared memory. This project could be expanded to give python fine grain control over registration stage parameters.

Difficulty: Basic to intermediate

Skills: C++, python

Mentors: Paolo Zaffino

Your project here

Description: We invite your project proposal related to medical image computation.

Difficulty: Varies

Skills: Varies

Mentors: Varies