Welcome

This is the Wiki of the Knee Segmentation and Registration Toolkit (KSRT)

Overview

KSRT is an open source software distribution of algorithms to automatically quantify cartilage thickness from magnetic resonance (MR) images and to perform related statistical analyses. The implemented algorithms are described in more detail here .

See also Liang Shan's PhD defense slides for an overview and explanation of the available algorithms.

In particular, the software allows to automatically perform the following analysis tasks:

1. Segmenting cartilage from knee Magnetic Resonance Imaging (MRI) data.
2. Computing cartilage thickness from the automatic cartilage segmentations.
3. Establishing spatial correspondence across MRI data appropriate for statistical analysis.
4. Performing statistical analysis of localized cartilage changes for cross-sectional and longitudinal data.

A software manual can be downloaded as a [PDF] or accessed through [HTML]

We just recently greatly simplified the build procedure (and are working to update the software manual accordingly). All that is needed now to build the toolbox are the following steps:

1. git clone git@bitbucket.org:marcniethammer/ksrt.git
2. mkdir ksrt-build
3. cd ksrt-build
4. ccmake ../ksrt
5. make

Resulting images can for example be visualized using Slicer. For a quick way of visualization turn on the ImageViewer compile option which will check out and compile Kitware's ImageViewer.

This work was supported by NIH grant: R21AR059890: Automatic Quantitative Analysis of MR images of the knee in osteoarthritis

Related references are:

• Niethammer M, Pohl K M, Janoos F, Wells, III W M. 2015. Active Mean Fields for Probabilistic Image Segmentation: Connections with Chan-Vese and Rudin-Osher-Fatemi Models. ArXiv e-prints.
• Shan L. 2014. Automatic localized analysis of longitudinal cartilage changes.
• Shan L, Zach C, Charles C, Niethammer M. 2014. Automatic Atlas-based Three-label Cartilage Segmentation from MR Knee Images. Medical Image Analysis.
• Huang C, Shan L, Charles C, Niethammer M, Zhu H. 2013. Diseased Region Detection of Longitudinal Knee MRI Data. Proceedings of the Conference on Information Processing in Medical Imaging (IPMI). 7917:632–643.
• Shan L, Charles C, Niethammer M. 2013. Longitudinal three-label segmentation of knee cartilage. Proceedings of the International Symposium on Biomedical Imaging (ISBI).
• Shan L, Charles C, Niethammer M. 2012. Automatic Multi-Atlas-Based Cartilage Segmentation from Knee MR Images. Proceedings of the International Symposium on Biomedical Imaging (ISBI).
• Shan L, Charles C, Niethammer M. 2011. Automatic Atlas-based Three-label Cartilage Segmentation from MR Knee Images. Proceedings of the Workshop on Mathematical Methods in Biomedical Image Analysis (MMBIA).
• Shan L, Zach C, Styner M, Charles C, Niethammer M. 2010. Automatic Bone Segmentation and Alignment from MR knee images. SPIE Medical Imaging.
• Shan L, Zach C, Niethammer M. 2010. Automatic three-label bone segmentation from knee MR images. International Symposium on Biomedical Imaging (ISBI).
• Zach C, Shan L, Frahm J-M, Niethammer M. 2009. Globally Optimal Finsler Active Contours. DAGM. :552–561.