Project: Data-Driven Spine Detection for Low Resolution Multi-Sequence MRI

by Daniel Kottke, Gino Gulamhussene.

Performing statistically significant analysis with medical images requires data from many test persons. Using medical imaging, this is only applicable by reducing quality. Furthermore, research on vertebrae deformation should not rely on top-down (model-based) methods. Our proposed algorithm automatically detects the central spinal curve with 3D data-driven methods on multi-sequence magnetic resonance imaging (MRI). Additionally, we implemented a naive edge operator for vertebrae border detection that can be used for the statistical evaluation with some fast user interaction. First, we segment the spinal canal, based on multi-sequence image analysis and extract a central polynomial. During the second step, we edit this curve by moving it towards the spine center. Finally, an edge operator pre-calculates border estimates for vertebrae, that are corrected by fast user interaction afterwards. We show, that our algorithm correctly detects more than 90% of all spines and allows statistical analysis of vertebrae heights with just 12 seconds interaction per spine. This approach enables further research on huge studies on spine deformation. Working on a segmentation algorithm for vertebrae, this will produce reliable results, that supports medical analysis dramatically.

To get access to the code, contact as. It is hosted at:
https://bitbucket.org/dkottke/parsos/

This work will be published soon.