Abstract #1387

Non-Rigid Registration of Diffusion Weighted Images Using Fibre Orientation Distributions

David Raffelt^1,2, J-Donald Tournier^3,4, Jurgen Fripp⁵, Natasha Lepore⁶, Stuart Crozier², Alan Connelly^3,4, Olivier Salvado¹

¹The Australian E-Health Research Centre, CSIRO, Brisbane, QLD, Australia; ²Department of Biomedical Engineering, University of Queensland, Brisbane, QLD, Australia; ³Brain Research Institute, Florey Neuroscience Institutes (Austin), Melbourne, VIC, Australia; ⁴Department of Medicine, University of Melbourne, Melbourne, VIC, Australia; ⁵The Australian E-Health Research Centre, CSIRO,, Brisbane, QLD, Australia; ⁶Laboratory of Neuro Imaging, David Geffen School of Medicine UCLA, Los Angeles, CA, USA

Fibre Orientation Distributions (FODs) obtained from diffusion weighted imaging (DWI) provide a non-invasive method to investigate white matter structure and connectivity. In order to compare white matter between individuals, spatial normalisation is required. We propose a free form deformation registration method using a novel similarity measure that optimises directly on FODs. Our technique compared favourably with existing T1 weighted and fractional anisotropy based methods: average images were sharper indicative of more accurate spatial alignment. Unlike existing scalar or diffusion tensor based techniques, this method exploits information provided by FODs in voxels with crossing fibres, independently of how FOD are computed.