Cost function evaluation for the registration of clinical DTI images onto the ICBM DTI81 white matter atlas. Academic Article uri icon



  • BACKGROUND AND PURPOSE: Various algorithms are available for the analysis of diffusion tensor (DTI) images. Many of these stand alone software packages require time-intensive user interactions not yet suited for routine clinical application Here, we demonstrate the use of the 'Analysis of Functional NeuroImages' (AFNI) software package, a standard for the analysis of functional magnetic resonance images (fMRI), to automatically align clinical DTI images onto the ICBM DTI81 atlas potentially enabling the combined presentation of fMRI and DTI results. METHODS: Fractional anisotropy (FA) maps from seven patients diagnosed with video/EEG defined complex partial seizures were retrospectively analyzed. Affine transformations parameters for seven different cost functions provided by the 3dAllineate software tool were calculated. Alignment quality and variations of the transformation parameters were assessed. RESULTS: Best alignment between the FA maps for each subject and the ICBM DTI81 atlas was achieved with cost functions utilizing the cost ratio (CR) (symmetrized* CR, symmetrized+ CR and unsymmetrized CR). Symmetrized* CR performed slightly better, in particular for lateral white matter structures. Relatively small variations in the transformation parameters emphasize the robustness of the transformations. CONCLUSIONS: Good alignment of FA maps to the ICBM DTI81 white matter atlas can be achieved using an automated affine transformation with software tools provided by AFNI potentially enabling the combined presentation of fMRI and DTI information. This procedure maybe readily be applied in clinical practice.

publication date

  • January 1, 2010



  • Brain Mapping
  • Diffusion Tensor Imaging
  • Epilepsy, Complex Partial
  • Image Interpretation, Computer-Assisted


Scopus Document Identifier

  • 77954596803

Digital Object Identifier (DOI)

  • 10.3233/THC-2010-0578

PubMed ID

  • 20495254

Additional Document Info


  • 18


  • 2