A Bayesian Cost Function Applied to Model-Based Registration of Sub-cortical Brain Structures
Morphometric analysis and anatomical correspondence across MR images is important in understanding neurological diseases as well as brain function. By registering shape models to unseen data, we will be able to segment the brain into its sub-cortical regions. A Bayesian cost function was derived for this purpose and serves to minimize the residuals to a planar intensity model. The aim of this paper is to explore the properties and justify the use of the cost function. In addition to a pure residual term (similar to correlation ratio) there are three additional terms, one of which is a growth term. We show the benefit of incorporating an additional growth term into a purely residual cost function. The growth term minimizes the size of the structure in areas of high residual variance. We further show the cost function’s dependence on the local intensity contrast estimate for a given structure.
KeywordsCost Function Growth Term Active Shape Model Correlation Ratio Statistical Shape Model
Unable to display preview. Download preview PDF.
- 2.Fischl, B., Salat, D.H., Busa, E., Albert, M., Dieterich, M., Haselgrove, C., van der Kouwe, A., Killiany, R., Kennedy, D., Klaveness, S., Montillo, A., Makris, N., Rosen, B., Dale, A.M.: Whole brain segmentation: Automated labeling of neuroanatomical structures in the human brain. Neuron 33, 341–355 (2002)CrossRefGoogle Scholar
- 4.Seidman, L.J., et al.: Left hippocampal volume as a vulnerable indicator for schizophrenia. Arch. Gen. Psychiatry 59 (2002)Google Scholar
- 6.Cootes, T.F., Taylor, C.J.: Statistical models of appearance for medical image analysis and computer vision. In: Proc. SPIE Medical Imaging (2001)Google Scholar
- 7.Jenkinson, M., Behrens, T., Woolrich, M., Crum, W., Hill, D., Smith, S.: A bayesian similarity function for segmentation using anatomical, shape-based models. In: Medical Image Understanding and Analysis (MIUA) (2005)Google Scholar
- 8.Jenkinson, M.: A bayesian similarity function for segmentation using anatomical, shape-based models. Technical report, FMRIB, University of Oxford (2005)Google Scholar