Difference between revisions of "Projects:BayesianMRSegmentation"
| Line 11: | Line 11: | ||
We use a Bayesian modeling approach, in which we build an parametric computational model of how an MRI image around the hippocampal area is generated. The model incorporates a ''prior'' distribution that makes predictions about where neuroanatomical labels typically occur throughout the image, and is based on a generalization of probabilistic atlases that uses a deformable, compact tetrahedrical mesh representation. The model also includes a ''likelihood'' distribution that predicts how a label image, where each voxel is assigned a unique neuroanatomical label, translates into an MRI image, where each voxel has an intensity. | We use a Bayesian modeling approach, in which we build an parametric computational model of how an MRI image around the hippocampal area is generated. The model incorporates a ''prior'' distribution that makes predictions about where neuroanatomical labels typically occur throughout the image, and is based on a generalization of probabilistic atlases that uses a deformable, compact tetrahedrical mesh representation. The model also includes a ''likelihood'' distribution that predicts how a label image, where each voxel is assigned a unique neuroanatomical label, translates into an MRI image, where each voxel has an intensity. | ||
| − | Given an image to be segmented, we first estimate the parameters of our model that are most probable in light of the data. This parameter estimation involves finding the deformation that optimally warps the mesh-based probabilistic atlas onto the image under study, as well as the | + | Given an image to be segmented, we first estimate the parameters of our model that are most probable in light of the data. This parameter estimation involves finding the deformation that optimally warps the mesh-based probabilistic atlas onto the image under study, as well as finding the mean intensity and the intensity variance for each of the hippocampal subfields in the image. Once these model parameters are estimated, the model is then used to obtain fully automated segmentations. |
= Results = | = Results = | ||
Revision as of 21:10, 16 May 2008
Home < Projects:BayesianMRSegmentationBack to NA-MIC Collaborations, MIT Algorithms
Model-Based Segmentation of Hippocampal Subfields in In Vivo MRI
Recent developments in MR data acquisition technology are starting to yield images that show anatomical features of the hippocampal formation at an unprecedented level of detail, providing the basis for hippocampal subfield measurement. Because of the role of the hippocampus in human memory and its implication in a variety of disorders and conditions, the ability to reliably and efficiently quantify its subfields through in vivo neuroimaging is of great interest to both basic neuroscience and clinical research. The aim of this project is to develop and validate a fully-automated method for segmenting the hippocampal subfields in ultra-high resolution MRI data.
Bayesian modeling and inference
We use a Bayesian modeling approach, in which we build an parametric computational model of how an MRI image around the hippocampal area is generated. The model incorporates a prior distribution that makes predictions about where neuroanatomical labels typically occur throughout the image, and is based on a generalization of probabilistic atlases that uses a deformable, compact tetrahedrical mesh representation. The model also includes a likelihood distribution that predicts how a label image, where each voxel is assigned a unique neuroanatomical label, translates into an MRI image, where each voxel has an intensity.
Given an image to be segmented, we first estimate the parameters of our model that are most probable in light of the data. This parameter estimation involves finding the deformation that optimally warps the mesh-based probabilistic atlas onto the image under study, as well as finding the mean intensity and the intensity variance for each of the hippocampal subfields in the image. Once these model parameters are estimated, the model is then used to obtain fully automated segmentations.
Results
Key Investigators
- MIT Algorithms: Koen Van Leemput, Polina Golland
