Image Processing and Segmentation
The Image Processing and Segmentation project is developing novel, well-validated approaches to brain image segmentation, structure analysis, and combined volume and surface registration.
By extending our earlier work on cortical surface modeling and expanding our existing algorithms, we will be able to improve our segmentation of the cortex. This will enable us to create better measures of cortical thickness, improve our methods for labeling landmarks on the cortex, and enable us to extend our work to the cerebellum.
The first stage of most brain atlasing efforts is the processing of neuroimaging data. Neuroanatomical structures must be identified clearly so that they can be further analyzed. Correspondences must be made between key features in the images or within the identified structures so that mappings can be constructed that allow information from different subjects to be aggregated and studied within statistical frameworks.
The Image Processing and Segmentation Project has the following aims:
- To further improve the performance and functionality of BrainSuite with particular emphasis on automated cortical thickness estimation, interactive sulcal labeling, and application to mapping of the cerebellum.
- To develop and validate methods and software for automatic labeling of cortical anatomy and intersubject cortical registration.
- To develop methods for combined surface and volumetric registration with the goal of producing an integrated approach to brain image analysis in which volumetric alignment of structures is achieved while simultaneously aligning sulcal and gyral features.
Ultimately, the goal of this project is to produce well-validated tools that identify and align brain features across subjects. We will integrate these tools into our interactive BrainSuite product and also develop stand-alone command line modules that can be used within the LONI Pipeline Environment or independently.