Diffusion tensor images segmentation in the context of mathematical morphology

• Grant number: 09/17130-9
• Support Opportunities: Scholarships in Brazil - Post-Doctoral
• Effective date (Start): March 01, 2010
• Effective date (End): May 31, 2014
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Principal Investigator: Roberto de Alencar Lotufo
• Grantee: Leticia Rittner
• Host Institution: Faculdade de Engenharia Elétrica e de Computação (FEEC). Universidade Estadual de Campinas (UNICAMP). Campinas , SP, Brazil
• Associated research grant: 13/07559-3 - BRAINN - The Brazilian Institute of Neuroscience and Neurotechnology, AP.CEPID
• Associated scholarship(s): 12/01250-8 - Graph-based segmentation and classification methods for diffusion tensor images (DTI) of the brain, BE.EP.PD

Abstract

This postdoctoral project proposes the improvement, validation and quantitative analysis of a segmentation method for diffusion tensor images (DTI) of the brain based on the watershed transform [Rittner and Lotufo, 2008]. It also foresees the application of this method in solving real problems of neuroscience, such as segmentation of thalamus nuclei, delineation of regions of the corpus callosum, classification of brain tissue and analysis of diffusion measures in several brain structures. The proposed method defines the tensorial morphological gradient (TMG) based on mathematical morphology, which holds the relevant information contained in the tensor and allows the segmentation of the image through the watershed transform. The results of segmentation based on TMG and hierarchical watershed are consistent with results of atlas-based segmentation. But in order to robustly segment the images of diffusion tensor brain and to solve real problems, it is necessary to improve the computation of the TMG and to propose a new method for selecting the watershed markers. Additionally, the method must be validated and its results objectively compared with other methods in the literature or with manual segmentations, in order to be used by physicians and researchers. This research project will result in a new version of the segmentation method based on watershed and TMG. The enhanced version of the method will be validated, quantitatively analyzed and used in solving problems of neuroscience. (AU)