Graph-based segmentation and classification methods for diffusion tensor images (DTI) of the brain

• Grant number: 12/01250-8
• Support Opportunities: Scholarships abroad - Research Internship - Post-doctor
• Effective date (Start): August 01, 2012
• Effective date (End): July 31, 2013
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Principal Investigator: Roberto de Alencar Lotufo
• Grantee: Leticia Rittner
• Supervisor: Oge Marques
• Host Institution: Faculdade de Engenharia Elétrica e de Computação (FEEC). Universidade Estadual de Campinas (UNICAMP). Campinas , SP, Brazil
• Research place: University of Pennsylvania, United States
• Associated to the scholarship: 09/17130-9 - Diffusion tensor images segmentation in the context of mathematical morphology, BP.PD

Abstract

This project aims to identify and adapt, if necessary, segmentation and pattern recognition methods suitable to work with diffusion tensor images (DTI) of the brain. As a starting point, we will explore graph-based methods, such as image operators defined through the Image Forest Transform (IFT) framework and the Optimum path forest (OPF) classifier, which calculates an optimal path forest in a training set and assigns to each sample the label of the root with which it is more closely connected. In order to apply these techniques to diffusion tensor images (DTI), we must investigate the necessary pre-processing steps and the best attributes, metrics, cost functions and evaluating criteria for this kind of image. Then, the adequate classifier configuration should be proposed for this imaging modality as well as its training procedure. Eventually, other image modalities can be also used to generate more robust methods. Finally, real segmentation and classification problems should be addressed using the proposed attributes and classifiers, such as: brain tissue classification; automatic identification of white matter lesions, identification of regions of white matter atrophy; among others. (AU)