Advanced search
Start date
Betweenand

Segmentation of medical images through complex networks

Grant number: 12/19092-0
Support type:Scholarships abroad - Research
Effective date (Start): March 01, 2013
Effective date (End): February 28, 2014
Field of knowledge:Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
Principal Investigator:João Do Espirito Santo Batista Neto
Grantee:João Do Espirito Santo Batista Neto
Host: Fernando Bello
Home Institution: Instituto de Ciências Matemáticas e de Computação (ICMC). Universidade de São Paulo (USP). São Carlos , SP, Brazil
Local de pesquisa : Imperial College London, England  

Abstract

Image segmentation plays an important role in many medical applications such as content-based image retrieval, detection of lesions, image registration, identification of structures for surgical simulation and image guided surgery. A wide variety of image segmentation methods can be found in the literature. Among those, there are some based on graphs, in which each node represents a pixel and edges depict a relationship among pixels in a neighborhood. Many graph-based techniques are, however, computationally expensive and fail to identify the best partitioning for the graph. Recent developments in Complex Network theory made possible a variety of graph-based pattern recognition techniques, more specifically community detection algorithms. The use of such algorithms, however, has not been explored for image segmentation purposes, mainly because of the excessive number of nodes while modelling large images, which make the processing impracticable. Nonetheless, recent work carried out by the author of this proposal has shown that satisfactory image segmentation can be attained by combining community detection algorithms with super pixel techniques. The main goal of this project is to extend this preliminary work, especially for 3D medical images, so as to contribute for the tasks of surgical simulation and image guided surgery. (AU)