Unsupervised segmentation of AM using independent component analysis

Oezkurt N., Oezkurt A.

IEEE 15th Signal Processing and Communications Applications Conference, Eskişehir, Turkey, 11 - 13 June 2007, pp.707-710 identifier

  • Publication Type: Conference Paper / Full Text
  • Volume:
  • City: Eskişehir
  • Country: Turkey
  • Page Numbers: pp.707-710
  • Dokuz Eylül University Affiliated: Yes

Abstract

In this study, an autonomous classification and segmentation algorithm to diagnose, and trace Multiple Sclerosis (MS), from Magnetic Resonance (MR) imaging is developed. In this method, new image stacks are derived by using three diffrent weighted MR images, and then, independent components are obtained from those images derived. A decision maker is developed in order to choose the most suitable independent component by using spatial tone and MRI tissue properties.