Home // PATTERNS 2010, The Second International Conferences on Pervasive Patterns and Applications // View article

Tissue Classification from CT of Liver Volumetric Dataset Using 3D Relational Features

Authors:
Wan Nural Jawahir Hj Wan Yussof
Hans Burkhardt

Keywords: invariant features, relational kernels, Computed Tomography (CT)

Abstract:
This paper proposes an extension of two dimensional relational features into three dimensions. In two dimensions, the relational features are extracted using a non-linear kernel function. This function is applied to the values of the points of two circles. To extract 3D relational features, we represent the points on the two spheres. We aim at classifying the tissue from Computed Tomography (CT) of liver datasets into three classes; normal, abnormal and others(i.e., kidneys, blood vessel, etc.).For this task, 100 known points from 6 CT datasets were used for training using the Support Vector Machine (SVM) classifier and 150 points from 10 CT datasets were used for validation. The results presented in this paper show that the relational features are promising.

Pages: 48 to 52

Copyright: Copyright (c) IARIA, 2010

Publication date: November 21, 2010

Published in: conference

ISSN: 2308-3557

ISBN: 978-1-61208-111-3

Location: Lisbon, Portugal

Dates: from November 21, 2010 to November 26, 2010