WIT Press


Interactive Border Detection In Cellular Images Using A Kohonen Neural Network

Price

Free (open access)

Volume

3

Pages

7

Published

1996

Size

876 kb

Paper DOI

10.2495/BSIM960181

Copyright

WIT Press

Author(s)

V. Barrios, A. Reigosa & G. Montilla

Abstract

During cancer tissue analysis, specialists frequently must observe parameters such as cellular size, shape, arrangement and the amount of cells present, in order to diagnose. The first step in this job is tracing the borders. This work develops an interactive auxiliary method that enables the physician to trace cells' borders as they are selected in a microscopic image. The method is based on gradient and gray-level analysis over a series of radial lines, originating at the selected cell's center. A series of points is obtained from such analysis. Several points are selected as part of the border, using similarity criteria and a slightly

Keywords