Valuable medical information can be achieved by analysing shape and appearance of human chromosomes. Karyotype, an image of collection of all 23 pairs of human chromosomes, is usually used for this purpose. Making a Karyotype is hard and time consuming, encouraging experts in the image processing and machine vision field to work towards an automatic Karyotyping method. The first step in automation of this process is to define the geometric (morphologic) and intensity based features of the chromosome originating mostly from its banding pattern. As part of a complete project, which is defined to develop a new knowledge based classification technique for chromosomes, a number of new features in addition to the commonly used geometric and intensity based features, are introduced in this paper. Some of the features are computed using the so-called medial axis transform (MAT). For an accurate determination of most of these features it is necessary, however, to identify some key points or landmarks in the image (mostly over the MAT). This paper describes novel algorithms developed to locate such landmarks as centromere, end points of chromosome and two points defined as branching points on the chromosome axis. The algorithms have been tested on the real images supplied by the cytogenetic laboratory of Cancer Institute, University of Tehran. The automatically defined positions of the landmarks have been compared to those manually identified by an expert. In most of the cases the results were in complete agreement.
Automatic landmark detection on chromosomes' images for feature extraction purposes
2003-01-01
284045 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Automatic Landmark Detection on Chromosomes' Images for Feature Extraction Purposes
British Library Conference Proceedings | 2003
|Automatic archaeological feature extraction from satellite VHR images
Online Contents | 2010
|Extraction of significant regions in color images for landmark identification
BASE | 2006
|Automatic archaeological feature extraction from satellite VHR images
Elsevier | 2009
|Landmark Detection in Orbital Images Using Salience Histograms
Online Contents | 2010