Abnormal Cell Detection Using the Choquet Integral

R. Joe Stanley, Missouri University of Science and Technology
J. M. Keller
C. W. Caldwell
Paul Gader

This document has been relocated to http://scholarsmine.mst.edu/ele_comeng_facwork/837

There were 1 downloads as of 27 Jun 2016.

Abstract

Automated Giemsa-banded chromosome image research has been largely restricted to classification schemes associated with isolated chromosomes within metaphase spreads. In normal human metaphase spreads, there are 46 chromosomes occurring in homologous pairs for the autosomal classes 1-22 and the X chromosome for females. Many genetic abnormalities are directly linked to structural and/or numerical aberrations of chromosomes within metaphase spreads. Cells with the Philadelphia chromosome contain an abnormal chromosome for class 9 and for class 22, leaving a single normal chromosome for each class. A data-driven homologue matching technique is applied to recognizing normal chromosomes from classes 9 and 22. Homologue matching integrates neural networks, dynamic programming and the Choquet integral for chromosome recognition. The inability to locate matching homologous pairs for classes 9 and 22 provides an indication that the cell is abnormal, potentially containing the Philadelphia chromosome. Applying this technique to 50 normal and to 48 abnormal cells containing the Philadelphia chromosome yields 100.0% correct abnormal cell detection with a 24.0% false positive rate.