earticle

영어

This paper proposes a modified bacterial foraging algorithm with a probabilistic derivative approach to detect edges in chromosome images. Chromosomal Edge Detection is fundamental for automatic karyotyping for noise reduction and getting useful messages from the edges. Subjected to staining and other imaging constraints, chromosomal banding patterns lack in resolution, contrast and suffer from noise. For this reason, chromosomal edge detection is highly preferred prior to the segmentation and classification of chromosomes. When the chromosomes occlude or overlap, edge detection becomes extremely difficult. Edge detection is highly challenging and this paper presents a Modified Bacterial Foraging Algorithm (MBFA) based on a probabilistic derivative methodology based on Ant Colony Optimization (ACO) for the detection of edges in chromosomes. Bacterium searches for the nutrients in the direction decided by a probabilistic derivative approach derived from ACO and the edge pixels are identified and traversed. The study reveals that MBFA gives the most promising results in detecting chromosomal edges, greatly reducing the computation time and memory requirements. Acceptable values of parameters for performance evaluation like Kappa (K) and Entropy (E) are achieved with the proposed algorithm in comparison to the other conventional methods of edge detection.