Extraction of Leukocyte Section from Digital Microscopy Picture with Image Processing Method

摘要

Appraisal of leukocyte sections is essential to predict the disease in humans. The increase in the leukocyte indirectly represents the disease in humans. This work proposes a methodology to extract the various leukocyte segments from the Microscopic Blood Smear (MBS) image with better accuracy. This work implemented an entropy assisted thresholding and morphology based segmentation to mine the segment. The threshold is executed with the Kapur's function with a Social Group Optimization (SGO). The mining of leukocyte is done with morphological extraction. The mined segment is then compared against the ground-truth (GT) and the performance metrics are then computed. Further, a confusion matrix is also constructed to confirm the performance of the proposed technique. In this analysis, the benchmark images of Leukocyte Images for Segmentation and Classification (LISC) is considered and the attained outcomes are authenticated with existing similar results in literature. The experimental outcome confirms that, proposed technique is efficient in extracting all the five leukocytes from the LISC catalog with improved image quality parameters. This confirms that, proposed system can be used to examine the clinical rank pictures in future.