Title: Research on segmentation and recognition algorithm of squamous carcinoma cells based on M-SVM

Authors: Hu Qi; Duan Jin; Wang LiNing

Addresses: Department of Electronics and Information Engineering, Changchun University of Science and Technology, Changchun, Jilin, China; Information Engineering Department Jilin Business and Technology College, Changchun, Jilin, China ' Department of Electronics and Information Engineering, Changchun University of Science and Technology, Changchun, Jilin, China ' College of Biological and Agricultural Engineering, Jilin University, Changchun, Jilin, China

Abstract: In the biomedical field, the accurate segmentation and recognition of cells is always one of the hot spots in the cell image. The errors in the segmentation process will propagate to higher-level processing stage, which has a significant impact on the recognition rate in the future, so the accuracy of segmentation is very important. The traditional way of using the probability density function to identify cells is greatly limited. Therefore, starting with the statistical machine learning theory, this paper presents a new classification detection algorithm for squamous carcinoma cells based on the multi-support vector machine (M-SVM), which uses the support vector machine to search for optimal classification surface, and eventually extracts and classifies the squamous carcinoma cells with high accuracy. From the experimental results, the recognition accuracy of the squamous carcinoma cells has been significantly improved, thereby providing a more robust theoretical support for the subsequent clinical diagnosis.

Keywords: multiple SVM; support vector machines; M-SVM; statistical machine learning; squamous carcinoma cells; cell segmentation; cell recognition; cell images; medical imaging; clinical diagnosis; biomedical engineering.

DOI: 10.1504/IJCSM.2016.078736

International Journal of Computing Science and Mathematics, 2016 Vol.7 No.4, pp.340 - 349

Received: 30 Mar 2016
Accepted: 25 Apr 2016

Published online: 01 Sep 2016 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article