Skip to content
Accessible Unlicensed Requires Authentication Published by De Gruyter 2020

14. Multilevel thresholding using crow search optimization for medical images

S. N. Kumar, A. Lenin Fred, Ajay H. Kumar, Parasuraman Padmanabhan and Balazs Gulyas

Abstract

Segmentation is the process of delineation of the desired region of interest and in the case of medical images; the region of interest are anatomical organs, tumor or cyst. The multilevel thresholding gains much importance for the images with complex objects and the role of the optimization algorithm is vital in the selection of threshold values. Thresholding is a classical segmentation algorithm and for the estimation of threshold values, Otsu’s or Kapur’s technique is used. This research work employs various optimization techniques like electromagnetic optimization, harmony search optimization and crow search optimization for the optimum selection of threshold values. The electromagnetism like optimization algorithm (EMO) is based on the electromagnetism laws of physics. The harmony search algorithm key concept is that, when musicians compose the harmony, they usually try various possible combinations of the music pitches stored in the memory, based on this concept, this algorithm was evolved. The bio-inspired optimization techniques are gaining prominence in many applications and the crow search optimization is based on the biological traits of crows. The multilevel thresholding, when coupled with crow search optimization, was found to yield efficient results. The less number of parameter tuning and less complexity makes crow search optimization, an efficient one for solving the real-world problems. The algorithms are developed in Matlab 2010a tested on real-time CT abdomen DICOM data sets. The performance metrics evaluation reveals the efficiency of crow search optimization in the multilevel thresholding segmentation approach.

© 2020 Walter de Gruyter GmbH, Berlin/Munich/Boston