An improved RIME optimization algorithm for lung cancer image segmentation

• Published in: Computers in biology and medicine Vol. 174; p. 108219
• Main Authors: Guo, Lei, Liu, Lei, Zhao, Zhiguang, Xia, Xiaodong
• Format: Journal Article
• Language: English
• Published: United States Elsevier Ltd 01.05.2024; Elsevier Limited
• Subjects: 2D Kapur's entropy; Accuracy; Algorithms; Convergence; COVID-19; Deep learning; Design; Diagnosis; Efficiency; Entropy; Experiments; Heuristic; Humans; Image Interpretation, Computer-Assisted - methods; Image processing; Image Processing, Computer-Assisted - methods; Image segmentation; Internal Medicine; Lung cancer; Lung Neoplasms - diagnostic imaging; Lung Neoplasms - pathology; Medical diagnosis; Medical imaging; Medical research; Meta-heuristic; Multilevel image segmentation; Neural networks; Optimization algorithms; Other; RIME optimization algorithm; Swarm intelligence; Tumors; RIME optimization algorithm; Meta-heuristic; Multilevel image segmentation; 2D Kapur's entropy
• ISSN: 0010-4825; 1879-0534; 1879-0534
• DOI: 10.1016/j.compbiomed.2024.108219

Lung cancer is a prevalent form of cancer worldwide, necessitating early and accurate diagnosis for successful treatment. Within medical imaging processing, image segmentation plays a vital role in medical diagnosis. This study applies swarm intelligence algorithms to segment lung cancer pathological images at three levels. The original algorithm incorporates the Whales' search prey mechanism and a random mutation strategy, resulting in an improved version named WDRIME, which aims to enhance convergence speed and avoid local optima (LO). Additionally, the study introduces a multilevel image segmentation method for lung cancer based on the improved algorithm. WDRIME's performance is showcased by comparing it to the state-of-the-art algorithms in IEEE CEC2014. To design a framework for lung cancer image segmentation, this paper combines the WDRIME algorithm with the multilevel segmentation method. Evaluation of the segmentation results employs metrics such as PSNR, SSIM, and FSIM. Overall, the analysis confirms that the proposed algorithm supersedes others regarding convergence speed and accuracy. This model signifies a high-quality segmentation method and offers practical support for in-depth exploration of lung cancer pathological images. •WDRIME (Whale search prey mechanism and Random mutation strategy-based RIME) is proposed.•WDRIME achieves a great improvement in solution quality and convergence problems.•The performance of WDRIME is verified by comparing it with other excellent algorithms.•WDRIME is applied to multilevel lung cancer image segmentation based on 2D histogram.