Medical Image Thresholding Using Modified Shuffled Frog Leaping Algorithm (SFLA)

This research explores medical image thresholding using a modified Shuffled Frog Leaping Algorithm (SFLA): In medical image processing, thresholding serves as a fundamental technique for segmenting images into distinct regions. This study focuses on enhancing medical image thresholding effectiveness through an improved SFLA approach. The SFLA algorithm mimics natural frog population behavior by simulating their foraging strategies to search for optimal threshold values. The modified SFLA implementation incorporates enhanced population initialization and leap-size adaptation mechanisms to improve convergence speed. Key algorithmic components include: - Fitness function evaluation using Otsu's inter-class variance maximization - Dynamic subgroup partitioning for local search optimization - Global information exchange through shuffled regrouping operations Our experimental framework employs diverse medical image datasets (including MRI, CT, and ultrasound images) to evaluate the modified SFLA's performance in threshold-based segmentation. The algorithm's effectiveness is measured through quantitative metrics including Dice coefficient, Jaccard index, and computational time analysis. This research aims to contribute more efficient thresholding methodologies to medical image processing, ultimately enhancing segmentation quality and diagnostic accuracy. The proposed approach demonstrates particular advantages in handling noisy medical images and maintaining boundary precision in heterogeneous tissue regions.