The segmentation of retinal images into multiple meaningful regions is a critical step in medical image interpretation, directly impacting diagnostic accuracy and clinical decision support. Conventional thresholding techniques, while foundational, often suffer from limitations in handling complex pixel distributions, particularly in high-dimensional search spaces where the risk of premature convergence and local stagnation increases with the number of thresholds. In response to these challenges, this work implements the Quantum-behaved Particle Swarm Optimization (QPSO) algorithm tailored for multilevel threshold-based segmentation. Distinct from classical PSO, QPSO integrates probabilistic modeling rooted in quantum theory, enabling a superior balance between global exploration and local exploitation. The segmentation framework is guided by Kapur’s entropy, which serves as a fitness criterion to optimize the selection of threshold values by maximizing inter-region information content. The QPSO is systematically applied to publicly available retinal image datasets, and its performance is benchmarked against a suite of seven contemporary metaheuristic algorithms. Quantitative evaluation is conducted using standard metrics, including PSNR, SSIM, FSIM, and MSE, while statistical robustness is established through non-parametric Wilcoxon signed-rank and Friedman tests. Experimental results consistently indicate the superiority of QPSO, achieving an SSIM of 0.88, FSIM of 0.89, and PSNR of 24.77, underscoring its potential as an effective and computationally efficient solution for complex medical image segmentation tasks.