New perspective in enhancing Papanicolaou-smear image using CLAHE and spider monkey optimization

Abstract

High-quality Papanicolaou (Pap) smear images are essential for reliable early detection of cervical cancer, yet low contrast and noise often hinder accurate interpretation. This study introduces spider monkey optimization (SMO)-contrast-limited adaptive histogram equalization (CLAHE), an optimized CLAHE framework guided by the SMO algorithm. A novel signal contrast (SC) objective function is proposed, combining perceptual enhancement contrast enhancement-based image quality (CEIQ) with fidelity preservation peak signal-to-noise ratio (PSNR) to adaptively tune CLAHE parameters. Experiments on the publicly available SIPaKMeD and Mendeley LBC datasets demonstrate that SMO-CLAHE consistently outperforms manual settings and flower pollination algorithm (FPA)-based optimization, and achieves performance comparable to pelican optimization algorithm (POA) across key quality metrics including entropy, structural similarity index (SSIM), PSNR, enhancement measure estimation (EME), root mean square contrast (RMSC), standard deviation (STD-DEV), and CEIQ. Furthermore, downstream evaluation using a MobileNetV3-S classifier shows that the enhanced images lead to improved cervical cancer classification performance. These results highlight SMO-CLAHE as a robust and clinically relevant preprocessing framework, offering a new perspective for Pap smear image enhancement and diagnostic support.