Geometry-aware Gaussian Prior and Axial Attention for Cervical Cytology Image Classification

2026-07-11

AI Summary

Cervical cytology image classification remains highly challenging due to complex cellular morphologies, subtle intra-class variations, and high inter-class similarities, which hinder existing methods from effectively modeling both local textures and long-range structural dependencies. This work proposes a novel classification framework that integrates geometric priors with axial attention: leveraging a pre-trained vision-language model for semantic feature extraction, it employs a Mixture-of-Gaussians module to generate axial structural priors and incorporates them into an axial self-attention mechanism. This design enhances the modeling of diagnostically critical structures such as spatial cell organization and nuclear arrangements. By innovatively embedding geometry-aware Gaussian priors into the attention mechanism, the approach enables structure-guided modeling of long-range dependencies, achieving state-of-the-art accuracy of 99.48% on Mendeley and 96.08% on SIPaKMeD, while significantly improving both performance and interpretability.

Abstract

Accurate cervical cytology image classification is a key component of automated cervical cancer screening, where reliable recognition of normal, precancerous, and cancer-associated cellular patterns from Pap smear images can improve screening efficiency and diagnostic consistency. However, this task remains challenging because cervical cells exhibit complex morphology, subtle intra-class variations, and strong inter-class similarities. Existing convolution-based models capture local texture well but have limited ability to model long-range relationships, whereas attention-based models provide broader context but often lack explicit structural guidance. To address these limitations, we propose a geometry-aware classification framework for cervical cancer screening-oriented cytology image analysis, incorporating semantic abstraction and structural priors learned from pre-trained vision-language features. The method uses Gaussian expert modules to generate axis-wise priors from global semantic information, capturing structural regularities such as nuclear alignment and cellular spatial organization. These priors are embedded into an axial self-attention module to modulate similarity computation along horizontal and vertical directions, improving long-range dependency modeling and structure-sensitive feature interaction. Experiments on the Mendeley liquid-based cytology and SIPaKMeD datasets show that the proposed method achieves 99.48% accuracy on the former and 96.08% on the latter, with balanced gains in recall, precision, and overall classification performance. Visual analysis further shows that the learned priors highlight diagnostically relevant cellular regions, demonstrating the potential of the proposed framework as a screening-oriented decision-support tool for cervical cytology.

Problem

• cervical cytology
• image classification
• morphological complexity
• inter-class similarity
• long-range dependency

Innovation

• geometry-aware prior
• axial attention
• Gaussian expert module
• cervical cytology classification
• structure-sensitive feature interaction