Style-Aligned Image Composition for Robust Detection of Abnormal Cells in Cytopathology

2025-06-26

AI Summary

Abnormal cell detection in cytopathology faces three key challenges: scarce annotations, severe class imbalance (long-tailed distribution), and inconsistent staining styles across laboratories. To address these, we propose a zero-shot style-aligned image synthesis method that requires no additional training. Our approach retrieves candidate abnormal cells from a curated cell repository using attribute-guided retrieval, then performs high-fidelity style transfer via high-frequency feature reconstruction. A large vision-language model (VLM) is further employed to filter synthesized images based on clinical plausibility and quality. By integrating multimodal priors—morphological attributes, spectral characteristics, and semantic knowledge—the method enables controllable, pathology-aware data augmentation. Experiments demonstrate substantial improvements in tail-class recognition accuracy and cross-staining generalization. The synthesized images achieve state-of-the-art performance across multiple clinically relevant metrics, validating the method’s practical utility and robustness in real-world diagnostic settings.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Code Generation / Program Synthesis from Natural Language

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Challenges such as the lack of high-quality annotations, long-tailed data distributions, and inconsistent staining styles pose significant obstacles to training neural networks to detect abnormal cells in cytopathology robustly. This paper proposes a style-aligned image composition (SAIC) method that composes high-fidelity and style-preserved pathological images to enhance the effectiveness and robustness of detection models. Without additional training, SAIC first selects an appropriate candidate from the abnormal cell bank based on attribute guidance. Then, it employs a high-frequency feature reconstruction to achieve a style-aligned and high-fidelity composition of abnormal cells and pathological backgrounds. Finally, it introduces a large vision-language model to filter high-quality synthesis images. Experimental results demonstrate that incorporating SAIC-synthesized images effectively enhances the performance and robustness of abnormal cell detection for tail categories and styles, thereby improving overall detection performance. The comprehensive quality evaluation further confirms the generalizability and practicality of SAIC in clinical application scenarios. Our code will be released at https://github.com/Joey-Qi/SAIC.

Problem

• Lack of high-quality annotations for abnormal cell detection
• Long-tailed data distributions in cytopathology images
• Inconsistent staining styles across pathological images

Innovation

• Style-aligned image composition for robust detection
• High-frequency feature reconstruction for fidelity
• Vision-language model filters high-quality synthesis