Two-Stage Cross-Domain Cervical Abnormality Screening with Cytopathological Image Synthesis and Knowledge Distillation

2026-06-25

AI Summary

This work addresses the challenge of cross-domain generalization in cervical cytopathology, where significant inter-institutional domain shifts and subtle visual differences across disease stages hinder model performance. To tackle this, the authors propose a two-stage framework: first, a Spatially-Continuous Unpaired Neural Schrödinger Bridge (SC-UNSB) is employed to synthesize an intermediate domain that mitigates domain discrepancy; second, a dual-level feature alignment strategy is integrated into knowledge distillation to simultaneously align shallow structural and deep semantic features, enabling invariant knowledge transfer from source to target domains. By innovatively combining entropy-regularized optimal transport–based image synthesis with dual-level alignment, the method effectively confronts the dual challenges of domain shift and class ambiguity, achieving substantial performance gains in cross-domain cervical abnormality screening.

Abstract

Cross-domain diagnosis remains a major challenge in cervical cell pathology due to pronounced domain shifts across institutions and the subtle visual differences among disease stages, which jointly impair model generalization. To address these issues, this paper proposes a two-stage framework for cross-domain cervical cell detection. In the first stage, we propose the Spatially-Continuous Unpaired Neural Schrödinger Bridge (SC-UNSB), which constructs a synthetic intermediate domain to mitigate cross-domain distribution shifts by modeling image translation as an entropy-regularized optimal transport process. In the second stage, we propose a dual-level feature alignment strategy within a knowledge distillation, which progressively aligns shallow structural features and deep semantic representations to facilitate the transfer of domain-invariant knowledge from the source to the target model. Experimental results demonstrate that the proposed method effectively mitigates domain shift and category ambiguity, improving the cross-domain detection performance.

Problem

• cross-domain diagnosis
• domain shift
• cervical cell pathology
• category ambiguity

Innovation

• cross-domain
• image synthesis
• knowledge distillation
• optimal transport
• feature alignment