深度学习在宫颈癌诊疗中的研究进展

Research Progress of Deep Learning in the Diagnosis and Treatment of Cervical Cancer

摘要: 宫颈癌是全球女性健康的主要威胁之一，其早期诊断和精准治疗是改善预后的关键。近年来，深度学习技术凭借其强大的特征提取与模式识别能力在宫颈癌诊疗的多个关键环节展现出重要价值。本文综述了深度学习在宫颈癌筛查、病理诊断、影像分析、治疗规划及预后预测等方面的研究进展，重点分析了卷积神经网络、Transformer及其混合模型在细胞学、组织病理学及多模态医学影像中的关键应用，涵盖了从弱监督学习到多模态融合等多种技术策略。同时，本文也探讨了当前模型面临的数据异质性、临床可解释性及前瞻性验证不足等核心挑战，并对未来发展方向进行了展望，通过算法优化与临床需求的深度融合，深度学习正持续推动宫颈癌诊疗向自动化、个性化和精准化迈进。

Abstract: Cervical cancer is one of the major threats to women’s health worldwide, and its early diagnosis and precise treatment are key to improving prognosis. In recent years, deep learning technology has demonstrated significant value in multiple key aspects of cervical cancer diagnosis and treatment due to its powerful feature extraction and pattern recognition capabilities. This article reviews the research progress of deep learning in cervical cancer screening, pathological diagnosis, image analysis, treatment planning, and prognosis prediction, with a focus on analyzing the key applications of Convolutional Neural Networks (CNN), Transformer, and their hybrid models in cytology, histopathology, and multimodal medical imaging, covering various technical strategies from weakly supervised learning to multimodal fusion. At the same time, this article also discusses the core challenges currently faced by models, such as data heterogeneity, clinical interpretability, and insufficient prospective validation, and looks forward to future development directions. Through the deep integration of algorithm optimization and clinical needs, deep learning is continuously advancing the diagnosis and treatment of cervical cancer towards automation, individualization, and precision.