Advancing Deep Learning Frameworks for Enhanced Cervical Cancer Diagnosis, Staging and Treatment

Cervical cancer detection and diagnosis are undergoing a transformation with the integration of advanced deep learning (DL) technologies. Despite significant progress, challenges remain in achieving high diagnostic accuracy, interpretability, data privacy, and workflow integration. This Research Topic aims to showcase innovative research that leverages federated learning, generative AI, self-supervised learning, explainable AI, computer vision (including augmented reality), multimodal fusion, and benchmarking to advance the detection and diagnosis of cervical cancer.We invite submissions that address the following focus areas:• Federated Learning for Privacy-Preserving Detection. We welcome papers that develop and evaluate federated learning frameworks that enable collaborative model training across multiple institutions without sharing sensitive patient data. This approach addresses privacy concerns and enhances the robustness and generalizability of DL models for cervical cancer detection and diagnosis.• Generative AI and Self-Supervised Learning. Manuscripts that employ generative adversarial networks (GANs) and diffusion models to synthesize rare-event data, augment datasets, and improve model performance in class-imbalanced scenarios are especially welcome. Papers that utilize self-supervised and contrastive learning to exploit unlabeled cervical cytology and histopathology images, reducing annotation costs and improving feature extraction, are also in scope of this collection.• Explainable AI (XAI) for Clinical Trust. Papers that integrate explainable AI techniques, such as attention maps and feature attribution, to provide transparency and interpretability in model predictions, fostering clinician trust and supporting clinical decision-making will be considered.• Computer Vision and Augmented Reality (AR) in Diagnosis and Surgery. Papers that advance computer vision algorithms for automated image analysis, segmentation, and classification of cervical lesions will be considered. Furthermore, papers that explore the use of AR for real-time surgical guidance, preoperative planning, and enhanced visualization of anatomical structures during procedures, as demonstrated in recent studies on brachytherapy and diagnostic workflows are also within scope of this collection. AR can overlay diagnostic insights directly onto the surgical field, improving precision and safety.• Multimodal Fusion and Benchmarking. Manuscripts that design and benchmark models that fuse imaging, clinical, and molecular data for comprehensive risk assessment and lesion classification are within scope of this collection. We welomce papers that aim to publish and evaluate new, annotated datasets to establish benchmarks and facilitate reproducibility.• Comparative Studies and Novel Method. Papers that conduct comparative analyses of state-of-the-art architectures (e.g., CNNs, Vision Transformers, GANs, diffusion models) and propose innovative solutions to persistent challenges in cervical cancer detection, such as domain adaptation and rare lesion identification, will also be considered for this collection.Topic Editors, Maroua Slouma and Pijush Kanti Dutta Pramanik, have declared they have no competing interests with relation to this topic theme.Please note: Manuscripts consisting solely of bioinformatics, computational analysis, or predictions of public databases which are not accompanied by validation (independent clinical or patient cohort, or biological validation in vitro or in vivo, which are not based on public databases) are not suitable for publication in this journal. Topic editors: Maroua SLOUMA, Pijush Kanti Dutta Pramanik, Hedia Bellali, Mohamed HAMROUN, Hedi Sakli Source: https://www.frontiersin.org/research-topics/74905/advancing-deep-learning-frameworks-for-enhanced-cervical-cancer-diagnosis-staging-and-treatment