ORIGINAL RESEARCH article
Front. Med.
Sec. Precision Medicine
This article is part of the Research TopicArtificial Intelligence-Assisted Radiotherapy for Pelvic and Abdominal MalignanciesView all 7 articles
A Transformer-Based Framework for Cervical Cancer Lesion Detection Toward Radiogenomic Integration
Provisionally accepted
- 1Affiliated Hospital of Nanjing University of Chinese Medicine, Nanjing, Jiangsu, Nanjing, China
- 2The Chinese Clinical Medicine Innovation Center of Obstetrics, Gynecology, and Reproduction in Jiangsu Province, Nanjing, Nanjing, China
- 3Suzhou TCM Hospital Affiliated to Nanjing University of Chinese Medicine, Suzhou, Suzhou, China
- 4Affiliated Hospital of Nanjing University of Chinese Medicine, Nanjing,, Nanjing, China
- 5The Chinese Clinical Medicine Innovation Center of Obstetrics, Gynecology, and Reproduction in Jiangsu Province, Nanjing, China
- 6Suzhou TCM Hospital Affiliated to Nanjing University of Chinese Medicine, Suzhou, China
Select one of your emails
You have multiple emails registered with Frontiers:
Notify me on publication
Please enter your email address:
If you already have an account, please login
You don't have a Frontiers account ? You can register here
Cervical cancer remains one of the leading causes of cancer-related deaths in women worldwide. Early detection and accurate lesion localization from cervical cytology images are crucial for improving diagnostic outcomes. In this work, we propose CerviFormer, a transformer-based deep learning framework designed for automated analysis of cervical cytology images. Unlike traditional convolutional neural networks, our model captures long-range dependencies through self-attention mechanisms, allowing more accurate detection of morphological abnormalities. We evaluate CerviFormer on a benchmark cytology image dataset and compare it against state-of-the-art object detection models, including Faster R-CNN and YOLO variants. Our results demonstrate that CerviFormer achieves superior performance in terms of mean Average Precision (mAP) and lesion localization accuracy. Furthermore, attention heatmaps are used to enhance model interpretability, providing insights into the decision-making process and improving clinical trust. While this study focuses on cytology image-based lesion detection, future extensions will explore integrating radiogenomic data for comprehensive cervical cancer progression modeling.
Keywords: Cervical Cancer Prediction, Curriculum Optimization, multimodal learning, Radiogenomic Integration, spatial attention, transformer architecture
Received: 25 Jul 2025; Accepted: 09 Feb 2026.
Copyright: © 2026 Sun, Ge, Deng, Wang, Wu, Wang and Ren. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
* Correspondence: Qingling Ren
Disclaimer: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.