ORIGINAL RESEARCH article
Front. Oncol.
Sec. Gastrointestinal Cancers: Gastric and Esophageal Cancers
Volume 15 - 2025 | doi: 10.3389/fonc.2025.1657159
FC-YOLO: A Fast Inference Backbone and Lightweight Attention Mechanism-Enhanced YOLO for Detecting Gastric Adenocarcinoma in Pathological Image
Provisionally accepted
- 1Baotou Medical College, Baotou, China
- 2Key Laboratory of Human Anatomy at Universities of Inner Mongolia Autonomous Region, Inner Mongolia, Baotou Medical College, Baotou, China
- 3School of Medical Technology and Anesthesia, Baotou Medical College, Baotou, 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
Gastric adenocarcinoma (GAC) is a major contributor to cancer-related deaths, yet its histopathological diagnosis is often hampered by complex images and a shortage of pathologists. While deep learning models offer potential solutions, many are computationally intensive and struggle with the fine-grained feature extraction required for GAC detection. This study addresses these challenges by proposing FC-YOLO, an optimized object detection framework for analyzing GAC histopathological images. Built upon the YOLOv11s architecture, FC-YOLO integrates three key components: (1) a FasterNet backbone for efficient multi-scale feature extraction; (2) a lightweight Mixed Local-Channel Attention (MLCA) mechanism for enhanced feature recalibration; and (3) Content-Aware ReAssembly of FEatures (CARAFE) for improved feature upsampling. These modifications are designed to enhance the model's feature extraction capabilities at a low computational cost while maintaining high computational efficiency. On a public dataset of 1,855 images, FC-YOLO achieved a mean Average Precision (mAP) of 82.8%, a 2.6% improvement over the baseline YOLOv11s, with an inference speed of 131.56 FPS. Further evaluation on an independent clinical dataset of 2,500 clinical H&E-stained slides yielded an mAP of 85.7%, indicating promising generalization capabilities. The results suggest that FC-YOLO's lightweight and efficient design presents a potential tool to assist pathologists by improving diagnostic accuracy and efficiency in resource-limited settings.
Keywords: gastric cancer, Pathological images, prediction, deep learning, target detection
Received: 01 Jul 2025; Accepted: 08 Sep 2025.
Copyright: © 2025 Zhang, jia, Wang, gao, zhang, Yi and Yan. 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:
xin xin Wang, Key Laboratory of Human Anatomy at Universities of Inner Mongolia Autonomous Region, Inner Mongolia, Baotou Medical College, Baotou, China
xun fei gao, School of Medical Technology and Anesthesia, Baotou Medical College, Baotou, China
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.