Teng Kehan, Ren Lihua, Yan Xiaoyu, Duan Yawei, Chen Zhe, Li Hansheng, Zhang Lihua, Cui Lei
Department of Pathology, Zhongda Hospital, Southeast University School of Medicine, Nanjing, Jiangsu, China.
Department of Gastroenterology, Zhongda Hospital, Southeast University School of Medicine, Nanjing, Jiangsu, China.
Front Med (Lausanne). 2025 Jun 11;12:1594614. doi: 10.3389/fmed.2025.1594614. eCollection 2025.
infection is considered to be a primary causative factor for gastric cancer and a common cause of chronic gastritis worldwide. Identifying infection through hematoxylin and eosin (H&E) staining is demanding and tedious for pathologists. We aimed to use artificial intelligence (AI) models to improve the accuracy and efficiency of diagnosis and to reduce the workload of pathologists.
Here, we developed three multi-instance learning (MIL) models: AB-MIL, DS-MIL, and Trans-MIL, to automatically detect infection. A total of 1,020 digitized histological whole-slide images (WSI) from 817 patients were used for training, validating and testing sets at a ratio of 3:1:1. Additionally, 100 cases (218 WSIs) were randomly selected from the test set for pathologists to identify under the microscope. The accuracy, specificity, sensitivity, false negative rate, false positive rate, and other metrics were calculated separately for the MIL models and the pathologists.
All three models demonstrated good diagnostic performance in predicting infection, with the DS-MIL classification model showing the best diagnostic performance, achieving an accuracy of 89.7% and an area under the curve (AUC) of 0.949, which is higher than the accuracy rate of senior pathologists at 81.7%. Furthermore, the model demonstrates superior performance in terms of sensitivity and specificity. The reliability of DS-MIL is confirmed through the Visual model.
Our research presents an AI - based predictive model for infection, which significantly enhances clinical efficiency and diagnostic accuracy. Currently, we are conducting multi-center validation to enhance the model's generalization capability.
感染被认为是胃癌的主要致病因素，也是全球慢性胃炎的常见病因。通过苏木精和伊红（H&E）染色来识别感染对病理学家来说既费力又繁琐。我们旨在使用人工智能（AI）模型来提高诊断的准确性和效率，并减轻病理学家的工作量。
在此，我们开发了三种多实例学习（MIL）模型：AB-MIL、DS-MIL和Trans-MIL，以自动检测感染。总共817例患者的1020张数字化组织学全切片图像（WSI）以3:1:1的比例用于训练集、验证集和测试集。此外，从测试集中随机选择100个病例（218张WSI）供病理学家在显微镜下识别感染。分别计算MIL模型和病理学家的准确率、特异性、敏感性、假阴性率、假阳性率等指标。
所有三种模型在预测感染方面均表现出良好的诊断性能，其中DS-MIL分类模型表现出最佳的诊断性能，准确率达到89.7%，曲线下面积（AUC）为0.949，高于高级病理学家81.7%的准确率。此外，该模型在敏感性和特异性方面表现优异。通过视觉模型证实了DS-MIL的可靠性。
我们的研究提出了一种基于AI的感染预测模型，显著提高了临床效率和诊断准确性。目前，我们正在进行多中心验证以增强模型的泛化能力。
