End-to-end deep learning model with multi-channel and attention mechanisms for multi-class diagnosis in CT-T staging of advanced gastric cancer

IF 3.3 3区医学 Q1 RADIOLOGY, NUCLEAR MEDICINE & MEDICAL IMAGING

European Journal of Radiology Pub Date : 2025-09-03 DOI:10.1016/j.ejrad.2025.112408

Bowen Liu , Pengcheng Jiang , Zehui Wang , Xiaoxiao Wang , Zhixuan Wang , Chen Peng , Zhanpeng Liu , Chao Lu , Donggang Pan , Xiuhong Shan

{"title":"End-to-end deep learning model with multi-channel and attention mechanisms for multi-class diagnosis in CT-T staging of advanced gastric cancer","authors":"Bowen Liu , Pengcheng Jiang , Zehui Wang , Xiaoxiao Wang , Zhixuan Wang , Chen Peng , Zhanpeng Liu , Chao Lu , Donggang Pan , Xiuhong Shan","doi":"10.1016/j.ejrad.2025.112408","DOIUrl":null,"url":null,"abstract":"<div><h3>Background</h3><div>Homogeneous AI assessment is required for CT-T staging of gastric cancer.</div></div><div><h3>Purpose</h3><div>To construct an End-to-End CT-based Deep Learning (DL) model for tumor T-staging in advanced gastric cancer.</div></div><div><h3>Materials and Methods</h3><div>A retrospective study was conducted on 460 cases of presurgical CT patients with advanced gastric cancer between 2011 and 2024. A Three-dimensional (3D)-Convolution (Conv)-UNet based automatic segmentation model was employed to segment tumors, and a SmallFocusNet-based ternary classification model was built for CT-T staging. Finally, these models were integrated to create an end-to-end DL model. The segmentation model’s performance was assessed using the Dice similarity coefficient (DSC), Intersection over Union (IoU) and 95 % Hausdorff Distance (HD_95), while the classification model’s performance was measured with thearea under the Receiver Operating Characteristic curve (AUC), sensitivity, specificity, and F1-score.Eventually, the end-to-end DL model was compared with the radiologist using the McNemar test.</div></div><div><h3>Results</h3><div>The data were divided into <strong><em>Dataset 1</em></strong>(423 cases for training and test set, mean age, 65.0 years ± 9.46 [SD]) and <strong><em>Dataset 2</em></strong>(37 cases for independent validation set, mean age, 68.8 years ± 9.28 [SD]). For segmentation task, the model achieved a DSC of 0.860 ± 0.065, an IoU of 0.760 ± 0.096 in test set of <strong><em>Dataset 1</em></strong>, and a DSC of 0.870 ± 0.164, an IoU of 0.793 ± 0.168 in <strong><em>Dataset 2</em></strong>. For classification task,the model demonstrated a macro-average AUC of 0.882(95 % <em>CI</em> 0.812–0.926), an average sensitivity of 76.9 % (95 % <em>CI</em> 67.6 %–85.3 %) in test set of <strong><em>Dataset 1</em></strong> and a macro-average AUC of 0.862(95 % <em>CI</em> 0.723–0.942), an average sensitivity of 76.3 % (95 % <em>CI</em> 59.8 %–90.0 %) in <strong><em>Dataset 2</em></strong>. Meanwhile, the DL model’s performance was better than that of radiologist (Accuracy was 91.9 %vs82.1 %, <em>P</em> = 0.007).</div></div><div><h3>Conclusion</h3><div>The end-to-end DL model for CT-T staging is highly accurate and consistent in pre-treatment staging of advanced gastric cancer.</div></div>","PeriodicalId":12063,"journal":{"name":"European Journal of Radiology","volume":"192 ","pages":"Article 112408"},"PeriodicalIF":3.3000,"publicationDate":"2025-09-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"European Journal of Radiology","FirstCategoryId":"3","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0720048X25004942","RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"RADIOLOGY, NUCLEAR MEDICINE & MEDICAL IMAGING","Score":null,"Total":0}

引用次数: 0

Abstract

Background

Homogeneous AI assessment is required for CT-T staging of gastric cancer.

Purpose

To construct an End-to-End CT-based Deep Learning (DL) model for tumor T-staging in advanced gastric cancer.

Materials and Methods

A retrospective study was conducted on 460 cases of presurgical CT patients with advanced gastric cancer between 2011 and 2024. A Three-dimensional (3D)-Convolution (Conv)-UNet based automatic segmentation model was employed to segment tumors, and a SmallFocusNet-based ternary classification model was built for CT-T staging. Finally, these models were integrated to create an end-to-end DL model. The segmentation model’s performance was assessed using the Dice similarity coefficient (DSC), Intersection over Union (IoU) and 95 % Hausdorff Distance (HD_95), while the classification model’s performance was measured with the area under the Receiver Operating Characteristic curve (AUC), sensitivity, specificity, and F1-score.Eventually, the end-to-end DL model was compared with the radiologist using the McNemar test.

Results

The data were divided into Dataset 1(423 cases for training and test set, mean age, 65.0 years ± 9.46 [SD]) and Dataset 2(37 cases for independent validation set, mean age, 68.8 years ± 9.28 [SD]). For segmentation task, the model achieved a DSC of 0.860 ± 0.065, an IoU of 0.760 ± 0.096 in test set of Dataset 1, and a DSC of 0.870 ± 0.164, an IoU of 0.793 ± 0.168 in Dataset 2. For classification task,the model demonstrated a macro-average AUC of 0.882(95 % CI 0.812–0.926), an average sensitivity of 76.9 % (95 % CI 67.6 %–85.3 %) in test set of Dataset 1 and a macro-average AUC of 0.862(95 % CI 0.723–0.942), an average sensitivity of 76.3 % (95 % CI 59.8 %–90.0 %) in Dataset 2. Meanwhile, the DL model’s performance was better than that of radiologist (Accuracy was 91.9 %vs82.1 %, P = 0.007).

Conclusion

The end-to-end DL model for CT-T staging is highly accurate and consistent in pre-treatment staging of advanced gastric cancer.

查看原文本刊更多论文

基于多通道关注机制的端到端深度学习模型在晚期胃癌CT-T分期中的多分类诊断

背景胃癌CT-T分期需要均匀AI评价。目的构建基于端到端ct的晚期胃癌肿瘤t分期深度学习（DL）模型。材料与方法对2011 ~ 2024年460例晚期胃癌术前CT患者进行回顾性研究。采用基于三维(3D)-卷积(Conv)-UNet的自动分割模型对肿瘤进行分割，建立基于smallfocusnet的CT-T分期三元分类模型。最后，将这些模型集成在一起，创建端到端的深度学习模型。采用Dice相似系数（DSC）、Intersection over Union （IoU）和95% Hausdorff Distance （HD_95）评价分割模型的性能，采用Receiver Operating Characteristic curve （AUC）下面积、灵敏度、特异性和f1评分来衡量分类模型的性能。最后，端到端深度学习模型与放射科医生使用McNemar测试进行比较。结果数据分为数据集1（训练和测试集423例，平均年龄65.0岁±9.46 [SD]）和数据集2（独立验证集37例，平均年龄68.8岁±9.28 [SD]）。对于分割任务，该模型在数据集1的DSC值为0.860±0.065，IoU值为0.760±0.096；在数据集2的DSC值为0.870±0.164，IoU值为0.793±0.168。对于分类任务，该模型在数据集1的测试集上的宏观平均AUC为0.882(95% CI 0.812-0.926)，平均灵敏度为76.9% (95% CI 67.6% - 85.3%)；在数据集2的测试集上的宏观平均AUC为0.862(95% CI 0.723-0.942)，平均灵敏度为76.3% （95% CI 59.8% - 90.0%）。同时，DL模型的表现优于放射科医生（准确率为91.9% vs82.1%, P = 0.007）。结论端到端DL模型对晚期胃癌CT-T分期具有较高的准确性和一致性。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

European Journal of Radiology 医学-核医学

CiteScore

6.70

自引率

3.00%

发文量

398

审稿时长

42 days

期刊介绍： European Journal of Radiology is an international journal which aims to communicate to its readers, state-of-the-art information on imaging developments in the form of high quality original research articles and timely reviews on current developments in the field. Its audience includes clinicians at all levels of training including radiology trainees, newly qualified imaging specialists and the experienced radiologist. Its aim is to inform efficient, appropriate and evidence-based imaging practice to the benefit of patients worldwide.