[1]陈超爽,曹洋森,朱晓斐,等.基于深度学习的胰腺肿瘤靶区自动分割[J].中国医学物理学杂志,2025,42(7):923-928.[doi:DOI:10.3969/j.issn.1005-202X.2025.07.012]
　CHEN Chaoshuang,CAO Yangsen,ZHU Xiaofei,et al.Automatic pancreatic cancer GTV segmentation based on deep learning[J].Chinese Journal of Medical Physics,2025,42(7):923-928.[doi:DOI:10.3969/j.issn.1005-202X.2025.07.012]
点击复制

基于深度学习的胰腺肿瘤靶区自动分割()

《中国医学物理学杂志》[ISSN:1005-202X/CN:44-1351/R]

卷:
42
期数:
2025年第7期
页码:
923-928
栏目:
医学影像物理
出版日期:
2025-07-25

文章信息/Info

Title:
Automatic pancreatic cancer GTV segmentation based on deep learning
文章编号:
1005-202X（2025）07-0923-06
作者:
陈超爽1; 曹洋森2; 朱晓斐2; 曾福斌2; 顾蕾2; 江林宫2; 张火俊2
1.中国人民解放军92493部队医院核医学肿瘤科，辽宁 葫芦岛 125000；2.中国人民解放军海军军医大学第一附属医院放射治疗科，上海 200433
Author(s):
CHEN Chaoshuang1;  CAO Yangsen2;  ZHU Xiaofei2;  ZENG Fubin2;  GU Lei2;  JIANG Lingong2;  ZHANG Huojun2
1. Department of Nuclear Medicine and Oncology, Chinese People’s Liberation Army 92493 Military Hospital, Huludao 125000, China;2. Department of Radiation Oncology, Changhai Hospital of Naval Medical University, Shanghai 200433, China
关键词:
胰腺癌; 肿瘤靶区; 分割; 深度学习; 卷积神经网络
Keywords:
pancreatic cancer;  gross tumor volume;  segmentation;  deep learning;  convolutional neural network
分类号:
R318.14
DOI:
DOI:10.3969/j.issn.1005-202X.2025.07.012
文献标志码:
A
摘要:
目的：研究基于胰腺增强CT的卷积神经网络进行胰腺肿瘤靶区（GTV）自动分割的可行性及准确性。方法：回顾性选取114例胰腺癌患者的定位增强CT影像，使用AccuContour进行GTV手动勾画后导入AccuLearning按照8:1:1的比例随机抽取数据作为训练集、验证集和测试集，分别使用Flex和Segresnet两种网络结构进行自动分割模型训练，每种网络结构固定训练参数不变连续训练3次。模型的评价指标包括戴斯相似系数（DSC）、95%豪斯多夫距离（HD95）、对称位置的平均表面距离（ASSD）、体素相对误差（RVD）。结果：模型训练阶段，Flex 组中 Flex-3 测试结果较差，最小 DSC 为0.14%，平均DSC为56.30%；Flex-1测试结果较优，最小DSC为47.90%，平均DSC为67.35%。Segresnet组中Segresnet-2测试结果较差，最小 DSC 为 0.00%，平均 DSC 为 42.46%；Segresnet-3 测试结果较优，最小 DSC 为 42.65%，平均 DSC 为63.28%。固定测试阶段，Segresnet-3组取得相对最优的DSC和RVD均值分别为63.88%和29.41%，Segresnet-2组取得相对最优的ASSD均值为4.43 mm，Segresnet-1组取得相对最优的HD95均值为12.87 mm。结论：Flex和Segresnet两种卷积神经网络结构都可用于胰腺肿瘤靶区的自动分割训练，Segresnet构建的分割模型的综合评估更优。
Abstract:
Abstract: Objective To investigate the feasibility and accuracy of convolutional neural networks for automaticallydelineating the pancreatic cancer gross target volume (GTV) in pancreatic enhanced CT. Methods The localizable enhancedCT images of 114 patients with pancreatic cancer were retrospectively selected, in which the GTV was manually delineatedusing AccuContour. The imaging data were then import to AccuLearning and randomly divided as the training set, validationset and test set at a ratio of 8:1:1. Flex and Segresnet were used to train the automatic segmentation model, with each networkstructure trained continuously 3 times using fixed training parameters. The model was evaluated in terms of Dice similaritycoefficient (DSC), 95% Hausdorff distance (HD95), average symmetric surface distance (ASSD) and relative volumedifference (RVD). Results In the model training phase, Flex-3 test results in Flex group were the worst, with a minimum DSCof 0.14% and an average DSC of 56.30%, while Flex-1 performed well, achieving a minimum DSC of 47.90% and anaverage DSC of 67.35%. Meanwhile, Segresnet-2 in Segresnet group had the worst test results, with a minimum DSC of0.00% and an average DSC of 42.46%, while Segresnet-3 test results were better, with a minimum DSC of 42.65% and anaverage DSC of 63.28%. In the fixed testing phase, the best results among all were as follows: average DSC and RVD valuesof 63.88% and 29.41% in Segresnet-3 group, average ASSD value of 4.43 mm in Segresnet-2 group, and average HD95 valueof 12.87 mm in Segresnet-1 group. Conclusion Both Flex and Segresnet architectures of convolutional neural network can beused for the automatic pancreatic tumor GTV segmentation training, with Segresnet performing better in comprehensiveevaluation.

相似文献/References:

[1]巩汉顺,王金媛,丛小虎,等.晚期胰腺癌适形、简化调强及螺旋断层放疗的剂量学评估[J].中国医学物理学杂志,2015,32(01):8.[doi:10.3969/j.issn.1005-202X.2015.01.003]
[2]郄帅,刘明,翟福山,等.局部晚期胰头癌不同调强放疗技术的剂量学分析[J].中国医学物理学杂志,2013,30(04):4257.[doi:10.3969/j.issn.1005-202X.2013.04.011]
[3]尚进,李东,闫英,等.胰腺癌螺旋断层调强放疗剂量学分析[J].中国医学物理学杂志,2013,30(06):4472.[doi:10.3969/j.issn.1005-202X.2013.06.001]
[4]朱夫海,吴伟章,王勇,等.胰腺癌患者螺旋断层放疗摆位误差分析[J].中国医学物理学杂志,2013,30(06):4480.[doi:10.3969/j.issn.1005-202X.2013.06.003]
[5]曹洋森,于春山,张火俊,等.射波刀Monte Carlo与Ray-Tracing算法在胰腺癌中的剂量学比较[J].中国医学物理学杂志,2016,33(10):1008.[doi:10.3969/j.issn.1005-202X.2016.10.007]
　[J].Chinese Journal of Medical Physics,2016,33(7):1008.[doi:10.3969/j.issn.1005-202X.2016.10.007]
[6]刘玉,刘晓娜,高维青,等.联合应用DCE-MRI和18F-FDG PET/CT 监测125I 粒子短距离放疗对胰腺癌移植瘤的疗效[J].中国医学物理学杂志,2017,34(1):1.[doi:10.3969/j.issn.1005-202X.2017.01.001]
　[J].Chinese Journal of Medical Physics,2017,34(7):1.[doi:10.3969/j.issn.1005-202X.2017.01.001]
[7]林淑莹,黄子成,郑彬彬,等.MRI与MSCTA诊断胰腺癌胰周血管侵犯的价值比较[J].中国医学物理学杂志,2020,37(9):1143.[doi:10.3969/j.issn.1005-202X.2020.09.012]
　LIN Shuying,HUANG Zicheng,ZHENG Binbin,et al.Comparison of diagnostic efficiency of MRI versus MSCTA in peripancreatic vascular invasionin pancreatic cancer[J].Chinese Journal of Medical Physics,2020,37(7):1143.[doi:10.3969/j.issn.1005-202X.2020.09.012]
[8]陈鑫龙,叶凯,周文策.人工智能在胰腺疾病新型诊疗模式中的应用及进展[J].中国医学物理学杂志,2022,39(8):1049.[doi:DOI:10.3969/j.issn.1005-202X.2022.08.023]
　CHEN Xinlong,YE Kai,et al.Keywords: artificial intelligence machine learning deep learning pancreatic diseasepancreatitis?ancreatic cancer precision medicine[J].Chinese Journal of Medical Physics,2022,39(7):1049.[doi:DOI:10.3969/j.issn.1005-202X.2022.08.023]
[9]李政晓,李昇霖,孙嘉晨,等.胰腺癌胰周血管侵犯的影像学进展[J].中国医学物理学杂志,2022,39(9):1114.[doi:DOI:10.3969/j.issn.1005-202X.2022.09.010]
　LI Zhengxiao,,et al.Advances in imaging of peripancreatic vascular invasion in pancreatic carcinoma[J].Chinese Journal of Medical Physics,2022,39(7):1114.[doi:DOI:10.3969/j.issn.1005-202X.2022.09.010]
[10]张丹凤,蒋俊,吴昊天,等.基于nnU-Net的宫颈癌近距离治疗中高危临床靶区及危及器官的自动勾画[J].中国医学物理学杂志,2023,40(12):1463.[doi:DOI:10.3969/j.issn.1005-202X.2023.12.003]
　ZHANG Danfeng,JIANG Jun,WU Haotian,et al.Auto-segmentation of high-risk clinical target volume and organs-at-risk for brachytherapy of cervical cancer based on nnUNet[J].Chinese Journal of Medical Physics,2023,40(7):1463.[doi:DOI:10.3969/j.issn.1005-202X.2023.12.003]

备注/Memo

备注/Memo:
【收稿日期】2025-02-12【基金项目】海军军医大学第一附属医院基础医学研究专项（2023PY22）【作者简介】陈超爽，主治医师，研究方向：恶性肿瘤的综合治疗以及放射治疗物理技术应用，E-mail: 258448423@qq.com【通信作者】曹洋森，硕士，物理师，副主任医师，研究方向：肿瘤放射物理，E-mail: caoyangsen@163.com
更新日期/Last Update: 2025-07-25