[1]陈爱德,胡亚恒,魏文存.基于深度学习的不同CT阈值下肺腺癌实性成分浸润性预测[J].中国医学物理学杂志,2026,43(1):60-64.[doi:DOI:10.3969/j.issn.1005-202X.2026.01.008]
　CHEN Aide,HU Yaheng,WEI Wencun.Deep learning based prediction of solid component infiltration in lung adenocarcinoma under different CT thresholds[J].Chinese Journal of Medical Physics,2026,43(1):60-64.[doi:DOI:10.3969/j.issn.1005-202X.2026.01.008]
点击复制

基于深度学习的不同CT阈值下肺腺癌实性成分浸润性预测()

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

卷:
43卷
期数:
2026年第1期
页码:
60-64
栏目:
医学影像物理
出版日期:
2026-01-26

文章信息/Info

Title:
Deep learning based prediction of solid component infiltration in lung adenocarcinoma under different CT thresholds
文章编号:
1005-202X（2026）01-0060-05
作者:
陈爱德; 胡亚恒; 魏文存
安康市中医医院设备科， 陕西 安康 725000
Author(s):
CHEN Aide;  HU Yaheng;  WEI Wencun
Department of Equipment, Ankang Hospital of Traditional Chinese Medicine, Ankang 725000, China
关键词:
肺腺癌; 实性成分; 肺结节; 深度学习; CT阈值
Keywords:
Keywords: lung adenocarcinoma solid component lung nodule deep learning CT threshold
分类号:
R318；R734.2
DOI:
DOI:10.3969/j.issn.1005-202X.2026.01.008
文献标志码:
A
摘要:
目的：探讨不同CT阈值对3种深度学习模型在肺结节分类任务中的影响，以期寻找能够提高模型分类性能的最佳阈值，从而为肺结节的精准诊断提供参考。方法：收集某三甲医院113例肺腺癌（176个肺结节）患者的CT影像数据，选取5个不同的CT阈值，分别基于卷积神经网络、残差网络和卷积长短期记忆网络3种模型对肺结节进行分类，并计算各模型在不同阈值下的AUC-ROC值、准确率、精确率和召回率。结果：随着CT阈值逐渐升高，3种模型的分类性能总体上呈现出上升趋势，特别是在-200 HU阈值下，所有模型的AUC-ROC值均达到最高点，卷积神经网络、残差网络、卷积长短期记忆网络分别为0.855、0.870和0.860。然而，阈值进一步提高至-100 HU后，模型的性能有所下降。结论：CT阈值的选择对深度学习模型在肺结节分类中的表现具有显著影响，-200 HU可能是最佳阈值。在该阈值下，模型的分类性能达到最优。未来的计算机辅助诊断系统应考虑在该阈值附近进行CT图像预处理，以提高肺结节的诊断准确率。
Abstract:
Abstract: Objective To explore the effects of different CT thresholds on the performance of 3 deep learning models in lung nodule classification tasks for identifying the optimal threshold that enhances model classification performance and thereby providing a reference for the accurate diagnosis of lung nodules. Methods CT imaging data were collected from 113 patients with lung adenocarcinoma (involving 176 lung nodules) in a Grade A tertiary hospital. Five different CT thresholds were selected, and the lung nodules were classified using convolutional neural network, residual network, and convolutional long short-term memory network. The area under the receiver operating characteristic curve (AUC-ROC), accuracy, precision, and recall of each model under different thresholds were calculated. Results The classification performance of all 3 models generally improved as the CT threshold gradually increased. Especially at the threshold of -200 HU, the AUC-ROC values of all models reached their peaks, which were 0.855 for convolutional neural network, 0.870 for residual network, and 0.860 for convolutional long short-term memory network. However, when the threshold was further increased to -100 HU, the model performance declined. Conclusion The selection of CT thresholds exerts significant effects on the performance of deep learning models in lung nodule classification. The threshold of -200 HU may be the optimal choice as it yields the best model performance. Future computer-aided diagnostic systems should consider performing CT image preprocessing near this threshold to improve the diagnostic accuracy of lung nodules.

相似文献/References:

[1]鲁晓腾,龚敬,聂生东. 基于CT图像特征的肺腺癌预后因素分析[J].中国医学物理学杂志,2019,36(3):291.[doi:DOI:10.3969/j.issn.1005-202X.2019.03.009]
　LU Xiaoteng,GONG Jing,NIE Shengdong. CT image feature-based analysis on the prognostic factors of lung adenocarcinoma[J].Chinese Journal of Medical Physics,2019,36(1):291.[doi:DOI:10.3969/j.issn.1005-202X.2019.03.009]
[2]邓靓娜,张国晋,张斌,等.周围型肺腺癌和周围型肺鳞癌的能谱CT表现[J].中国医学物理学杂志,2021,38(9):1097.[doi:10.3969/j.issn.1005-202X.2021.09.009]
　DENG Liangna,,et al.Spectral CT findings of peripheral lung adenocarcinoma and peripheral lung squamous cellcarcinoma[J].Chinese Journal of Medical Physics,2021,38(1):1097.[doi:10.3969/j.issn.1005-202X.2021.09.009]
[3]刘立威,林俊豪,陈逢生.ARHGAP10基因在肺腺癌中的表达及其机制[J].中国医学物理学杂志,2023,40(5):653.[doi:1005-202X（2023）05-0653-08]
　LIU Liwei,LIN Junhao,CHEN Fengsheng.Expression of ARHGAP10 in lung adenocarcinoma and relevant mechanisms[J].Chinese Journal of Medical Physics,2023,40(1):653.[doi:1005-202X（2023）05-0653-08]
[4]董雨晴,刘浩然,孙继宏,等.基于生物信息学构建铜代谢相关基因肺腺癌预后模型及免疫分析[J].中国医学物理学杂志,2024,41(10):1296.[doi:DOI:10.3969/j.issn.1005-202X.2024.10.015]
　DONG Yuqing,LIU Haoran,SUN Jihong,et al.Prognostic model and immune analysis of copper metabolism related genes in lung adenocarcinoma based on bioinformatics[J].Chinese Journal of Medical Physics,2024,41(1):1296.[doi:DOI:10.3969/j.issn.1005-202X.2024.10.015]

备注/Memo

备注/Memo:
【收稿日期】2025-07-21 【基金项目】陕西省自然科学基金基础研究计划（2023-JC-QN-0849） 【作者简介】陈爱德，副主任技师，研究方向：临床生物医学工程，医疗设备维修，E-mail: C55319hen@163.com 【通信作者】魏文存，副主任技师，研究方向：临床生物医学工程，医疗设备维修，E-mail: 475424146@qq.com
更新日期/Last Update: 2026-01-27