CT影像组学联合临床影像学特征评估混合磨玻璃结节样肺腺癌的浸润性

doi:10.13481/j.1671-587X.20260428

吉林大学学报(医学版) ›› 2026, Vol. 52 ›› Issue (4): 1151-1160.doi: 10.13481/j.1671-587X.20260428

• 影像学 • 上一篇

CT影像组学联合临床影像学特征评估混合磨玻璃结节样肺腺癌的浸润性

史鸿宇¹,徐磊¹(),赵磊²

^1.内蒙古医科大学附属医院呼吸与危重症学科，内蒙古呼和浩特 010050
^2.内蒙古医科大学附属医院影像诊断科，内蒙古呼和浩特 010050

收稿日期:2025-10-30 接受日期:2025-12-23 出版日期:2026-07-28 发布日期:2026-07-27
通讯作者: 徐磊 E-mail:lxu81@126.com
作者简介:史鸿宇（1999—），女，内蒙古自治区呼和浩特市人，住院医师，医学硕士，主要从事肺结节和肺血管基础及临床方面的研究。
基金资助:
内蒙古自治区科技厅自然科学基金项目(2023MS08014);内蒙古医学科学院重点项目(2024GLLH0291);内蒙古自治区科技计划项目(2022YFSH0092);内蒙古自治区卫生健康科技计划项目(202201335)

Evaluation of CT radiomics combined with clinical imaging features on invasiveness of mixed ground-glass nodule-like pulmonary adenocarcinoma

Hongyu SHI¹,Lei XU¹(),Lei ZHAO²

^1.Department of Respiratory and Critical Care Medicine，Affiliated Hospital，Inner Mongolia Medical University，Hohhot 010050，China
^2.Department of Diagnostic Imaging，Affiliated Hospital，Inner Mongolia Medical University，Hohhot 010050，China

Received:2025-10-30 Accepted:2025-12-23 Online:2026-07-28 Published:2026-07-27
Contact: Lei XU E-mail:lxu81@126.com

摘要/Abstract

摘要：

目的探讨混合磨玻璃结节（mGGN）样肺腺癌浸润性的影像学危险因素，构建临床影像特征联合影像组学的机器学习模型，阐明其诊断效能。方法回顾性收集术后病理明确的mGGN样肺腺癌患者261例，符合纳入标准的mGGN共278枚，根据病理结果分为惰性组（n=73）和浸润组（n=205）。收集2组患者临床资料和影像学资料。所有患者接受薄层计算机断层扫描（CT）扫描，提取影像组学特征，通过z.score标准化联合LASSO回归对数据进行特征降维，并利用10折交叉完成筛选，确定最优组学特征，构建逻辑回归（LR）机器学习算法的影像组学模型；采用单因素分析和多因素Logistic回归分析筛选mGGN样肺腺癌患者浸润性有意义的影像学特征，联合最优组学特征绘制列线图，使用受试者工作特征（ROC）曲线下面积（AUC）、校准曲线和临床决策曲线（DCA）评估模型预测性能。采用自抽样进行模型验证。结果纳入的278例mGGN中，共计205例患者经病理结果证实为浸润性mGGN样肺腺癌。与惰性组比较，浸润组患者提取出的影像组学中最大三维（3D）直径、第90百分位数、逆差距归一和表面积与体积比为相关最优组学特征。单因素Logistic分析，浸润组患者中毛刺征、分叶征、支气管充气征、胸膜凹陷征、棘状突起、长径、短径和平均CT值的发生率高于惰性组（P<0.05）。多因素Logistic回归分析，分叶征与分组风险显著相关［比值比（OR）=3.724，95%置信区间（CI）：1.373~10.100，P=0.010］；长径每增加1 mm，分组风险增加31.8%（OR=1.318，95%CI：1.152~1.509，P<0.001）；平均CT值每升高1 HU，浸润性风险增加0.6%（OR=1.006，95%CI：1.003~1.009，P<0.001）。结合分叶征、结节长径、平均CT值及影像组学评分构建的融合列线图AUC为0.858（95%CI：0.813~0.914），自模型校正后一致性指数（C-index）为0.835。DCA曲线显示模型具有较好的临床适用性。Hosmer-Lemeshow检验，列线图具有较好的拟合度。结论分叶征、结节长径和平均CT值为mGGN样肺腺癌浸润性的危险因素，融合列线图对于mGGN样肺腺癌浸润性具有较好的诊断效能，可能为临床提供无创诊断和个体化治疗提供新方法。

关键词: 混合磨玻璃结节, 肺腺肿瘤, 肺癌早筛, 影像组学, 危险因素, 风险预测模型

Abstract:

Objective To discuss the imaging risk factors for invasiveness of mixed ground-glass nodule （mGGN）-like pulmonary adenocarcinoma， to construct a machine learning model combining clinical imaging features and radiomics， and to clarify its diagnostic efficacy. Methods A total of 261 patients with postoperative pathologically confirmed mGGN-like pulmonary adenocarcinoma were retrospectively collected， and 278 mGGNs met the inclusion criteria. According to the pathological results， they were divided into indolent group （n=73） and invasive group （n=205）. The clinical data and imaging data of the patients in two groups were collected. All the patients underwent thin-slicecomputed tomography（CT） scanning； radiomic features were extracted. Feature dimensionality reduction was performed on the data via Z-score standardization combined with LASSO regression， and a 10-fold cross-validation was adopted to complete the screening process. The optimal omics features were identified， and a radiomics model based on the machine learning algorithm of Logistic Regression （LR） was constructed. Univariate and multivariate logistic regression analyses were used to identify the imaging features with significance for invasiveness of mGGN-like pulmonary adenocarcinoma； nomogram was drawn by combining the optimal radiomic features； the area under the receiver operating characteristic （ROC） curve （AUC）， calibration curve， and decision curve analysis （DCA） were used to evaluate the predictive performance of the model. Bootstrap method was used for model validation. Results Among the 278 included mGGNs， a total of 205 patients were pathologically confirmed as invasive mGGN-like pulmonary adenocarcinoma. Compared with indolent group， the optimal radiomic features extracted from the patients in invasive group included maximum three dimension（3D） diameter， the 90th percentile， inverse difference moment normalized， and surface area to volume ratio. The univariate Logistic analysis results showed that the incidences of spiculation， lobulation， air bronchogram sign， pleural indentation sign， spinous process， long diameter， short diameter， and mean CT value in the patients in invasive group were higher than those in indolent group （P<0.05）. The multivariate Logistic regression analysis results showed that lobulation was significantly associated with the risk of group classification ［odds ratio（OR）=3.724， 95% confidence interval（CI）： 1.373-10.100， P=0.010］； for every 1 mm increase in long diameter， the risk of group classification was increased by 31.8% （OR=1.318， 95%CI： 1.152-1.509， P<0.001）； for every 1 HU increase in mean CT value， the risk of invasive carcinoma was increased by 0.6% （OR=1.006， 95%CI： 1.003-1.009， P<0.001）. The AUC of the fusion nomogram constructed by combining lobulation， nodule long diameter， mean CT value and radiomics score was 0.858 （95%CI： 0.813-0.914）， and the bootstrap concordance index （C-index） of model was 0.835. The DCA curve results showed that the model had good clinical applicability. The Hosmer-Lemeshow test results showed that the nomogram had a good fit. Conclusion Lobulation， nodule long diameter and mean CT value are the risk factors for invasiveness of mGGN-like pulmonary adenocarcinoma. The fusion nomogram has good diagnostic efficacy for invasiveness of mGGN-like pulmonary adenocarcinoma， and may provide a new method for non-invasive diagnosis and individualized treatment in clinical practice.

Key words: Mixed ground-glass nodules, Lung neoplasm, Early screening for lung cancer, Radiomics, Risk factor, Risk prediction model

中图分类号:

R734.2

史鸿宇,徐磊,赵磊. CT影像组学联合临床影像学特征评估混合磨玻璃结节样肺腺癌的浸润性[J]. 吉林大学学报(医学版), 2026, 52(4): 1151-1160.

Hongyu SHI,Lei XU,Lei ZHAO. Evaluation of CT radiomics combined with clinical imaging features on invasiveness of mixed ground-glass nodule-like pulmonary adenocarcinoma[J]. Journal of Jilin University(Medicine Edition), 2026, 52(4): 1151-1160.

图/表 10

图1

表1

图2

图3

图4

表2

图5

图6

图7

图8

参考文献 28

[1]	FAN X， LIANG C， MA X Q， et al. Clinical， imaging， and pathological-molecular characteristics associated with stage IA invasive lung adenocarcinoma recurrence after sub-lobar resection［J］. Acad Radiol， 2025， 32（1）： 450-459.
[2]	OH J， PIAO Z， CHO H J， et al. CT-based three-dimensional invasiveness analysis of adenocarcinoma presenting as pure ground-glass nodules［J］. Transl Cancer Res， 2023， 12（4）： 765-773.
[3]	曹立武，谢召勇，朱志磊，等. 磨玻璃结节肺腺癌组织学与CT表现特点对比研究［J］. 中国CT和MRI杂志， 2022， 20（7）： 48-51.
[4]	MENG Q C， ZHENG C B， GUO L W， et al. Construction and validation of a risk score system for diagnosing invasive adenocarcinoma presenting as pulmonary pure ground-glass nodules： a multi-center cohort study in China［J］. Quant Imaging Med Surg， 2024， 14（7）： 4864-4877.
[5]	ZHENG H， CHEN W， QI W Y， et al. Enhancing the prediction of the invasiveness of pulmonary adenocarcinomas presenting as pure ground-glass nodules： Integrating intratumor heterogeneity score with clinical-radiological features via machine learning in a multicenter study［J］. Digit Health， 2024， 10： 20552076241289181.
[6]	TANG Y H， LI M Z， LIN B K， et al. Deep learning-assisted development and validation of an algorithm for predicting the growth of persistent pure ground-glass nodules［J］. Transl Lung Cancer Res， 2023， 12（12）： 2494-2504.
[7]	中华医学会呼吸病学分会，中国肺癌防治联盟专家组.肺结节诊治中国专家共识（2024年版）［J］中华结核和呼吸杂志， 2024，47（8）：716-729
[8]	GRADISHAR W J， MORAN M S， ABRAHAM J， et al. Breast Cancer， Version 3.2024， NCCN Clinical Practice Guidelines in Oncology［J］. J Natl Compr Canc Netw， 2024， 22（5）： 331-357.
[9]	HU X F， GONG J， ZHOU W， et al. Computer-aided diagnosis of ground glass pulmonary nodule by fusing deep learning and radiomics features［J］. Phys Med Biol， 2021， 66（6）： 065015.
[10]	GAO R J， GAO Y H， ZHANG J， et al. A nomogram for predicting invasiveness of lung adenocarcinoma manifesting as pure ground-glass nodules： incorporating subjective CT signs and histogram parameters based on artificial intelligence［J］. J Cancer Res Clin Oncol， 2023， 149（17）： 15323-15333.
[11]	LIU J N， QI L L， WANG Y W， et al. Development of a combined radiomics and CT feature-based model for differentiating malignant from benign subcentimeter solid pulmonary nodules［J］. Eur Radiol Exp， 2024， 8（1）： 8.
[12]	LI Q Q， XIAO T J， LI J D， et al. The diagnosis and management of multiple ground-glass nodules in the lung［J］. Eur J Med Res， 2024， 29（1）： 305.
[13]	BARTLETT E C， KEMP S V， RAWAL B， et al. Defining growth in small pulmonary nodules using volumetry： results from a “coffee-break” CT study and implications for current nodule management guidelines［J］. Eur Radiol， 2022， 32（3）： 1912-1920.
[14]	GONG H P， SI H F， XING J L， et al. Value of MSCT multiplanar reconstruction in the differential diagnosis of ground glass nodular lung adenocarcinoma［J］ Chin J CT MRI， 2023， 21（2）： 52-54.
[15]	ZHENG C Q， WANG H F， LIU Q Z， et al. Application effect of low-dose spiral CT on pulmonary nodules and its diagnostic value for benign and malignant nodules［J］. Am J Transl Res， 2023， 15（1）： 256-263.
[16]	洪静静，文戈，张飞，等. 预测肺腺癌浸润性：基于临床特征和普通CT平扫影像学特征的联合模型［J］. 分子影像学杂志， 2024， 47（11）： 1225-1230.
[17]	KWON H J， PARK U H， GOH C J， et al. Enhancing lung cancer classification through integration of liquid biopsy multi-omics data with machine learning techniques［J］. Cancers， 2023， 15（18）： 4556.
[18]	HU K M， GAO L L， ZHANG R Y， et al. Clinical application of serum seven tumour-associated autoantibodies in patients with pulmonary nodules［J］. Heliyon， 2024， 10（9）： e30576.
[19]	DONG N， YAN Y M， LI Y X， et al. CT-based habitat radiomics for preoperative differentiation of adenocarcinoma in situ/minimally invasive adenocarcinoma from invasive adenocarcinoma manifesting as ground-glass nodules： a multicenter study［J］. Front Oncol， 2025， 15： 1660071.
[20]	WU S J， FAN L F， WU Y M， et al. Deep transfer learning radiomics combined with explainable machine learning for preoperative thymoma risk prediction based on CT［J］. Eur J Radiol， 2025， 190： 112266.
[21]	QI S L， ZUO Y F， CHANG R S， et al. Using CT radiomic features based on machine learning models to subtype adrenal adenoma［J］. BMC Cancer， 2023， 23（1）： 111.
[22]	DONG H， XI Y Z， LIU K， et al. A Radiological-Radiomics model for differentiation between minimally invasive adenocarcinoma and invasive adenocarcinoma less than or equal to 3 cm： a two-center retrospective study［J］. Eur J Radiol， 2024， 176： 111532.
[23]	刘建，师金，田国. 基于SEER数据库的年轻肺腺癌晚期患者预后预测及风险分层的模型构建［J］. 解放军医学杂志， 2024， 49（8）： 889-896.
[24]	NOBEL T B， CARR R A， CASO R， et al. Primary lung cancer in women after previous breast cancer［J］. BJS Open， 2021， 5（6）： zrab115.
[25]	CHEN P F， GONG H Y， ZHANG L， et al. CT radiomics combined with neural networks predict the malignant degree of pulmonary grinding glass nodules［J］. Front Med， 2025， 12： 1603472.
[26]	YANG X L， WANG J Y， WANG P， et al. Deep learning model using CT images for longitudinal prediction of benign and malignant ground-glass nodules［J］. Eur J Radiol， 2025， 190： 112252.
[27]	YANG Y T， XU J， WANG W， et al. Meta-analysis of the correlation between CT-based features and invasive properties of pure ground-glass nodules［J］. Asian J Surg， 2023， 46（9）： 3405-3416.
[28]	HE S Y， CHEN C E， WANG Z G， et al. The use of the mean computed-tomography value to predict the invasiveness of ground-glass nodules： a meta-analysis［J］. Asian J Surg， 2023， 46（2）： 677-682.

Characteristic	Indolent group (n=73)	Invasive group (n=205)	P
Nodule location [n(%)]			0.480
Right upper lobe	25 (34.25)	93 (45.37)
Right middle lobe	5 (6.85)	12 (5.85)
Right lower lobe	10 (13.70)	29 (14.15)
Left upper lobe	23 (31.51)	49 (23.90)
Left lower lobe	10 (13.70)	22 (10.73)
Spiculation [n(%)]			0.001
Absent	55 (75.34)	110 (53.66)
Present	18 (24.66)	95 (46.34)
Lobulation [n(%)]			<0.001
Absent	24 (32.88)	10 (4.88)
Present	49 (67.12)	195 (95.12)
Vacuole sign [n(%)]			0.695
Absent	63 (86.30)	173 (84.39)
Present	10 (13.70)	32 (15.61)
Air bronchogram sign [n(%)]			0.005
Absent	51 (69.86)	104 (50.73)
Present	22 (30.14)	101 (49.27)
Pleural indentation [n(%)]			<0.001
Absent	39 (53.42)	64 (31.22)
Present	34 (46.58)	141 (68.78)
Spinous process [n(%)]			<0.001
Absent	72 (98.63)	161 (78.54)
Present	1 (1.37)	44 (21.46)
Vascular convergence [n(%)]			0.222
Absent	69 (94.52)	184 (89.76)
Present	4 (5.48)	21 (10.24)
Vessel penetration [n(%)]			0.368
Absent	71 (97.26)	192 (93.66)
Present	2 (2.74)	13 (6.34)
Long diameter [M(P25,P75),mm]	14.00 (11.00, 17.00)	19.00 (16.00, 23.00)	<0.001
Short diameter [M(P25,P75),mm]	12.00 (9.00, 13.00)	14.00 (11.00, 17.00)	<0.001
Mean CT value [M(P25,P75),HU]	-420.50 (-500.30, -356.20)	-327.40 (-420.30, -234.40)	<0.001

Characteristic	Univariable			Multivariable
Characteristic	OR	95% CI	P	OR	95% CI	P
Nodule location
Right upper lobe	—	—
Right middle lobe	0.645	0.208, 2.003	0.448
Right lower lobe	0.780	0.335, 1.812	0.563	-	-	-
Left upper lobe	0.573	0.295, 1.112	0.100
Left lower lobe	0.591	0.248, 1.409	0.236
Spiculation
Absent	—	—		—	—
Present	2.639	1.450, 4.803	0.001	0.933	0.414, 2.102	0.867
Lobulation
Absent	—	—		—	—
Present	9.551	4.285, 21.287	<0.001	3.724	1.373, 10.100	0.010
Vacuole sign
Absent	—	—
Present	1.165	0.542, 2.508	0.696	-	-	-
Air bronchogram sign
Absent	—	—		—	—
Present	2.251	1.273, 3.981	0.005	0.515	0.235, 1.125	0.096
Pleural indentation
Absent	—	—		—	—
Present	2.527	1.463, 4.365	<0.001	0.849	0.410, 1.759	0.659
Spinous process
Absent	—	—		—	—
Present	19.677	2.659, 145.604	0.004	4.406	0.533, 36.417	0.169
Vascular convergence
Absent	—	—
Present	1.969	0.652, 5.941	0.229	-	-	-
Vessel penetration
Absent	—	—		-	-	-
Present	2.404	0.529, 10.918	0.256
Long diameter (mm)	1.334	1.224, 1.454	<0.001	1.318	1.152, 1.509	<0.001
Short diameter (mm)	1.292	1.180, 1.415	<0.001	1.010	0.862, 1.184	0.900
Mean CT value (HU)	1.006	1.004, 1.009	<0.001	1.006	1.003, 1.009	<0.001

CT影像组学联合临床影像学特征评估混合磨玻璃结节样肺腺癌的浸润性

Evaluation of CT radiomics combined with clinical imaging features on invasiveness of mixed ground-glass nodule-like pulmonary adenocarcinoma

RichHTML

PDF (PC)

赞

可视化

摘要/Abstract

引用本文

使用本文

图/表 10

参考文献 28

相关文章 15

Metrics

本文评价

推荐阅读 0

[1]	李蕊,郝晓燕,杨柳,刘家云,何睦. 基于临床实验室常规检测数据的食管癌机器学习预测模型的构建和验证[J]. 吉林大学学报(医学版), 2026, 52(4): 1086-1095.
[2]	于晶,黄胤倬,周中伟,谢雨欣,毛骞. 2型糖尿病患者并发下肢动脉粥样硬化的影响因素分析及列线图模型构建[J]. 吉林大学学报(医学版), 2026, 52(3): 806-812.
[3]	李裕洋,李想,时赫,秦一文,肖兆融,陈思浩,赵聪,刘炜炜. 口腔颌面外科患者术后感染病原学特征及其危险因素分析[J]. 吉林大学学报(医学版), 2026, 52(2): 507-512.
[4]	李晓,王莉,武文,王瑞,张爱英,张烁,姜汝佳,孟亚宁. 大疱性类天疱疮患者发生感染的临床转归及危险因素分析[J]. 吉林大学学报(医学版), 2025, 51(5): 1312-1317.
[5]	冯佳慧,刘仁杰,陈儇. 动脉瘤性蛛网膜下腔出血患者发生急性脑积水的危险因素分析[J]. 吉林大学学报(医学版), 2025, 51(3): 763-769.
[6]	袁素珍,靳焰,汪雯雯. 卵巢透明细胞癌和卵巢型子宫内膜异位症患者的临床特征分析[J]. 吉林大学学报(医学版), 2024, 50(6): 1677-1682.
[7]	于佳园,赵娣,姜新,徐静,姜丽丽,姜宏宇. 外周血淋巴细胞亚群与颈动脉粥样硬化发生的关联性分析[J]. 吉林大学学报(医学版), 2024, 50(5): 1400-1405.
[8]	齐鹏,孟宪瑛,朴美花,张强. 甲状腺微小乳头状癌复发危险因素的网状Meta分析[J]. 吉林大学学报(医学版), 2023, 49(6): 1504-1512.
[9]	冯秀南,陈露,龙钰,姜振宇,赵令. 干燥综合征并发间质性肺疾病患者临床特征及其危险因素分析[J]. 吉林大学学报(医学版), 2023, 49(6): 1513-1518.
[10]	董雪, 臧金凤, 徐彩凤, 刘贺冰, 程兆华. 重症急性胰腺炎患者肠内营养喂养不耐受风险预测模型的构建和验证[J]. 吉林大学学报(医学版), 2023, 49(6): 1586-1592.
[11]	沈童,王洋,金巍,林志辉,闫力. 老年慢性病患者认知衰弱的危险因素分析及风险模型构建[J]. 吉林大学学报(医学版), 2023, 49(5): 1304-1309.
[12]	巨名飞,刘超,马志刚,赵娟,王涂,王智昊. 残余胆固醇水平对急性冠脉综合征患者发病的预测价值[J]. 吉林大学学报(医学版), 2023, 49(3): 765-769.
[13]	高珊,王宇彤,陆敏秋,石磊,褚彬,丁月华,王梦真,鲍立. 新药时代多发性骨髓瘤患者早期死亡原因及其危险因素分析[J]. 吉林大学学报(医学版), 2022, 48(3): 783-789.
[14]	王一品, 赫东芸, 黄琼卫, 盛敏佳. 卵巢癌风险预测模型在上皮性卵巢癌早期诊断中的应用价值[J]. 吉林大学学报(医学版), 2019, 45(04): 905-910.
[15]	张昊龙, 于镇滔, 高子涵, 吴元玉, 房学东. 胃癌术后胃瘫综合征患者相关危险因素及其临床治疗[J]. 吉林大学学报(医学版), 2019, 45(03): 673-677.