Prediction of the properties of pulmonary ground glass nodules by CT texture features

doi:10.3877/cma.j.issn.1674-6902.2025.02.008

Abstract

Abstract:

Objective

To analyze the predictive value of clinical features and computed tomography（CT） texture features in the invasiveness of pulmonary glass nodule.

Methods

A total of 93 patients with ground glass nodules detected by chest CT in our hospital from July 2021 to September 2024 were selected as subjects. 65 patients （69.89%） were randomly selected as the training set and 28 patients （30.11%） as the verification set. According to postoperative pathology，patients with non-invasive adenocarcinoma and invasive adenocarcinoma were divided into two groups. Clinical features and CT texture features were collected，and the difference in parameters between the two groups was compared. Factors predicting the invasiveness of pulmonary ground glass nodules were screened by Logistic regression analysis，and a prediction model was built. The predictive value of the model was analyzed by receiver operating characteristic curve （ROC） and verified by empirical set.

Results

In the training set，nodule shape （χ² =4.317，P＜0.05），air bronchial sign （χ² =6.717，P ＜0.05），internal vascular sign （χ² = 11. 405，P ＜0. 05） of 27 patients with non-invasive adenocarcinoma and 38 patients with invasive adenocarcinoma were analyzed. Lobus （χ²=4.636，P＜0.05），diameter （t=4.341，P＜0.05），volume （t=6.713，P＜0.05），average CT value （t=2.936，P＜0.05），consolidation rate （t=8.371，P＜0.05），CT eigenvalue contrast （t=4.144，P＜0.05），correlation （t=4.031，P＜0.05），entropy （t= 4.797，P ＜0.05） and energy （t= 2.651，P ＜0.05） were statistically significant.Multivariate Logistic regression analysis showed that nodule volume （OR=1.371，95%CI： 1.005 ～1.869），average CT value （OR=1.032，95%CI： 1.002 ～1.062），consolidation rate （OR=2.433，95%CI： 1.256 ～4.712），CT eigenvalue contrast （OR=4.177，95%CI： 1.554～11.226），entropy （OR=1.798，95%CI： 1.035～3.123） and energy （OR=3.071，95%CI： 1.033～9.133） were the risk factors affecting the invasiveness of pulmonary glass nodule （P ＜0.05）. The result of Hosmer-Lemeshow test based on clinical and CT texture features was 12.069，P=0.148. ROC curve showed that the area under curve （AUC） of the combined model in the training set and validation set were 0.967 and 0.959 respectively，with sensitivity 92.62% and 83.33%，specificity 96.55% and 97.56%，and accuracy 91.39% and 88.76%. It was higher than the predicted value of clinical features and CT texture features.

Conclusion

Clinical features and CT texture features have high accuracy in predicting the invasiveness of pulmonary ground glass nodules，and their combined prediction improves accuracy and reliability，providing support for early invasive judgment and individualized treatment.

Key words: Pulmonary glass nodules, Clinical features, Texture feature, Electronic computed tomography, Diagnostic model

Juhua Si, Xiaofeng Jin, Longqun Liu, Weiwei Zhang, Zican Wang, Ziwen Yin. Prediction of the properties of pulmonary ground glass nodules by CT texture features[J]. Chinese Journal of Lung Diseases(Electronic Edition), 2025, 18(02): 246-250.

/ / Recommend

Add to citation manager EndNote|Ris|BibTeX

URL: https://zhfbjbzz.cma-cmc.com.cn/EN/10.3877/cma.j.issn.1674-6902.2025.02.008

https://zhfbjbzz.cma-cmc.com.cn/EN/Y2025/V18/I02/246

Figures/Tables 3

临床资料	浸润性腺癌（n=38）	非浸润性腺癌（n=27）	χ ² /t 值	P值
形状［n（％）］			4.317	0.038
圆形/ 类圆形	14（36.84）	17（62.96）
不规则	24（63.16）	10（37.04）
空气支气管征［n（％）］	19（50.00）	5（18.82）	6.717	0.010
内部血管征［n（％）］	26（68.42）	7（25.93）	11.405	0.001
分叶［n（％）］	20（52.63）	7（25.93）	4.636	0.031
直径［mm，（ $\overset{ˉ}{x}$ ± s ）］	14.31±3.19	11.26±2.47	4.341	0.000
体积［mm³，（ $\overset{ˉ}{x}$ ± s ）］	1 349.71±405.20	829.47±213.28	6.713	0.000
平均CT值［HU，（ $\overset{ˉ}{x}$ ± s ）］	-462.15±61.48	-514.24±76.25	2.936	0.005
实变率［％，（ $\overset{ˉ}{x}$ ± s ）］	24.31±7.15	13.45±3.02	8.371	0.000
对比度（ $\overset{ˉ}{x}$ ± s ）	9.42±1.20	8.24±1.08	4.144	0.000
相关性（ $\overset{ˉ}{x}$ ± s ）	0.61±0.13	0.72±0.09	4.031	0.000
熵（ $\overset{ˉ}{x}$ ± s ）	9.04±0.51	8.41±0.53	4.797	0.000
能量（ $\overset{ˉ}{x}$ ± s ）	0.08±0.02	0.07±0.01	2.651	0.010

临床资料	浸润性腺癌（n=38）	非浸润性腺癌（n=27）	χ ² /t 值	P值
形状［n（％）］			4.317	0.038
圆形/ 类圆形	14（36.84）	17（62.96）
不规则	24（63.16）	10（37.04）
空气支气管征［n（％）］	19（50.00）	5（18.82）	6.717	0.010
内部血管征［n（％）］	26（68.42）	7（25.93）	11.405	0.001
分叶［n（％）］	20（52.63）	7（25.93）	4.636	0.031
直径［mm，（ $\overset{ˉ}{x}$ ± s ）］	14.31±3.19	11.26±2.47	4.341	0.000
体积［mm³，（ $\overset{ˉ}{x}$ ± s ）］	1 349.71±405.20	829.47±213.28	6.713	0.000
平均CT值［HU，（ $\overset{ˉ}{x}$ ± s ）］	-462.15±61.48	-514.24±76.25	2.936	0.005
实变率［％，（ $\overset{ˉ}{x}$ ± s ）］	24.31±7.15	13.45±3.02	8.371	0.000
对比度（ $\overset{ˉ}{x}$ ± s ）	9.42±1.20	8.24±1.08	4.144	0.000
相关性（ $\overset{ˉ}{x}$ ± s ）	0.61±0.13	0.72±0.09	4.031	0.000
熵（ $\overset{ˉ}{x}$ ± s ）	9.04±0.51	8.41±0.53	4.797	0.000
能量（ $\overset{ˉ}{x}$ ± s ）	0.08±0.02	0.07±0.01	2.651	0.010

变　量	β	标准差	Wald	P值	OR	95％CI
变　量	β	标准差	Wald	P值	OR	下限	上限
节形状	0.685	1.440	0.226	0.634	1.983	0.118	33.346
空气支气管征	1.014	1.814	0.313	0.576	2.757	0.079	96.476
内部血管征	0.269	0.153	3.094	0.079	1.309	0.970	1.767
分叶	0.280	0.223	1.574	0.210	1.324	0.854	2.051
直径	0.219	0.262	0.700	0.403	1.245	0.745	2.078
体积	0.315	0.158	3.962	0.047	1.371	1.005	1.869
平均CT 值	0.031	0.015	4.353	0.037	1.032	1.002	1.062
实变率	0.889	0.337	6.946	0.008	2.433	1.256	4.712
对比度	1.430	0.504	8.034	0.005	4.177	1.554	11.226
相关性	-0.448	0.618	0.525	0.469	0.639	0.190	2.145
熵	0.586	0.282	4.330	0.037	1.798	1.035	3.123
能量	1.122	0.556	4.071	0.044	3.071	1.033	9.133
常数	-74.468	33.857	4.969

相关指标	AUC	敏感性（％）	特异性（％）	95％CI	准确率（％）	Youden指数
训练集
体积	0.786	69.67	80.46	0.724～0.840	70.33	0.501
平均CT 值	0.757	80.33	66.67	0.693～0.814	74.16	0.470
实变率	0.812	66.39	93.10	0.753～0.863	72.73	0.595
对比度	0.763	62.30	78.16	0.700～0.819	68.42	0.405
熵	0.838	70.49	82.76	0.781～0.885	77.51	0.533
能量	0.729	66.39	66.67	0.663～0.788	67.94	0.331
联合模型	0.967	92.62	96.55	0.933～0.987	91.39	0.892
验证集
体积	0.780	66.67	82.93	0.680～0.861	70.79	0.496
平均CT 值	0.737	83.33	56.10	0.633～0.825	70.79	0.394
实变率	0.794	62.50	97.56	0.695～0.872	76.40	0.601
对比度	0.712	81.52	60.98	0.614～0.809	69.66	0.422
熵	0.810	85.42	70.73	0.713～0.885	76.40	0.562
能量	0.677	75.00	60.98	0.570～0.773	68.54	0.360
联合模型	0.959	83.33	97.56	0.894～0.990	88.76	0.809

References 23

1	Liang X，Zhang C，Ye X. Overdiagnosis and overtreatment of groundglass nodule-like lung cancer［J］. Asia Pac J Clin Oncol，Asia Pac J Clin Oncol，2025，21（1）： 108-114.
2	Chen K，Bai J，Reuben A，et al. Multiomics analysis reveals distinct immunogenomic features of lung cancer with ground-glass opacity［J］. Am J Respir Crit Care Med，2021，204（10）： 1180-1192.
3	Hino H，Shinohara S，Okano Y，et al. Solitary ground-glass nodule mimicking lung cancer due to focal progression of usual interstitial pneumonia［J］. Int J Surg Pathol，2023，31（5）： 656-661.
4	Xue G，Jia W，Wang G，et al. Lung microwave ablation： Postprocedure imaging features and evolution of pulmonary ground-glass nodule-like lung cancer［J］. J Cancer Res Ther，2023，19（6）：1654-1662.
5	Alshammari A，Patel A，Boyle M，et al. Prevalence of invasive lung cancer in pure ground glass nodules less than 30？ mm：A systematic review［J］. Eur J Cancer，2024，213： 115116.
6	Ren J，Wang Y，Liu C，et al.Correlation analysis of clinical，pathological，imaging and genetic features of ground-glass nodule featured lung adenocarcinomas between high-risk and non-high-risk individuals［J］. Eur J Med Res，2023，28（1）： 478.
7	Feng H，Shi G，Xu Q，et al. Radiomics-based analysis of CT imaging for the preoperative prediction of invasiveness in pure ground-glass nodule lung adenocarcinomas［J］. Insights Imaging，2023，14（1）：24.
8	Kao TN，Hsieh MS，Chen LW，et al. CT-based radiomic analysis for preoperative prediction of tumor invasiveness in lung adenocarcinoma presenting as pure ground-glass nodule［J］. Cancers（Basel），2022，14（23）： 5888.
9	Luo W，Ren Y，Liu Y，et al.Imaging diagnostics of pulmonary groundglass nodules： a narrative review with current status and future directions［J］. Quant Imaging Med Surg，2024，14（8）： 6123-6146.
10	Bai C，Choi CM，Chu CM，et al. Evaluation of pulmonary nodules：clinical practice consensus guidelines for Asia［J］. Chest，2016，150（4）： 877-893.
11	Zhong F，Xu J，Wu L，et al. Comparative long-term prognosis of early surgery and surgery after surveillance for patients with groundglass nodule adenocarcinomas［J］. Sci Rep，2024，14（1）： 18785.
12	Gu M，Liu Y，Zheng W，et al. Combined targeting of senescent cells and senescent macrophages： A new idea for integrated treatment of lung cancer［J］. Semin Cancer Biol，2024，106-107：43-57.
13	Wang Q，Ba W，Yin K，et al. Predicting lung adenocarcinoma invasiveness by measurement of pure ground-glass nodule roundness by using multiplanar reformation： a retrospective analysis［J］. Clin Radiol，2022，77（1）： e20-e26.
14	Wang Z，Zhu W，Lu Z，et al. Invasive adenocarcinoma manifesting as pure ground glass nodule with different size： radiological characteristics differ while prognosis remains the same［J］. Transl Cancer Res，2021，10（6）： 2755-2766.
15	Zhang Y，Qu L，Zhang H，et al. Construction of a predictive model of 2-3 cm ground-glass nodules developing into invasive lung adenocarcinoma using high-resolution CT ［ J ］. Front Med（Lausanne），2024，11： 1403020.
16	Liu C，He Y，Luo J. Application of chest CT imaging feature model in distinguishing squamous cell carcinoma and adenocarcinoma of the lung［J］. Cancer Manag Res，2024，16： 547-557.
17	Wu J，Wang K，Deng L，et al. Growth prediction of ground-glass nodules based on pulmonary vascular morphology nomogram［J］.Acad Radiol，2024： S1076-6332（24）00887-0.
18	Zhu M，Yang Z，Wang M，et al. A computerized tomography-based radiomic model for assessing the invasiveness of lung adenocarcinoma manifesting as ground-glass opacity nodules［J］. Respir Res，2022，23（1）： 96.
19	Han T，Liu Y，Zhou J，et al. Development of an invasion score based on metastasis-related pathway activity profiles for identifying invasive molecular subtypes of lung adenocarcinoma［J］. Sci Rep，2024，14（1）： 1692.
20	Ding Y，He C，Zhao X，et al. Adding predictive and diagnostic values of pulmonary ground-glass nodules on lung cancer via novel non-invasive tests［J］. Front Med （Lausanne），2022，9： 936595.
21	Bin J，Wu M，Huang M，et al. Predicting invasion in early-stage ground-glass opacity pulmonary adenocarcinoma： a radiomics-based machine learning approach［J］. BMC Med Imaging，2024，24（1）：240.
22	熊廷伟，陶阳，李王佳，等. 分析MPVR 技术在诊断肺磨玻璃结节浸润性中的临床意义［J/CD］. 中华肺部疾病杂志（电子版），2024，17（4）： 575-579.
23	解良婕，王剑，阳韬. 采用AI 的CT 影像特征对肺结节良恶性鉴别的价值分析［J/CD］. 中华肺部疾病杂志（电子版），2024，17（2）： 242-246.

[1]	Qian Yang, Qiuyang Li, Nan Li, Yukunn Luo, Jie Tang. A radiotranscriptomics approach for prediction of prostate cancer based on ultrasound image texture features [J]. Chinese Journal of Medical Ultrasound (Electronic Edition), 2024, 21(03): 319-326.
[2]	Chao Zhang, Zhen Zhang, Liang Ma, Huanhuan Mu, Cailing Liu. Clinical characteristics and influencing factors of patients with grade C pancreatic fistula after laparoscopic pancreaticoduodenectomy [J]. Chinese Journal of Operative Procedures of General Surgery(Electronic Edition), 2024, 18(06): 675-678.
[3]	Xiangyu Qu, Yigang Zhang, Haoling Li, Tian Qiu, Yi Tan. Diagnostic and prognostic role of USP24 and its co-expression of tumor metabolic genes in hepatocellular carcinoma [J]. Chinese Journal of Operative Procedures of General Surgery(Electronic Edition), 2024, 18(06): 659-662.
[4]	Jia Gao, Jixun Jiang, Xin Wang, Ting Wu, Jiangming Xiang. Single center clinical analysis of pathological nipple discharge with 168 cases report [J]. Chinese Journal of Operative Procedures of General Surgery(Electronic Edition), 2024, 18(03): 323-326.
[5]	Liu Yang, Jia Chen, Yajuan Sun, Jiao Chen, Mingchao Tan, Mingfu Gong. Chest CT and clinical features of antineutrophil cytoplasmic antibody-associated vasculitis [J]. Chinese Journal of Lung Diseases(Electronic Edition), 2024, 17(05): 744-749.
[6]	Pan Yang, Xiaohan Huang, Caixia Deng, Lihang Zhou, Xiangdong Zhou, Hu Luo. Clinical features and prognosis of SMARCA4-deficient thoracic undifferentiated tumors: a single-center, retrospective clinical study [J]. Chinese Journal of Lung Diseases(Electronic Edition), 2024, 17(04): 529-534.
[7]	Jun He, Shuhua Yi, Tingting Chen, Yu Yang, Hongyu Li, Fei Xie, Jian He. Clinical features and diagnostic value of procalcitonin and systemic immunoinflammatory index in pregnancy complicated with community-acquired pneumonia [J]. Chinese Journal of Lung Diseases(Electronic Edition), 2024, 17(03): 421-425.
[8]	Yating You, Xi Liu, Li Bai, Liang Guo. Analysis of imaging and clinical features of patients with pulmonary aspergillosis [J]. Chinese Journal of Lung Diseases(Electronic Edition), 2024, 17(02): 229-233.
[9]	Guanglin Li, Kai Tang, Xiuzhen Lu. The research progress of partially accommodative esotropia [J]. Chinese Journal of Ophthalmologic Medicine(Electronic Edition), 2024, 14(03): 177-181.
[10]	Liangfu Zhou. Diagnosis and treatment of blast-related traumatic brain injury [J]. Chinese Journal of Neurotraumatic Surgery(Electronic Edition), 2024, 10(03): 129-131.
[11]	Yu Liu, Shaobin Duan, Zhixiang Ding, Zhitao Shi. Expression of miR-34a-5p in colon cancer patients and its correlation with clinical characteristics and prognosis [J]. Chinese Journal of Digestion and Medical Imageology(Electronic Edition), 2024, 14(06): 485-490.
[12]	Huihui Li, Xiangchun Lin. Clinical features and risk factors of early-onset colorectal cancer [J]. Chinese Journal of Clinicians(Electronic Edition), 2024, 18(02): 195-200.
[13]	Wenjuan Li, Fan Yang, Anqi Zhang, Xiaoqing Yin, Yuanyuan Li, Fangfan Zheng. Correlation between the imaging manifestation of spina bifida occulta and clinical symptoms and curative effect in children with nocturnal enuresis [J]. Chinese Journal of Interventional Radiology(Electronic Edition), 2024, 12(04): 356-361.
[14]	Yu Wang, Zekai Zhang, Mingsheng Wu, Gaoxiang Wang, Xiaohui Sun, Jun Wang, Meiqing Xu, Tian Li, Shibin Xu, Mingran Xie. Clinical characteristics of 323 patients with postoperative pathological diagnosis of benign pulmonary nodules [J]. Chinese Journal of Thoracic Surgery(Electronic Edition), 2024, 11(03): 167-174.
[15]	Hua Huang, Zhijian Liang. Research advances on malignancy related hemorrhagic stroke [J]. Chinese Journal of Cerebrovascular Diseases(Electronic Edition), 2024, 18(06): 611-615.

Please choose a citation manager

Content to export