基于CE-T1WI影像组学的机器学习模型鉴别结核性与布鲁氏菌性脊柱炎的多中心研究

doi:10.19982/j.issn.1000-6621.20250524

摘要/Abstract

摘要：

目的: 探讨基于CE-T₁WI影像组学的机器学习模型在鉴别结核性脊柱炎与布鲁氏菌性脊柱炎中的价值。方法: 回顾性收集4所医学中心确诊的274例患者(结核性脊柱炎 131例,布鲁氏菌性脊柱炎 143例)的临床资料与CE-T₁WI图像。使用3D-Slicer勾画感兴趣区域(region of interest, ROI)并提取影像组学特征,经Mann-Whitney U检验(P<0.05)、Spearman相关性分析(相关系数>0.9)及LASSO算法进行特征筛选。按8∶2比例将样本随机分为训练集(219例)与验证集(55例),采用随机森林(Random Forest)、额外树(Extra Trees)、极端梯度提升决策树(XGBoost)、轻量级梯度提升决策树(LightGBM)4种机器学习算法构建模型,并通过受试者工作特征(receiver operating characteristic curve,ROC)曲线评估诊断效能。结果: 从CE-T₁WI影像中共提取107个影像组学特征,经统计学检验及LASSO算法筛选后,最终保留11个最优特征用于模型构建,包括7个纹理特征、3个形态学特征及1个一阶统计特征。基于上述特征构建的Random Forest、Extra Trees、XGBoost和LightGBM模型在验证集中的受试者工作特征曲线下面积(area under curve,AUC)分别为0.773、0.822、0.838和0.702。其中,XGBoost模型表现出最佳的综合诊断效能,准确率、敏感度和特异度均处于较高水平。结论: 基于CE-T₁WI影像组学特征构建的机器学习模型能有效鉴别结核性脊柱炎与布鲁氏菌性脊柱炎,其中XGBoost模型辅助诊断效能最佳,有望成为辅助临床决策的有效工具。

关键词: 磁共振成像, 人工智能, 结核,脊柱, 布鲁氏菌病

Abstract:

Objective: To evaluate the value of a contrast-enhanced T₁-weighted imaging (CE-T₁WI) based radiomics model in differentiating tuberculous spondylitis (TS) from brucella spondylitis (BS). Methods: Clinical data and CE-T₁WI images from 274 confirmed patients (131 TS, 143 BS) in 4 medical centers were collected. Radiomics features were extracted from manually delineated regions of interest (ROIs) using 3D-Slicer. Feature selection was performed using the Mann-Whitney U test (P<0.05), Spearman correlation analysis (coefficient>0.9), and the LASSO algorithm. The dataset was randomly split into a training cohort (n=219) and a validation cohort (n=55) in an 8∶2 ratio. Four machine learning models (Random Forest, Extra Trees, XGBoost, LightGBM) were established, and their diagnostic performances were evaluated using ROC curves. Results: A total of 107 radiomics features were initially extracted from CE-T₁WI images. After feature selection, 11 optimal features were retained for model construction, including 7 texture features, 3 shape features, and 1 first-order feature. In the validation cohort, the AUC values of Random Forest, Extra Trees, XGBoost, and LightGBM models established based on these features were 0.773, 0.822, 0.838, and 0.702, respectively. Among them, the XGBoost model demonstrated the best overall diagnostic performance with relatively high accurate rate, sensitivity and specificity. Conclusion: The CE-T₁WI-based radiomics machine learning models are effective for differentiating TS from BS. The XGBoost model showed best diagnostic performance, making it a promising tool for clinical decision-making.

Key words: Magnetic resonance imaging, Artificial intelligence, Tuberculosis, spinal, Brucellosis

中图分类号:

R445.2

郭晓雯, 焦天宇, 孙丽, 倪聪慧, 王梓霖, 王昊. 基于CE-T₁WI影像组学的机器学习模型鉴别结核性与布鲁氏菌性脊柱炎的多中心研究[J]. 中国防痨杂志, 2026, 48(8): 1184-1191. doi: 10.19982/j.issn.1000-6621.20250524

Guo Xiaowen, Jiao Tianyu, Sun Li, Ni Conghui, Wang Zilin, Wang Hao. Multi-center study on CE-T₁WI radiomics-based machine learning model in differentiating tuberculous spondylitis from brucella spondylitis[J]. Chinese Journal of Antituberculosis, 2026, 48(8): 1184-1191. doi: 10.19982/j.issn.1000-6621.20250524

图/表 8

图1

图5

图6

图7

图8

表1

表2

图9

参考文献 29

[1]	屈燕, 李涛, 马文斌, 等. 世界卫生组织《2025年全球结核病报告》解读[J]. 结核与肺部疾病杂志, 2025, 6(6): 613-623. doi:10.19983/j.issn.2096-8493.20250178.
[2]	Garg R K, Somvanshi D S. Spinal tuberculosis: a review[J]. J Spinal Cord Med, 2011, 34(5): 440-454. doi:10.1179/2045772311y.0000000023. pmid: 22118251
[3]	Jin M, Fan Z, Gao R, et al. Research progress on complications of Brucellosis[J]. Front Cell Infect Microbiol, 2023, 13: 1136674. doi:10.3389/fcimb.2023.1136674.
[4]	Cordero M, Sánchez I. Brucellar and tuberculous spondylitis. A comparative study of their clinical features[J]. J Bone Joint Surg Br, 1991, 73(1): 100-103. doi:10.1302/0301-620x.73b1.1991738.
[5]	Duarte R M, Vaccaro A R. Spinal infection: state of the art and management algorithm[J]. Eur Spine J, 2013, 22(12): 2787-2799. doi:10.1007/s00586-013-2850-1. pmid: 23756630
[6]	奇丽霞, 张凤翔, 杜海, 等. 基于CT的影像组学列线图对布鲁氏菌性脊柱炎和脊柱结核的鉴别价值[J]. 中国实验诊断学, 2025, 29(4): 440-448. doi:10.3969/j.issn.1007-4287.2025.04.013.
[7]	蔡一, 李睿涵, 伍倩, 等. 机器学习结合影像组学在结核性脊柱炎诊断与鉴别诊断的系统评价[J]. 实用放射学杂志, 2025, 41(8): 1348-1351,1360. doi:10.3969/j.issn.1002-1671.2025.08.020.
[8]	Wang W, Fan Z, Zhen J. MRI radiomics-based evaluation of tuberculous and brucella spondylitis[J]. J Int Med Res, 2023, 51(8): 3000605231195156. doi:10.1177/03000605231195156.
[9]	Guo H, Lan S, He Y, et al. Differentiating brucella spondylitis from tuberculous spondylitis by the conventional MRI and MR T2 mapping: a prospective study[J]. Eur J Med Res, 2021, 26(1): 125. doi:10.1186/s40001-021-00598-4. pmid: 34711265
[10]	Mayerhoefer M E, Materka A, Langs G, et al. Introduction to Radiomics[J]. J Nucl Med, 2020, 61(4): 488-495. doi:10.2967/jnumed.118.222893.
[11]	Lambin P, Rios-Velazquez E, Leijenaar R, et al. Radiomics: extracting more information from medical images using advanced feature analysis[J]. Eur J Cancer, 2012, 48(4): 441-446. doi:10.1016/j.ejca.2011.11.036. pmid: 22257792
[12]	Kumar V, Gu Y, Basu S, et al. Radiomics: the process and the challenges[J]. Magn Reson Imaging, 2012, 30(9): 1234-1248. doi:10.1016/j.mri.2012.06.010. pmid: 22898692
[13]	帕哈提·吐逊江, 杨来红, 何雄, 等. 影像组学在脊柱疾病中的应用[J]. 磁共振成像, 2022, 13(5): 162-166. doi:10.12015/issn.1674-8034.2022.05.035.
[14]	帕哈提·吐逊江, 杨来红, 何雄, 等. 基于FS-T₂WI序列联合机器学习对布鲁氏菌性脊柱炎与结核性脊柱炎的鉴别诊断[J]. 中华地方病学杂志, 2023, 42(5): 356-362. doi:10.3760/cma.j.cn231583-20220628-00237.
[15]	何雄, 陈艳丽, 帕哈提·吐逊江, 等. MR T₁WI影像组学对结核性脊柱炎与布鲁菌性脊柱炎的诊断价值[J]. 分子影像学杂志, 2023, 46(3):442-447. doi:10.12122/j.issn.1674-4500.2023.03.09.
[16]	Wu S, Wei Y, Li H, et al. A Predictive Clinical-Radiomics Nomogram for Differentiating Tuberculous Spondylitis from Pyogenic Spondylitis Using CT and Clinical Risk Factors[J]. Infect Drug Resist, 2022, 15: 7327-7338. doi:10.2147/idr.s388868. pmid: 36536861
[17]	Currie S, Galea-Soler S, Barron D, et al. MRI characteristics of tuberculous spondylitis[J]. Clin Radiol, 2011, 66(8): 778-787. doi:10.1016/j.crad.2011.02.016. pmid: 21570065
[18]	刘涛, 孙建民, 崔新刚, 等. MRI及病理学鉴别早期化脓性脊柱炎及布氏杆菌脊柱炎中的应用及价值[J]. 中国组织工程研究, 2014(4): 499-504. doi:10.3969/j.issn.2095-4344.2014.04.002.
[19]	陈鹰, 郭辉, 郑欢露, 等. 布鲁菌性脊柱炎临床诊断和治疗后分析[J]. 实用医学影像杂志, 2018, 19(6): 469-472. doi:10.16106/j.cnki.cn14-1281/r.2018.06.002.
[20]	Yokota Y, Fushimi Y, Okada T, et al. Evaluation of image quality of pituitary dynamic contrast-enhanced MRI using time-resolved angiography with interleaved stochastic trajectories (TWIST) and iterative reconstruction TWIST (IT-TWIST)[J]. J Magn Reson Imaging, 2020, 51(5): 1497-1506. doi:10.1002/jmri.26962. pmid: 31625655
[21]	Lee S, Choi Y H, Cho Y J, et al. Quantitative evaluation of Crohn’s disease using dynamic contrast-enhanced MRI in children and young adults[J]. Eur Radiol, 2020, 30(6): 3168-3177. doi:10.1007/s00330-020-06684-1.
[22]	Mui A W L, Lee A W M, Lee V H F, et al. Prognostic and therapeutic evaluation of nasopharyngeal carcinoma by dynamic contrast-enhanced (DCE), diffusion-weighted (DW) magnetic resonance imaging (MRI) and magnetic resonance spectroscopy (MRS)[J]. Magn Reson Imaging, 2021, 83: 50-56. doi:10.1016/j.mri.2021.07.003. pmid: 34246785
[23]	Yin P, Xu J, Sun X, et al. Intravoxel incoherent motion and dynamic contrast-enhanced magnetic resonance imaging for neoadjuvant chemotherapy response evaluation in patients with osteosarcoma[J]. Eur J Radiol, 2023, 162: 110790. doi:10.1016/j.ejrad.2023.110790.
[24]	Gleißner C, Kaczmarz S, Kufer J, et al. Hemodynamic MRI parameters to predict asymptomatic unilateral carotid artery stenosis with random forest machine learning[J]. Front Neuroimaging, 2022, 1: 1056503. doi:10.3389/fnimg.2022.1056503.
[25]	张德唯, 王梓延, 马楚塬, 等. 增强CT影像组学特征构建随机森林机器学习模型鉴别诊断小细胞肺癌与非小细胞肺癌的应用价值[J]. 影像研究与医学应用, 2025, 9(7): 50-52. doi:10.20267/j.issn.2096-3807.2025.07.017.
[26]	李立娟. 基于多器官融合和LightGBM的影像组学方法研究及其在食管静脉曲张中的应用[D]. 济南: 山东师范大学, 2021.
[27]	Greener J G, Kandathil S M, Moffat L, et al. A guide to machine learning for biologists[J]. Nat Rev Mol Cell Biol, 2022, 23(1): 40-55. doi:10.1038/s41580-021-00407-0.
[28]	郑紫涵. 基于机器学习的CT影像组学模型术前预测肝细胞癌GPC3表达状态[D]. 大理: 大理大学, 2024.
[29]	Kucukakcali Z, Akbulut S, Colak C. Machine Learning-based Prediction of HBV-related Hepatocellular Carcinoma and Detection of Key Candidate Biomarkers[J]. Medeni Med J, 2022, 37(3): 255-263. doi:10.4274/MMJ.galenos.2022.39049.

模型	准确率	AUC值	95%CI值	敏感度	特异度	阳性预测值
Random Forest	0.829	0.898	0.854~0.941	0.851	0.811	0.787
Extra Trees	0.891	0.955	0.929~0.981	0.897	0.887	0.867
XGBoost	0.891	0.955	0.929~0.981	0.839	0.934	0.912
LightGBM	0.549	0.677	0.614~0.740	0.000	1.000	0.000

模型	准确率	AUC值	95%CI值	敏感度	特异度	阳性预测值
Random Forest	0.716	0.773	0.670~0.876	0.523	0.946	0.920
Extra Trees	0.753	0.822	0.733~0.913	0.682	0.838	0.833
XGBoost	0.765	0.838	0.753~0.924	0.727	0.811	0.821
LightGBM	0.457	0.702	0.606~0.798	0.000	1.000	0.000

基于CE-T₁WI影像组学的机器学习模型鉴别结核性与布鲁氏菌性脊柱炎的多中心研究

Multi-center study on CE-T₁WI radiomics-based machine learning model in differentiating tuberculous spondylitis from brucella spondylitis

RichHTML

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

图/表 8

参考文献 29

相关文章 15

编辑推荐

Metrics

本文评价

[1]	孙郑, 刘炯雅, 陈驰, 于全骥, 李亚男, 陈诚, 竺丽梅. 人工智能辅助CT阅片系统在综合医院肺结核早发现中的应用[J]. 中国防痨杂志, 2026, 48(5): 586-593.
[2]	付旭文, 何维曜臻, 李翔. 男性生殖系统结核影像学表现[J]. 中国防痨杂志, 2026, 48(2): 188-196.
[3]	杜润泽, 董士铭, 艾尔帕提·玉素甫, 徐韬, 蔡晓宇, 王婷, 牙克甫·阿卜力孜, 盛伟斌, 买尔旦·买买提. 布鲁氏菌性脊柱炎的精确诊断：多因素Logistic回归预测模型的构建与验证[J]. 中国防痨杂志, 2025, 47(S1): 17-20.
[4]	中国防痨协会影像专业分会, 中华医学会结核病学分会, 中国防痨协会标准化专业分会, 中国防痨协会结核病控制专业分会. 胸部影像学检查人工智能辅助读片技术在肺结核患者发现中的应用专家共识[J]. 中国防痨杂志, 2025, 47(9): 1093-1104.
[5]	焦家欢, 孙长峰, 吴刚, 黄富礼, 盛云建. 基于机器学习算法的诊断模型对结核性胸腔积液的应用价值[J]. 中国防痨杂志, 2025, 47(8): 1053-1061.
[6]	阿布都热苏力·吐尔孙, 阿布都克尤木江·阿布力孜, 帕提曼·买买提, 黄陈翠, 沈玲燕, 马依迪丽·尼加提. 基于纵向胸部CT的影像组学深度学习预测肺结核患者疗效的研究[J]. 中国防痨杂志, 2025, 47(8): 1044-1052.
[7]	朱庆东, 赵春艳, 谢周华, 宋树林, 宋畅. 基于人工智能的CT影像组学在结核病诊断和治疗反应监测中应用的研究进展[J]. 中国防痨杂志, 2025, 47(8): 1068-1076.
[8]	谭黄圣, 蓝志明, 付远飞, 赖居易, 冯华龙, 蒋勇, 邓鹏伟, 何升华. 颈椎布鲁氏菌感染一例[J]. 中国防痨杂志, 2025, 47(3): 380-383.
[9]	宁锋钢, 房坤, 王珏, 吕岩, 贺伟, 侯代伦. 磁共振三维颅脑容积成像增强扫描在颅内结核影像诊断中的应用[J]. 中国防痨杂志, 2025, 47(11): 1489-1494.
[10]	李婷婷, 刘欢庆, 雷倩, 尤著宏, 赵国连. 基于机器学习算法的诊断模型对结核性胸膜炎的应用价值[J]. 中国防痨杂志, 2025, 47(11): 1508-1514.
[11]	杨敏, 王新文. 间质性肺病影像学诊断研究进展[J]. 中国防痨杂志, 2024, 46(S1): 316-318.
[12]	李翔, 浦英, 付旭文, 杞敏, 魏佳璐, 寸新华. 误诊为脊柱结核的真菌性脊柱炎临床及影像特征分析[J]. 中国防痨杂志, 2024, 46(9): 1109-1114.
[13]	李汶翰, 杨静, 李春华. 人工智能在肺结核影像诊断及耐药性预测中的研究进展[J]. 中国防痨杂志, 2024, 46(9): 1098-1103.
[14]	刘雪艳, 王芳, 李春华, 唐光孝, 郑娇凤, 王惠秋, 李玉蕊, 王佳男, 舒伟强, 吕圣秀. 基于深度学习的继发性肺结核CT辅助诊断模型构建及验证[J]. 中国防痨杂志, 2024, 46(3): 279-287.
[15]	李多, 吕平欣. 推动人工智能技术在肺结核影像领域的发展和运用[J]. 中国防痨杂志, 2024, 46(3): 253-259.