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

doi:10.19982/j.issn.1000-6621.20250047

中国防痨杂志 ›› 2025, Vol. 47 ›› Issue (8): 1044-1052.doi: 10.19982/j.issn.1000-6621.20250047

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

阿布都热苏力·吐尔孙¹^,², 阿布都克尤木江·阿布力孜¹^,², 帕提曼·买买提¹, 黄陈翠³, 沈玲燕³, 马依迪丽·尼加提²^,⁴()

¹新疆喀什地区第一人民医院影像中心,喀什844000
²新疆人工智能影像辅助诊断重点实验室,喀什844000
³杭州深睿技术有限公司研发中心研究合作部,杭州311101
⁴新疆医科大学第四附属医院影像中心,乌鲁木齐830000

收稿日期:2025-02-07 出版日期:2025-08-10 发布日期:2025-08-01
通信作者: 马依迪丽·尼加提,Email: mydl0911@163.com
基金资助:
国家自然科学基金(82360359);喀什地区科技计划(2022017);第二批“天山英才”-青年托举人才项目(2023TSYCQNTJ0009);自治区卫生健康青年医学科技人才专项科研项目(WJWY-202304);结核病诊断技术研发与检测体系部署(2024B0202010005);“产学研医”一体化建设推动公立医院高质量发展(KS2023012)

Predicting pulmonary tuberculosis treatment outcomes using longitudinal chest CT radiomics and deep learning

Abuduresuli Tu’ersun¹^,², Abudukeyoumujiang Abulizi¹^,², Patiman Maimaiti¹, Huang Chencui³, Shen Lingyan³, Mayidili Nijiati²^,⁴()

¹Imaging Center, The First People’s Hospital of Kashgar Region, Kashgar 844000, China
²Xinjiang Key Laboratory of Artificial Intelligence Assisted Medical Imaging Diagnosis, Kashgar 844000, China
³Research Collaboration Department, Hangzhou Deepwise and League of PhD Technology Co., Ltd., Hangzhou 311101, China
⁴Imaging Center, Fourth Affiliated Hospital of Xinjiang Medical University, Urumqi 830000, China

Received:2025-02-07 Online:2025-08-10 Published:2025-08-01
Contact: Mayidili Nijiati, Email: mydl0911@163.com
Supported by:
National Natural Science Foundation of China(82360359);Kashi Regional Science and Technology Plan(2022017);The Second Batch of “Tianshan Talents”-Youth Lifting Talents Project(2023TSYCQNTJ0009);The Autonomous Region Health and Medical Youth Scientific Talent Special Research Project(WJWY-202304);Development of Tuberculosis Diagnostic Technologies and Deployment of Detection Systems(2024B0202010005);The Integrated Construction of “Industry-Academia-Research-Medicine” Promotes High-Quality Development of Public Hospitals(KS2023012)

摘要/Abstract

摘要：

目的: 探索利用多期胸部CT扫描的深度学习模型,预测肺结核药物治疗效果。方法: 收集2020年1月至2024年1月在新疆喀什地区第一人民医院住院接受治疗的279例肺结核患者的多时序胸部CT影像数据,包括治疗前CT和治疗后CT。提取3386个影像组学特征,构建包括基于治疗前、治疗后、结合治疗前后的胸部CT影像组学模型、基于治疗后与治疗前影像特征变化、基于治疗前后的深度学习ResNet18模型等5种预测模型。模型性能通过受试者工作特征曲线下面积(AUC)等指标进行评估,并采用SHAP和Grad-CAM技术进行模型解释和可视化。结果: 结合治疗前和治疗后CT数据的模型,在训练集和测试集上分别取得了0.845和0.770的AUC。深度学习模型在训练集和验证集上的AUC分别达到0.883和0.858。结论: 基于纵向胸部CT影像的深度学习模型在肺结核疗效预测中展现出良好的性能,能够为个性化治疗和资源优化提供支持。

关键词: 结核, 肺, 治疗结果, 预测, 人工智能, 放射摄影术

Abstract:

Objective: To develop an effective imaging-based prediction tool for reducing overmedication in pulmonary tuberculosis patients. Methods: We collected longitudinal chest CT imaging data from 279 pulmonary tuberculosis patients who received treatment at the First People’s Hospital of Kashgar between January 2020 and January 2024. The dataset included both pre-treatment and post-treatment CT scans. A total of 3386 radiomic features were extracted to construct five distinct predictive models: (1) a pre-treatment CT-based model, (2) a post-treatment CT-based model, (3) a combined pre- and post-treatment CT radiomics model, (4) a delta radiomics model based on temporal changes in imaging features between pre- and post-treatment scans, (5) a deep learning model utilizing ResNet18 architecture incorporating both pre- and post-treatment CT images. Model performance was evaluated using metrics including the area under the receiver operating characteristic curve (AUC). Additionally, SHapley Additive exPlanations (SHAP) and Gradient-weighted Class Activation Mapping (Grad-CAM) techniques were employed for model interpretation and visualization. Results: The combined model utilizing both pre- and post-treatment CT data achieved AUCs of 0.845 and 0.770 in the training and test sets, respectively. The deep learning model demonstrated superior performance with AUCs of 0.883 and 0.858 in the training and validation sets, respectively. Conclusion: The deep learning model based on longitudinal chest CT imaging demonstrated robust performance in predicting tuberculosis treatment outcomes, offering valuable support for personalized treatment strategies and resource optimization.

Key words: Tuberculosis, pulmonary, Treatment outcome, Prediction, Artificial intelligence, Radiography

中图分类号:

R521
R81

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. doi: 10.19982/j.issn.1000-6621.20250047

Abuduresuli Tu’ersun, Abudukeyoumujiang Abulizi, Patiman Maimaiti, Huang Chencui, Shen Lingyan, Mayidili Nijiati. Predicting pulmonary tuberculosis treatment outcomes using longitudinal chest CT radiomics and deep learning[J]. Chinese Journal of Antituberculosis, 2025, 47(8): 1044-1052. doi: 10.19982/j.issn.1000-6621.20250047

图/表 8

图1

表1

图2

图3

图4

表2

图5

图6

参考文献 20

[1]	Chen Y, Chen W, Cheng Z, et al. Global burden of HIV-negative multidrug- and extensively drug-resistant tuberculosis based on Global Burden of Disease Study 2021. Sci One Health, 2024, 3: 100072. doi:10.1016/j.soh.2024.100072.
[2]	Solans BP, Béranger A, Radtke K, et al. Effectiveness and Pharmacokinetic Exposures of First-Line Drugs Used to Treat Drug-Susceptible Tuberculosis in Children: A Systematic Review and Meta-Analysis. Clin Infect Dis, 2023, 76(9): 1658-70 fc. doi:10.1093/cid/ciac973. pmid: 36609692
[3]	Hussain OA, Junejo KN. Predicting treatment outcome of drug-susceptible tuberculosis patients using machine-learning models. Inform Health Soc Care, 2019, 44(2): 135-151. doi:10.1080/17538157.2018.1433676. pmid: 29461901
[4]	Abdelbary BE, Garcia-Viveros M, Ramirez-Oropesa H, et al. Predicting treatment failure, death and drug resistance using a computed risk score among newly diagnosed TB patients in Tamaulipas, Mexico. Epidemiol Infect, 2017, 145(14): 3020-3034. doi:10.1017/S0950268817001911. pmid: 28903800
[5]	Sharma MK, Stobart M, Akochy PM, et al. Evaluation of Whole Genome Sequencing-Based Predictions of Antimicrobial Resistance to TB First Line Agents: A Lesson from 5 Years of Data. Int J Mol Sci, 2024, 25(11): 6245. doi:10.3390/ijms25116245.
[6]	Zhang K, Liu X, Shen J, et al. Clinically Applicable AI System for Accurate Diagnosis, Quantitative Measurements, and Prognosis of COVID-19 Pneumonia Using Computed Tomography. Cell, 2020, 181(6): 1423-1433.e11. doi:10.1016/j.cell.2020.04.045. pmid: 32416069
[7]	Chen RY, Dodd LE, Lee M, et al. PET/CT imaging correlates with treatment outcome in patients with multidrug-resistant tuberculosis. Sci Transl Med, 2014, 6(265): 265ra166. doi:10.1126/scitranslmed.3009501.
[8]	曹振东, 李磊. 补中益气汤辅助治疗肺结核患者的效果及不良反应评价. 中国防痨杂志, 2024, 46(S1): 70-72.
[9]	中华人民共和国卫生部疾病预防控制局, 中华人民共和国卫生部医政司, 中国疾病预防控制中心. 中国结核病防治规划实施工作指南(2008年版). 北京: 中国协和医科大学出版社, 2009.
[10]	国家感染性疾病临床医学研究中心/深圳市第三人民医院,《中国防痨杂志》编辑委员会. 肺结核活动性判断规范及临床应用专家共识. 中国防痨杂志, 2020, 42(4): 301-307. doi:10.3969/j.issn.1000-6621.2020.04.001.
[11]	王昌浩, 李静. 推行结核病防治服务新体系前后肺结核病诊疗规范性分析. 疾病预防控制通报, 2018, 33(1): 40-42. doi:10.13215/j.cnki.jbyfkztb.1710017.
[12]	Han X, Sun J, Gao Y, et al. A nomogram for predicting unfavorable outcomes of antituberculosis treatment among individuals with AIDS combined with pulmonary tuberculosis in China. Front Immunol, 2025, 16: 1594107. doi:10.3389/fimmu.2025.1594107.
[13]	Nijiati M, Guo L, Tuersun A, et al. Deep learning on longitudinal CT scans: automated prediction of treatment outcomes in hospitalized tuberculosis patients. iScience, 2023, 26(11): 108326. doi:10.1016/j.isci.2023.108326.
[14]	Shen E, Wang Z, Lin T, et al. DRFNet: a deep radiomic fusion network for nAMD/PCV differentiation in OCT images. Phys Med Biol, 2024, 69(7). doi:10.1088/1361-6560/ad2ca0.
[15]	Carrasquinha E, Santinha J, Mongolin A, et al. Regularization Techniques in Radiomics: A Case Study on the Prediction of pCR in Breast Tumours and the Axilla; proceedings of the Computational Intelligence Methods for Bioinformatics and Biostatistics. Berlin: Springer International Publishing, 2020.
[16]	Ribeiro GAS, da Silva MCB, et al. Clinical validation of an artificial intelligence algorithm for classifying tuberculosis and pulmonary findings in chest radiographs. Front Artif Intell, 2025, 8: 1512910. doi:10.3389/frai.2025.1512910.
[17]	Ahamed Fayaz S, Babu L, Paridayal L, et al. Machine learning algorithms to predict treatment success for patients with pulmonary tuberculosis. PLoS One, 2024, 19(10): e0309151. doi:10.1371/journal.pone.0309151.
[18]	Kheirandish M, Catanzaro D, Crudu V, et al. Integrating landmark modeling framework and machine learning algorithms for dynamic prediction of tuberculosis treatment outcomes. J Am Med Inform Assoc, 2022, 29(5): 900-908. doi:10.1093/jamia/ocac003. pmid: 35139541
[19]	张修磊, 王倩, 夏丽, 等. 肺结核X线胸片智能辅助诊断系统在基层医院的临床效能评价. 结核与肺部疾病杂志, 2022, 3(2): 96-101. doi:10.19983/j.issn.2096-8493.20210129.
[20]	Nijiati M, Guo L, Abulizi A, et al. Deep learning and radiomics of longitudinal CT scans for early prediction of tuberculosis treatment outcomes. Eur J Radiol, 2023, 169: 111180. doi:10.1016/j.ejrad.2023.111180.

特征	有效组(202例)	无效组(62例)	统计检验值	P值
年龄[岁,M(Q₁,Q₃)]	67.0(56.0,73.0)	71.5(63.0,78.0)	U=5913.5	0.003
体质量指数[kg/m²,M(Q₁,Q₃)]	21.0(20.0,22.0)	21.0(20.0,23.0)	U=5742.0	0.127
性别[例(构成比,%)]			χ²=1.645	0.199
男性	132(65.3)	46(74.2)
女性	70(34.7)	16(25.8)

模型	组别	曲线下面积	曲线下面积 95%CI值	准确率 (%)	敏感度 (%)	特异度 (%)	F1值 (%)	阳性预测值(%)	阴性预测值(%)
放射组学_CT治疗前
	训练集	0.724	0.659~0.789	77.4	88.4	35.6	86.1	83.9	35.6
	验证集	0.640	0.471~0.810	65.1	82.1	33.3	75.4	69.7	33.3
	测试集	0.604	0.395~0.812	69.8	80.0	25.0	81.2	82.4	25.0
放射组学_CT治疗后
	训练集	0.721	0.656~0.787	74.6	81.7	47.5	83.6	85.5	47.5
	验证集	0.645	0.464~0.825	74.4	81.3	54.5	82.5	83.9	54.5
	测试集	0.646	0.473~0.819	74.4	90.0	38.5	83.1	77.1	38.5
放射组学_CT治疗前+治疗后
	训练集	0.726	0.661~0.791	70.0	72.8	59.3	79.3	87.2	59.3
	验证集	0.677	0.508~0.847	67.4	74.2	50.0	76.7	79.3	50.0
	测试集	0.675	0.484~0.866	67.4	71.4	50.0	78.1	86.2	50.0
δ-放射组学_CT治疗前+治疗后
	训练集	0.845	0.782~0.909	77.4	72.6	78.8	58.8	49.5	78.8
	验证集	0.767	0.589~0.946	74.4	54.5	81.3	52.2	50.0	81.3
	测试集	0.770	0.553~0.986	74.4	57.1	77.8	42.1	33.3	77.8
深度学习_CT治疗前+治疗后
	训练集	0.883	0.835~0.931	85.2	89.9	67.5	90.5	91.2	67.5
	验证集	0.858	0.722~0.993	81.4	84.2	60.0	88.9	94.1	60.0
	测试集	0.850	0.730~0.970	81.4	85.3	66.7	87.9	90.6	66.7

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

Predicting pulmonary tuberculosis treatment outcomes using longitudinal chest CT radiomics and deep learning

RichHTML

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

图/表 8

参考文献 20

相关文章 15

编辑推荐

Metrics

本文评价

友情链接

[1]	杜润泽, 艾尔帕提·玉素甫, 董士铭, 徐韬, 蔡晓宇, 王婷, 牙克甫·阿卜力孜, 盛伟斌, 买尔旦·买买提. Lasso-logistic回归在感染性脊柱炎鉴别诊断中的应用[J]. 中国防痨杂志, 2025, 47(S1): 1-4.
[2]	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.
[3]	刘敏, 祁丹, 石峰, 李兵, 郝瑞霞, 王东, 白俊. ARIMA-SVM组合模型在肺结核发病预测中的应用[J]. 中国防痨杂志, 2025, 47(S1): 21-25.
[4]	李松柏. PLT和HCT对腹腔镜胆囊切除术患者术后并肺部感染的预测价值[J]. 中国防痨杂志, 2025, 47(S1): 121-124.
[5]	周春燕, 蓝小春, 邱慧, 方玉林. 血糖控制水平对糖尿病并发肺结核患者抗结核治疗效果的影响[J]. 中国防痨杂志, 2025, 47(S1): 160-163.
[6]	刘玲俊, 陈丹华. 预测性护理对预防重症医学科患者呼吸机相关性肺炎发生的作用[J]. 中国防痨杂志, 2025, 47(S1): 264-266.
[7]	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.
[8]	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.
[9]	李玉红, 梅金周, 苏伟, 阮云洲, 刘玉舒, 赵雁林, 刘小秋. 2015—2021年全国65岁及以上老年利福平耐药肺结核患者治疗转归及影响因素分析[J]. 中国防痨杂志, 2025, 47(4): 408-415.
[10]	吴璇, 张艳秋, 徐吉英, 孟丹, 孙定勇. 2019—2023年河南省肺结核合并糖尿病患者治疗转归影响因素分析[J]. 中国防痨杂志, 2025, 47(4): 425-431.
[11]	安源, 白云龙, 赵庆龙, 马建军, 姜雪, 潘艳, 高迎, 高智慧. 2018—2022年吉林省肺结核合并糖尿病患者治疗转归情况及影响因素分析[J]. 中国防痨杂志, 2025, 47(4): 432-438.
[12]	杨海霞, 张小燕, 黄毅, 黎秋. 活动性肺结核合并胸部肿瘤调强放射治疗的安全性及疗效分析[J]. 中国防痨杂志, 2025, 47(3): 312-321.
[13]	刘菊秀, 张建华, 温均筠, 姜晓双. 吉林市结核分枝杆菌耐药情况及趋势预测分析[J]. 中国防痨杂志, 2025, 47(3): 348-354.
[14]	王大福, 马晓雪, 王芸, 李福栋, 饶文, 龚婷婷, 李世军, 李进岚. 2016—2023年贵州省合并糖尿病的肺结核患者就诊延迟及不良结局危险因素分析[J]. 中国防痨杂志, 2025, 47(2): 201-209.
[15]	李福栋, 马晓雪, 周建, 王大福, 张玥颖, 龚婷婷, 饶文, 洪峰, 李世军, 李进岚. 2018—2023年贵州省利福平敏感MTB/HIV双重感染患者流行特征及治疗转归分析[J]. 中国防痨杂志, 2025, 47(1): 36-43.