作者投稿
专家审稿
编辑办公
邮件订阅
RSS
CT-based Radiomics to Predict Recurrence of Bladder Cancer after Resection in One Year: A Preliminary Study
-
摘要:
目的 探究基于CT图像的影像组学模型预测膀胱癌术后1年复发的可行性。 方法 回顾性纳入2014年5月至2018年7月于北京协和医院行手术治疗的膀胱癌患者，并对其进行随访，记录疾病复发状况。收集膀胱癌患者术前CT泌尿系成像实质期图像，经滤波处理后进行影像组学特征提取；采用JMIM特征选择算法识别与膀胱癌术后1年复发相关的最佳影像组学特征，采用随机森林模型、自适应增强模型、梯度提升树模型以及3个模型构成的组合模型构建膀胱癌术后1年复发的预测模型，并基于10次10折交叉验证法对各模型进行验证。采用受试者工作特征曲线对各模型的预测性能进行评定。 结果 共228例符合纳入和排除标准的膀胱癌患者入选本研究。随访1年时51例患者复发，177例患者未复发。经交叉验证，随机森林模型、自适应增强模型、梯度提升树模型和组合模型预测膀胱癌术后1年复发的曲线下面积分别为0.729(95% CI: 0.649~0.809)、0.710(95% CI: 0.627~0.793)、0.709(95% CI: 0.624~0.793)、0.732(95% CI: 0.651~0.812)，准确度分别为76.8%(95% CI: 70.6%~82.0%)、73.7%(95% CI: 67.4%~79.2%)、61.8%(95% CI: 54.7%~67.7%)、75.0%(95% CI: 68.8%~80.4%)，灵敏度分别为52.9%(95% CI: 38.6%~66.8%)、62.7%(95% CI: 48.1%~75.5%)、80.4%(95% CI: 64.3%~88.2%)、58.8%(95% CI: 44.2%~72.1%)，特异度分别为83.6%(95% CI: 77.1%~88.6%)、76.8%(95% CI: 69.8%~82.7%)、56.5%(95% CI: 48.9%~63.9%)、79.7%(95% CI: 72.8%~85.2%)。 结论 有机结合基于CT图像构建的多个影像组学模型可预测膀胱癌术后1年的复发风险。 Abstract:Objective To investigate the feasibility of the CT-based radiomics model to predict the recurrence of bladder cancer in one year postoperatively. Methods Patients with bladder cancer that received surgical treatment in Peking Union Medical College Hospital from May 2014 to July 2018 were retrospectively enrolled and followed up the recurrence of the disease. Nephrographic phase images of preoperative CT urography(CTU) performed in our hospital were collected. The images were filtered before radiomic feature extraction, and JMIM was used to identify the best radiomic features related to recurrence of bladder cancer. Random forest, AdaBoost, gradient boosting, and their combined model were used to build the model for predicting recurrence of bladder cancer after resection in one year. We applied 10-fold cross validation to validate each model and performed receiver operator characteristic curves to analyze the performance of each model. Results A total of 228 cases were included in this study according to inclusion and exclusion criteria. Fifty-one patients had recurrence and the rest 177 patients had no recurrence in one year during postoperative follow-up. In the cross validation, the random forest model, AdaBoost model, gradient boosting model and combined model predicted the recurrence of bladder cancer with AUC of 0.729(95% CI: 0.649-0.809), 0.710(95% CI: 0.627-0.793), 0.709(95% CI: 0.624-0.793)and 0.732(95% CI: 0.651-0.812), accuracy of 76.8%(95% CI: 70.6%-82.0%), 73.7%(95% CI: 67.4%-79.2%), 61.8%(95% CI: 54.7%-67.7%)and 75.0%(95% CI: 68.8%-80.4%), sensitivity of 52.9%(95% CI: 38.6%-66.8%), 62.7%(95% CI: 48.1%-75.5%), 80.4%(95% CI: 64.3%-88.2%)and 58.8%(95% CI: 44.2%-72.1%), specificity of 83.6%(95% CI: 77.1%-88.6%), 76.8%(95% CI: 69.8%-82.7%), 56.5%(95% CI: 48.9%-63.9%)and 79.7%(95% CI: 72.8%-85.2%), respectively. Conclusion Integration of CT-based radiomics prediction models can predict the recurrence risk of bladder cancer in one year postoperatively. -
Key words:
- bladder cancer /
- recurrence /
- radiomics /
- prediction model /
- computed tomography
作者贡献：张古沐阳负责研究设计、数据分析、论文撰写；许梨梨负责临床数据收集及整理；毛丽负责数据分析和论文修改；李秀丽负责数据分析指导；孙昊负责指导研究设计与数据分析及论文修改；金征宇负责研究项目设计。利益冲突：本研究所使用影像组学平台由深睿医疗人工智能研究院提供， 毛丽、李秀丽为深睿医疗人工智能研究院员工，其余作者均无利益冲突 -
图 1 3个基本模型中影像组学特征的基尼重要性分析
A.随机森林模型，贡献度最高的3个影像组学特征分别为经σ参数为4 mm的拉普拉斯高斯滤波器变换后图像ROI内GLSZM纹理特征的灰度不均匀性，经XYZ方向分别为低通、低通和高通的小波滤波器滤波后图像GLCM纹理特征的联合熵和运行熵; B.自适应增强模型，贡献度最高的3个影像组学特征分别为经XYZ方向分别为低通、高通和高通的小波滤波器滤波后图像ROI内的最大值，原始图像上GLCM纹理特征的和熵，经XYZ方向分别为高通、高通和高通的小波滤波器滤波后图像ROI内的最大值; C. 梯度提升树模型，贡献度最高的3个影像组学特征分别为经σ参数为4 mm的拉普拉斯高斯滤波器变换后图像ROI内GLSZM纹理特征的灰度不均匀性，经XYZ方向分别为低通、低通和高通的小波滤波器滤波后图像的GLCM纹理特征的联合熵，原始图像上GLCM纹理特征的和熵
ROI：感兴趣区；GLSZM：灰度区域大小矩阵；GLCM：灰度共生矩阵表 1 膀胱癌术后1年复发组与未复发组患者临床资料比较
指标 复发组(n=51) 未复发组(n=177) P值 年龄(x±s，岁) 65.5±9.9 64.1±11.5 0.475 最大病灶直径[M(P25, P75)，mm] 17.1(10.7, 25.6) 13.4(9.7, 18.5) 0.017 CT图像上病灶数量[n(%)] 0.005 单发 33(64.7) 146(82.5) 多发 18(35.3) 31(17.5) 最大病灶CT值[M(P25, P75)，HU] 66.8(60.5, 77.2) 66.2 (57.0, 75.5) 0.354 T分期[n(%)] 0.583 ≤T1期 44(86.3) 147(83.1) ≥T2期 7(13.7) 30(16.9) 
下载: 导出CSV
表 2 影像组学模型在训练组和经交叉验证对膀胱癌术后1年复发的预测效能
预测模型 AUC(95% CI) 准确度(95% CI) 灵敏度(95% CI) 特异度(95% CI) 训练集 随机森林模型 1.000(1.000~1.000) 99.6%(97.2%~100%) 100%(91.3%~100%) 99.4%(96.4%~100%) 自适应增强模型 0.952(0.924~0.979) 86.0%(80.6%~90.1%) 88.2%(75.4%~95.1%) 85.3%(79.0%~90.0%) 梯度提升树模型 0.984(0.972~0.995) 73.7%(67.4%~79.2%) 100%(91.3%~100%) 66.1%(58.6%~72.9%) 组合模型 1.000(0.999~1.000) 96.1%(92.4%~98.1%) 100%(91.3%~100%) 94.9%(90.3%~97.5%) 交叉验证 随机森林模型 0.729(0.649~0.809) 76.8%(70.6%~82.0%) 52.9%(38.6%~66.8%) 83.6%(77.1%~88.6%) 自适应增强模型 0.710(0.627~0.793) 73.7%(67.4%~79.2%) 62.7%(48.1%~75.5%) 76.8%(69.8%~82.7%) 梯度提升树模型 0.709(0.624~0.793) 61.8%(54.7%~67.7%) 80.4%(64.3%~88.2%) 56.5%(48.9%~63.9%) 组合模型 0.732(0.651~0.812) 75.0%(68.8%~80.4%) 58.8%(44.2%~72.1%) 79.7%(72.8%~85.2%) AUC：曲线下面积
下载: 导出CSV
-
[1] Bray F, Ferlay J, Soerjomataram I, et al. Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries[J]. CA Cancer J Clin, 2018, 68: 394-424. doi: 10.3322/caac.21492 [2] Pang C, Guan Y, Li H, et al. Urologic cancer in China[J]. JPN J Clin Oncol, 2016, 46: 497-501. doi: 10.1093/jjco/hyw034 [3] 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: 441-446. doi: 10.1016/j.ejca.2011.11.036 [4] Kumar V, Gu Y, Basu S, et al. Radiomics: the process and the challenges[J]. Magnetic Resonance Imaging, 2012, 30: 1234-1248. doi: 10.1016/j.mri.2012.06.010 [5] Aerts HJ, Velazquez ER, Leijenaar RT, et al. Decoding tumour phenotype by noninvasive imaging using a quantita-tive radiomics approach[J]. Nat Commun, 2014, 5: 4006. doi: 10.1038/ncomms5006 [6] Gillies RJ, Kinahan PE, Hricak H. Radiomics: Images Are More than Pictures, They Are Data[J]. Radiology, 2016, 278: 563-577. doi: 10.1148/radiol.2015151169 [7] Spiess PE, Agarwal N, Bangs R, et al. Bladder Cancer, Version 5.2017, NCCN Clinical Practice Guidelines in Oncology[J]. J Natl Compr Canc Netw, 2017, 15: 1240-1267. doi: 10.6004/jnccn.2017.0156 [8] Kluth LA, Black PC, Bochner BH, et al. Prognostic and Prediction Tools in Bladder Cancer: A Comprehensive Review of the Literature[J]. Eur Urol, 2015, 68: 238-253. doi: 10.1016/j.eururo.2015.01.032 [9] Fernandez-Gomez J, Madero R, Solsona E, et al. Predicting nonmuscle invasive bladder cancer recurrence and progression in patients treated with bacillus Calmette-Guerin: the CUETO scoring model[J]. J Urol, 2009, 182: 2195-2203. doi: 10.1016/j.juro.2009.07.016 [10] Sylvester RJ, van der Meijden AP, Oosterlinck W, et al. Predicting recurrence and progression in individual patients with stage Ta T1 bladder cancer using EORTC risk tables: a combined analysis of 2596 patients from seven EORTC trials[J]. Eur Urol, 2006, 49: 466-475. doi: 10.1016/j.eururo.2005.12.031 [11] Oderda M, Ricceri F, Pisano F, et al. Prognostic factors including Ki-67 and p53 in Bacillus Calmette-Guerin-treated non-muscle-invasive bladder cancer: a prospective study[J]. Urol Int, 2013, 90: 184-190. doi: 10.1159/000343431 [12] van Kessel KEM, van der Keur KA, Dyrskjøt L, et al. Molecular Markers Increase Precision of the European Associa-tion of Urology Non-Muscle-Invasive Bladder Cancer Progression Risk Groups[J]. Clin Cancer Res, 2018, 24: 1586-1593. doi: 10.1158/1078-0432.CCR-17-2719 [13] Tilki D, Burger M, Dalbagni G, et al. Urine markers for detection and surveillance of non-muscle-invasive bladder cancer[J]. Euro Urol, 2011, 60: 484-492. doi: 10.1016/j.eururo.2011.05.053 [14] Ueno Y, Takeuchi M, Tamada T, et al. Diagnostic Accuracy and Interobserver Agreement for the Vesical Imaging-Reporting and Data System for Muscle-invasive Bladder Cancer: A Multireader Validation Study[J]. Eur Urol, 2019, 76: 54-56. doi: 10.1016/j.eururo.2019.03.012 [15] Wu S, Zheng J, Li Y, et al. A Radiomics Nomogram for the Preoperative Prediction of Lymph Node Metastasis in Bladder Cancer[J]. Clin Cancer Res, 2017, 23: 6904-6911. doi: 10.1158/1078-0432.CCR-17-1510 [16] Zhang G, Xu L, Zhao L, et al. CT-based radiomics to predict the pathological grade of bladder cancer[J]. Eur Radiol, 2020, 30: 6749-6756. doi: 10.1007/s00330-020-06893-8 [17] Zhang G, Sun H, Shi B, et al. Quantitative CT texture analysis for evaluating histologic grade of urothelial carcinoma[J]. Abdom Radiol (NY), 2017, 42: 561-568. doi: 10.1007/s00261-016-0897-2 [18] Garapati SS, Hadjiiski L, Cha KH, et al. Urinary bladder cancer staging in CT urography using machine learning[J]. Med Phys, 2017, 44: 5814-5823. doi: 10.1002/mp.12510 -
点击查看大图
图(2) / 表(2)
计量
- 文章访问数: 400
- HTML全文浏览量: 56
- PDF下载量: 49
- 被引次数: 0
