切换至 "中华医学电子期刊资源库"
高级检索
中华腔镜泌尿外科杂志(电子版)

中华腔镜泌尿外科杂志(电子版) ›› 2022, Vol. 16 ›› Issue (01) : 7 -13. doi: 10.3877/cma.j.issn.1674-3253.2022.01.002

临床研究

基于机器学习模型分析不进行穿刺活检的前列腺根治术的可行性
陶金1, 于栓宝1, 范雅峰1, 董彪1, 洪国栋1, 屈武功1, 任选义2, 张雪培3,†()   
  1. 1. 450052　郑州大学第一附属医院泌尿外科
    2. 475000　河南，开封市中心医院泌尿外科
    3. 450052　郑州大学第一附属医院泌尿外科；450052　河南省慢性肾脏疾病精准诊疗重点实验室
  • 收稿日期:2020-04-13 出版日期:2022-02-01
  • 通信作者: 张雪培
  • 基金资助:
    河南省医学科技攻关计划联合共建项目(LHGJ20190181，LHGJ20200334)

The feasibility of radical prostatectomy without prostate biopsy based on machine-learning models

Jin Tao1, Shuanbao Yu1, Yafeng Fan1, Biao Dong1, Guodong Hong1, Wugong Qu1, Xuanyi Ren2, Xuepei Zhang3,†()   

  1. 1. Department of Urology, the First Affiliated Hospital of Zhengzhou University, Zhengzhou 450052, China
    2. Department of Urology, the Central Hospital of Kaifeng, Henan 475000, China
    3. Department of Urology, the First Affiliated Hospital of Zhengzhou University, Zhengzhou 450052, China; Key Laboratory of Precision Diagnosis and Treatment for Chronic Kidney Disease in Henan Province, Zhengzhou 450052, China
  • Received:2020-04-13 Published:2022-02-01
  • Corresponding author: Xuepei Zhang
全文 ( ) 下载PDF ( ) 导出引用
引用本文：

陶金, 于栓宝, 范雅峰, 董彪, 洪国栋, 屈武功, 任选义, 张雪培. 基于机器学习模型分析不进行穿刺活检的前列腺根治术的可行性[J/OL]. 中华腔镜泌尿外科杂志(电子版), 2022, 16(01): 7-13.

Jin Tao, Shuanbao Yu, Yafeng Fan, Biao Dong, Guodong Hong, Wugong Qu, Xuanyi Ren, Xuepei Zhang. The feasibility of radical prostatectomy without prostate biopsy based on machine-learning models[J/OL]. Chinese Journal of Endourology(Electronic Edition), 2022, 16(01): 7-13.

目的

基于机器学习算法构建前列腺癌（PCa）和临床显著前列腺癌（CSPCa）的诊断模型，并探讨不进行穿刺活检的前列腺根治术的可行性。

方法

回顾性收集2016年4月至2020年3月期间符合纳入标准的688例前列腺穿刺患者的临床资料，包括年龄、PSA、前列腺体积、磁共振报告和穿刺活检病理结果，基于人工神经网络（ANN）、支持向量机（SVM）、随机森林（RF）和Logistic回归（LR）算法，构建PCa和CSPCa的诊断模型，采用受试者工作特征曲线评价各模型的诊断准确性，并分析不进行穿刺活检的前列腺根治术的可行性。

结果

ANN（AUC=0.891和0.911）、SVM（AUC=0.903和0.925）、RF（AUC=0.897和0.916）和LR（AUC=0.894和0.917）模型预测前列腺癌和临床显著前列腺癌的诊断效能均优于PSA检测（AUC=0.805和0.792）和多参数磁共振检查（AUC=0.773和0.807） （P均<0.05）。SVM模型预测PCa和CSPCa的诊断效能均高于其他模型，但各模型间的诊断效能差异不显著（P均>0.05）。基于验证数据集，SVM的PCa预测模型能够使44例（44/208，21%）患者直接进行前列腺根治术，包括41例（41/65，63%）CSPCa和3例（3/12，25%）GS=3+3的PCa病例，而不过度治疗前列腺穿刺结果为良性的病例，优于ANN（33/208，16%）、RF（17%，36/208）和LR（35/208，17%）模型。

结论

机器学习模型预测PCa和CSPCa的诊断效能优于多参数磁共振等单一临床检查，能够使超过60%的CSPCa病例直接进行前列腺根治术，而不会导致过度治疗良性前列腺病例。但需开展前瞻性研究进一步验证该研究结果。

关键词: 前列腺癌, 机器学习, 穿刺, 算法, 模型
Objective

To develop machine-learning models for prediction of prostate cancer (PCa) and clinically significant prostate cancer (CSPCa), and explore the feasibility of radical prostatectomy without prostate biopsy based on machine-learning models.

Methods

The total of 688 patients who underwent prostate biopsy according to inclusion criteria between April 2016 and March 2020 were enrolled retrospectively, and the clinical data including age, PSA test, prostate volume (PV), reports of mpMRI examination, and pathological results of prostate biopsy were collected. The diagnostic models of PCa and CSPCa were constructed based on Artificial Neural Network (ANN), Support Vector Machine (SVC), Random Forest (RF) and Logistic Regression (LR) algorithms; and examined with receiver operating characteristic curves. Additionally, we analyzed the feasibility of radical prostatectomy without prostate biopsy based on machine-learning models.

Results

The ANN (AUC=0.89 and 0.911), SVM (AUC=0.903 and 0.925), RF (AUC=0.897 and 0.916), and LR (AUC=0.894 and 0.917) models outperformed PSA test (AUC=0.805 and 0.792) and mpMRI examination (AUC=0.773 and 0.807) in prediction of PCa and CSPCa (all P<0.05). The SVM models had higher diagnostic accuracy than other models for PCa and CSPCa, but the difference was not significant among the models (each P>0.05). The SVM model for PCa enabled 44(44/208, 21%) patients to undergo radical prostatectomy without prostate biopsy, including 41(41/65, 63%) cases of CSPCa and 3(3/12, 25%) cases of PCa (GS=3+3), without overtreatment of cases with benign biopsy in the validation cohort.

Conclusions

The machine-learning models outperformed single clinical examination such as mpMRI in prediction of PCa and CSPCa, enabling more than 60% of CSPCa cases to undergo radical prostatectomy without prostate biopsy, and would not lead to overtreatment of benign prostate cases. Further prospective validation is required.

Key words: Prostate cancer, Machine learning, Piopsy, Algorithms, Model
表1 不同Gleason评分的前列腺穿刺病例的临床特征比较[M(IQR)]
  例数 年龄（岁） tPSA（ng/ml） f/tPSA PSAD（ng/ml2） PV（ml）
良性前列腺 443 66(61~72) 10.5(6.65~17.0) 0.15(0.10~0.21) 0.18(0.11~0.29) 58(39~82)
GS=3+3 45 70(66~75) 14.2(7.70~19.4) 0.14(0.10~0.18) 0.28(0.16~0.47) 47(30~70)
GS=3+4 43 69(62~77) 19.7(11.8~27.0) 0.10(0.07~0.13) 0.57(0.36~0.68) 35(24~56)
GS=4+3 61 70(63~75) 25.5(11.9~34.1) 0.11(0.07~0.17) 0.62(0.31~1.06) 35(26~49
GS≥4+4 96 70(64~76) 23.8(10.4~34.7) 0.12(0.08~0.21) 0.44(0.25~0.72) 41(30~63)
P值   <0.001 <0.001 <0.001 <0.001 <0.001
  磁共振诊断前列腺癌[例(%)] 磁共振诊断精囊腺侵犯[例(%)] 磁共振诊断淋巴结转移[例(%)]
阴性 不确定 可疑 阴性 不确定 可疑 阴性 不确定 可疑
良性前列腺 250(56) 95(21) 98(22) 439(99) 1(0.2) 3(0.7) 432(98) 11(2) 0(0)
GS=3+3 14(31) 9(20) 22(49) 42(93) 0(0) 3(7) 43(96) 1(2) 1(2)
GS=3+4 11(26) 8(19) 24(56) 38(88) 1(2) 4(9) 41(95) 0(0) 2(5)
GS=4+3 6(10) 3(5) 52(85) 40(66) 5(8) 16(26) 52(85) 3(5) 6(10)
GS≥4+4 6(6) 9(9) 81(84) 43 (45) 1(1) 52(54) 66(69) 14(15) 16(17)
P值 <0.001 <0.001 <0.001
图1 不同模型及参数预测前列腺癌和临床显著前列腺癌的ROC曲线。（a）不同模型和参数预测前列腺癌的ROC曲线；（b）不同模型和参数预测临床显著前列腺癌的ROC曲线注：ANN：人工神经网络；SVM：支持向量机；RF：随机森林；LR：Logistic回归；mpMRI：多参数磁共振；PSA：前列腺特异性抗原检测
图2 机器学习算法的PCa和CSPCa模型预测不同Gleason评分穿刺病例的概率分布：（a）人工神经网络PCa模型的预测概率分布；（b）支持向量机PCa模型的预测概率分布；（c）随机森林PCa模型的预测概率分布；（d）Logistic回归PCa模型的预测概率分布；（e）人工神经网络CSPCa模型的预测概率分布；（f）支持向量机CSPCa模型的预测概率分布；（g）随机森林CSPCa模型的预测概率分布；（h）Logistic回归CSPCa模型的预测概率分布注：GS：Gleason评分；三条灰色虚线（从上到下）分别代表各模型诊断PCa的特异度分别为100%，99%和95%
表2 基于验证队列，机器学习和Logistic回归模型在特定前列腺癌检测特异度条件下直接前列腺癌根治术的比例和过度治疗良性前列腺病例的比例
模型 前列腺癌检测特异度 预测概率阈值 前列腺癌检测灵敏度 直接前列腺根治术(n=208)[例(%)] 过度治疗良性前列腺(n=131)[例(%)] 直接前列腺癌根治术[例(%)]
GS=3+3(n=12) GS=3+4(n=11) GS=4+3(n=18) GS≥4+4(n=36)
前列腺癌预测模型                
ANN 131/131(100%) 79% 33/77(43%) 33(16) 0(0) 1(8) 2(18) 11(61) 19(53)
SVM 131/131(100%) 65% 44/77(57%) 44(21) 0(0) 3(25) 3(27) 14(78) 24(67)
RF 131/131(100%) 69% 36/77(47%) 36(17) 0(0) 1(8) 1(9) 12(67) 22(61)
LR 131/131(100%) 83% 35/77(45% 35(17) 0(0) 1(8) 2(18) 12(67) 20(56)
ANN 130/131(99%) 75% 34/77(44%) 35(17) 1(1) 2(17) 2(18) 11(61) 19(53)
SVM 130/131(99%) 61% 45/77(58%) 46(22) 1(1) 3(25) 4(36) 14(78) 24(67)
RF 130/131(99%) 56% 43/77(56%) 44(21) 1(1) 2(17) 2(18) 14(78) 25(69)
LR 130/131(99%) 70% 44/77(57%) 45(22) 1(1) 3(25) 3(27) 14(78) 24(67)
ANN 125/131(95%) 57% 47/77(61%) 53(25) 6(5) 3(25) 4(36) 15(83) 25(69)
SVM 125/131(95%) 47% 50/77(65%) 56(27) 6(5) 4(33) 5(45) 15(83) 26(72)
RF 125/131(95%) 49% 52/77(68%) 58(28) 6(5) 3(25) 5(45) 15(83) 29(81)
LR 125/131(95%) 53% 51/77(66%) 57(27) 6(5) 4(33) 5(45) 16(89) 26(72)
临床显著前列腺癌预测模型              
ANN 131/131(100%) 70% 27/77(35%) 27(13) 0(0) 0(0) 0(0) 10(56) 17(47)
SVM 131/131(100%) 44% 43/77(56%) 43(21) 0(0) 3(25) 3(27) 14(78) 23(64)
RF 131/131(100%) 62% 31/77(40%) 31(15) 0(0) 1(8) 1(9) 11(61) 18(50)
LR 131/131(100%) 66% 36/77(47%) 36(17) 0(0) 2(17) 2(18) 12(67) 20(56)
ANN 130/131(99%) 55% 37/77(48%) 38(18) 1(1) 1(8) 1(9) 13(72) 22(61)
SVM 130/131(99%) 41% 44/77(57%) 45(22) 1(1) 3(25) 3(27) 14(78) 24(67)
RF 130/131(99%) 51% 40/77(52%) 41(20) 1(1) 2(17) 1(9) 14(78) 23(64)
LR 130/131(99%) 59% 41/77(53%) 42(20) 1(1) 2(17) 2(18) 13(72) 24(67)
ANN 125/131(95%) 43% 44/77(57%) 50(24) 6(5) 2(17) 3(27) 14(78) 25(69)
SVM 125/131(95%) 31% 49/77(64%) 55(26) 6(5) 4(33) 5(45) 14(78) 26(72)
RF 125/131(95%) 42% 49/77(64%) 55(26) 6(5) 3(25) 5(45) 14(78) 27(75)
LR 125/131(95%) 40% 51/77(66%) 57(27) 6(5) 5(42) 5(45) 15(83) 26(72)
[1]
International Agency for Research on Cancer. Global cancer observatory[EB/OL]. (2020-7-15).

URL    
[2]
方友强, 周祥福. 2020版美国国立综合癌症网络前列腺癌临床实践指南要点解读 [J/CD]. 中华腔镜泌尿外科杂志(电子版), 2020, 14(6): 405-8.
[3]
黄健, 王建业, 孔垂泽,等. 中国泌尿外科和男科疾病诊断治疗指南 [M]. 第一版，北京: 科学出版社, 2020.
[4]
Ahmed HU, EL-shater bosaily A, Brown LC, et al. Diagnostic accuracy of multi-parametric MRI and TRUS biopsy in prostate cancer (PROMIS): a paired validating confirmatory study[J]. Lancet (London, England), 2017, 389(10071): 815-822.
[5]
Loeb S, Vellekoop A, Ahmed HU, et al. Systematic review of complications of prostate biopsy [J]. Eur Urol, 2013, 64(6): 876-892.
[6]
Drost FH, Osses DF, Nieboer D, et al. Prostate MRI, with or without MRI-targeted biopsy, and systematic biopsy for detecting prostate cancer[J]. Cochrane Database Syst Rev, 2019, 4(4): Cd012663.
[7]
Scheltema MJ, Chang JI, Stricker PD, et al. Diagnostic accuracy of (68) Ga-prostate-specific membrane antigen (PSMA) positron-emission tomography (PET) and multiparametric (mp)MRI to detect intermediate-grade intra-prostatic prostate cancer using whole-mount pathology: impact of the addition of (68) Ga-PSMA PET to mpMRI[J]. BJU Int, 2019, 124 Suppl 1: 42-49.
[8]
Donato P, Roberts MJ, Morton A, et al. Improved specificity with (68)Ga PSMA PET/CT to detect clinically significant lesions "invisible" on multiparametric MRI of the prostate: a single institution comparative analysis with radical prostatectomy histology [J]. Eur J Nucl Med Mol Imaging, 2019, 46(1): 20-30.
[9]
Berger I, Annabattula C, Lewis J, et al. (68)Ga-PSMA PET/CT vs. mpMRI for locoregional prostate cancer staging: correlation with final histopathology[J]. Prostate Cancer Prostatic Dis, 2018, 21(2): 204-211.
[10]
Alberts AR, Roobol MJ, Verbeek JFM, et al. Prediction of high-grade prostate cancer following multiparametric magnetic resonance imaging: improving the rotterdam european randomized study of screening for prostate cancer risk calculators[J]. Eur Urol, 2019, 75(2): 310-318.
[11]
Yu S, Hong G, Tao J, et al. Multivariable models incorporating multiparametric magnetic resonance imaging efficiently predict results of prostate biopsy and reduce unnecessary biopsy[J]. Front In col, 2020, 10(2442).
[12]
Hu X, Cammann H, Meyer h- A, et al. Artificial neural networks and prostate cancer-tools for diagnosis and management [J]. Nat Rev Urol, 2013, 10(3): 174-182.
[13]
Delong e R, Delong DM, Clarke-pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach [J]. Biometrics, 1988, 44(3): 837-845.
[14]
Xing NZ, Wang MS, Fu Q, et al. Feasibility of prostatectomy without prostate biopsy in the era of new imaging technology and minimally invasive techniques [J]. World J Clin Cases, 2019, 7(12): 1403-1409.
[15]
European Association of Urology. EAU guidelines on prostate cancer [EB/OL]. (2020-5-8).

URL    
[16]
陈东, 李志勇, 李永红, 等. 机器人辅助前列腺癌根治术对比开放前列腺癌根治术:孰优孰劣? [J/CD]. 中华腔镜泌尿外科杂志(电子版), 2019, 13(4): 217-220.
[17]
Hung AJ, Chen J, Che Z, et al. Utilizing machine learning and automated performance metrics to evaluate robot-assisted radical prostatectomy performance and predict outcomes[J]. J Endourol, 2018, 32(5): 438-444.
[18]
罗进阳, 郑嘉欣, 蔡宗龙, 等. 基于第二版PI-RADS建立的预测模型对前列腺穿刺活检结果的预测价值 [J]. 中华泌尿外科杂志, 2019, 40(9): 673-679.
[19]
Wu YS, Zhang N, Liu SH, et al. The Huashan risk calculators performed better in prediction of prostate cancer in Chinese population: a training study followed by a validation study[J]. Asian J Androl, 2016, 18(6): 925-929.
[20]
Chen R, Xie L, Xue W, et al. Development and external multicenter validation of chinese prostate cancer consortium prostate cancer risk calculator for initial prostate biopsy[J]. Urol Oncol, 2016, 34(9): 416.e1-7.
[21]
Verbeek JFM, Bangma CH, Kweldam CF, et al. Reducing unnecessary biopsies while detecting clinically significant prostate cancer including cribriform growth with the ERSPC Rotterdam risk calculator and 4Kscore [J]. Urol Oncol, 2019, 37(2): 138-144.
[22]
Tomlins SA, Day JR, Lonigro RJ, et al. Urine TMPRSS2:ERG Plus PCA3 for individualized prostate cancer risk assessment[J]. European urology, 2016, 70(1): 45-53.
[23]
Hansen J, Auprich M, Ahyai SA, et al. Initial prostate biopsy: development and internal validation of a biopsy-specific nomogram based on the prostate cancer antigen 3 assay[J]. European urology, 2013, 63(2): 201-209.
[1] 罗辉, 方晔. 品管圈在提高甲状腺结节细针穿刺检出率中的应用[J/OL]. 中华医学超声杂志（电子版）, 2024, 21(10): 972-977.
[2] 孙佳丽, 金琳, 沈崔琴, 陈晴晴, 林艳萍, 李朝军, 徐栋. 机器人辅助超声引导下经皮穿刺的体外实验研究[J/OL]. 中华医学超声杂志（电子版）, 2024, 21(09): 884-889.
[3] 曹雨欣, 毛卓君, 梁嘉赫, 伊江浦, 张泽凯, 马文帅, 陈云涛, 李晓倩, 张宇新, 曹铁生, 袁丽君. 3D打印心脏模型在模拟左心耳封堵术临床教学中的应用价值[J/OL]. 中华医学超声杂志（电子版）, 2024, 21(06): 602-607.
[4] 奚玲, 仝瀚文, 缪骥, 毛永欢, 沈晓菲, 杜峻峰, 刘晔. 基于肌少症构建的造口旁疝危险因素预测模型[J/OL]. 中华普外科手术学杂志(电子版), 2025, 19(01): 48-51.
[5] 熊鹰, 林敬莱, 白奇, 郭剑明, 王烁. 肾癌自动化病理诊断：AI离临床还有多远？[J/OL]. 中华腔镜泌尿外科杂志(电子版), 2024, 18(06): 535-540.
[6] 李伟, 宋子健, 赖衍成, 周睿, 吴涵, 邓龙昕, 陈锐. 人工智能应用于前列腺癌患者预后预测的研究现状及展望[J/OL]. 中华腔镜泌尿外科杂志(电子版), 2024, 18(06): 541-546.
[7] 石兵, 张智, 陈金海, 唐文. 基于电磁跟踪和手术导航系统的实时超声引导下两种经皮肾盏穿刺方法的应用[J/OL]. 中华腔镜泌尿外科杂志(电子版), 2024, 18(06): 572-578.
[8] 公宇, 廖媛, 尚梅. 肝细胞癌TACE术后复发影响因素及预测模型建立[J/OL]. 中华肝脏外科手术学电子杂志, 2024, 13(06): 818-824.
[9] 刘敏思, 李荣, 李媚. 基于GGT与Plt比值的模型在HBV相关肝细胞癌诊断中的作用[J/OL]. 中华肝脏外科手术学电子杂志, 2024, 13(06): 831-835.
[10] 关小玲, 周文营, 陈洪平. PTAAR在乙肝相关慢加急性肝衰竭患者短期预后中的预测价值[J/OL]. 中华肝脏外科手术学电子杂志, 2024, 13(06): 841-845.
[11] 韦巧玲, 黄妍, 赵昌, 宋庆峰, 陈祖毅, 黄莹, 蒙嫦, 黄靖. 肝癌微波消融术后中重度疼痛风险预测列线图模型构建及验证[J/OL]. 中华临床医师杂志(电子版), 2024, 18(08): 715-721.
[12] 蔡晓雯, 李慧景, 丘婕, 杨翼帆, 吴素贤, 林玉彤, 何秋娜. 肝癌患者肝动脉化疗栓塞术后疼痛风险预测模型的构建及验证[J/OL]. 中华临床医师杂志(电子版), 2024, 18(08): 722-728.
[13] 孙铭远, 褚恒, 徐海滨, 张哲. 人工智能应用于多发性肺结节诊断的研究进展[J/OL]. 中华临床医师杂志(电子版), 2024, 18(08): 785-790.
[14] 董晟, 郎胜坤, 葛新, 孙少君, 薛明宇. 反向休克指数乘以格拉斯哥昏迷评分对老年严重创伤患者发生急性创伤性凝血功能障碍的预测价值[J/OL]. 中华临床医师杂志(电子版), 2024, 18(06): 541-547.
[15] 黄圣楷, 许斌, 苏健, 孙龙. 海南省2010~2020年乙型肝炎流行趋势的时间序列分析及预测[J/OL]. 中华临床医师杂志(电子版), 2024, 18(06): 555-561.
阅读次数
全文


摘要

版权所有 中华医学会 中华医学电子音像出版社有限责任公司  京ICP备14006079号-1  网络出版服务许可证：(署)网出证(京)字第075号