SVM预测模型在原发性肝癌病理分化程度判断中的应用

摘要/Abstract

摘要： 目的研究线性支持向量机(SMV)预测模型在原发性肝癌(HCC)病理分化程度判断中的应用价值。方法选择2018年6月—2020年6月在江南大学附属无锡五院确诊为HCC的患者78例。所有患者行CT纹理分析,并选取动脉期特征,采用SMV预测模型进行HCC病理分化程度判断。以术后病理检测结果作为金标准,采用Kappa检验CT纹理分析中动脉期SVM预测肿瘤分化程度的一致性进行判定。结果 78例HCC患者肿瘤分化程度分别为低分化、中分化、高分化,分别占总数的25.64 %、34.62 %、39.74 %。动脉期SVM判断HCC患者肿瘤低分化、中分化、高分化与病理检测结果比较,有较高的一致性,Kappa值分别为0.835、0.860、0.892。动脉期SVM预测HCC低分化的灵敏度为85.71 %、特异度为96.49 %、准确率为93.59 %;预测HCC中分化的灵敏度为89.29 %、特异度为96.00 %、准确率为93.59 %;预测HCC高分化的灵敏度为96.55 %、特异度为93.88 %、准确率为94.87 %。结论 SMV预测模型对判断HCC病理分化程度有较高价值。

关键词: CT纹理分析, 线性支持向量机, 原发性肝癌, 肿瘤分化程度

Abstract: Objective To study the application value of linear support vector machine (SMV) prediction model in judging the degree of pathological differentiation of primary hepatocellular carcinomaliver cancer (HCC). Methods A total of 78 patients who were diagnosed with HCC from June 2018 to June 2020 in Wuxi Fifth People's Hospital were selected. All patients underwent CT texture analysis, selected the characteristics of arterial phase, and used the SMV prediction model to judge the degree of HCC pathological differentiation. Taking postoperative pathological examination results as the gold standard, Kappa test was used to determine the consistency of the arterial phase SVM in predicting the degree of tumor differentiation in CT texture analysis. Results The degree of tumor differentiation of 78 HCC patients was poorly differentiated, moderately differentiated, and well differentiated, accounting for 25.64 %, 34.62 %, and 39.74 % of the total. The arterial phase SVM judged that the tumors of HCC patients were poorly differentiated, moderately differentiated, and highly differentiated, and the results of pathological examination were relatively consistent. The Kappa values were 0.835, 0.860, and 0.892, respectively. The sensitivity, specificity and accuracy of arterial SVM in predicting HCC differentiation were 85.71 %, 96.49 % and 93.59 %, respectively. The sensitivity, specificity and accuracy of in predicting HCC differentiation were 89.29 %, 96.00 % and 93.59 %, respectively, . The sensitivity, specificity and accuracy oinf predicting high differentiation of HCC were 96.55 %, 93.88 % and 94.87 %, respectively. Conclusion The SMV prediction model has a high value in judging the degree of HCC pathological differentiation.

Key words: CT texture analysis, lLinear support vector machine, pPrimary hepatocellular carcinoma liver cancer, tTumor differentiation

陈永莹, 彭攀, 许磊磊. SVM预测模型在原发性肝癌病理分化程度判断中的应用[J]. 肝脏, 2023, 28(7): 785-788.

CHEN Yong-ying, PENG Pan, XU Lei-lei. Application of SVM prediction model in judging the degree of pathological differentiation of primary hliver cancerepatocellular carcinoma[J]. Chinese Hepatolgy, 2023, 28(7): 785-788.

参考文献

[1] Takamoto T, Makuuchi M. Precision surgery for primary liver cancer. Cancer Biol Med, 2019, 16(3):475-485.
[2] Nderitu P, Bosco C, Garmo H, et al. The association between individual metabolic syndrome components, primary liver cancer and cirrhosis: a study in the swedish amoris cohort. Int J Cancer, 2017, 141(6):1148-1160.
[3] Kirts S, Nam BH, Panagopoulos OP, et al. Settlement prediction using support vector machine (svm)-based compressibility models: a case study. Int J Civ Eng, 2019, 17(10):1547-1557.
[4] 中国抗癌协会肝癌专业委员会, 中华医学会肝病学分会肝癌学组, 中国抗癌协会病理专业委员会, 等. 原发性肝癌规范化病理诊断指南(2015年版). 解放军医学杂志, 2015, 40(11):241-246.
[5] 曾磊, 徐静, 房娜, 等. 18F-FDGPET/CT联合同机CT增强检查在原发性肝细胞癌及分化程度诊断中的价值. 临床放射学杂志, 2016, 35(3):390-394.
[6] Hodgdon T, Mcinnes M, Schieda N, et al. Can quantitative ct texture analysis be used to differentiate fat-poor renal angiomyolipoma from renal cell carcinoma on unenhanced CT images. Radiology, 2015, 276(3):787-796.
[7] 迟剑宁, 于晓升, 张艺菲. 融合深度网络和浅层纹理特征的甲状腺结节癌变超声图像诊断. 中国图象图形学报, 2018, 23(10):1582-1593.
[8] 赵文芮, 许茂盛, 王世威, 等. DCE-MRI及DWI影像特征对乳腺癌病理组织学分级及Ki-67表达的预测研究. 中国生物医学工程学报, 2019, 38(2):176-183.
[9] 万小婷, 包晗, 黎超, 等. CT纹理分析与肝癌病理分化程度的相关性研究. 临床放射学杂志, 2019, 38(7):1239-1244.
[10] 李炳荣, 张坤, 严小斌, 等. 2018版MRI肝脏影像报告和数据系统对 ≤ 3 cm肝细胞癌与其他肝脏原发恶性肿瘤的鉴别诊断价值. 中华放射学杂志, 2021, 55(2):166-172.
[11] 杨伯文, 廖锦堂, 王译斌, 等. 胆管细胞癌与不同分化程度肝细胞癌CEUS特征的对比研究. 中国医学影像技术, 2017, 33(5):75-79.
[12] Alibek S, Uder M. Correlation of dynamic multidetector CT findings with pathological grades of hepatocellular carcinoma. Diagn Interv Radiol, 2012, 18(2):236-236.
[13] 梁长华, 梁盼, 刘甲, 等. 原发性肝癌320排容积CT全肿瘤灌注参数值与微血管密度及其与病理分级的相关性研究. 临床放射学杂志, 2017, 36(4):517-522.
[14] Simpson AL, Adams LB, Allen PJ, et al. Texture Analysis of Preoperative CT Images for Prediction of Postoperative Hepatic Insufficiency: A Preliminary Study. J Am Coll Surg, 2015, 220(3):339-346.
[15] Raman S P, Schroeder J L, Huang P, et al. Preliminary data using computed tomography texture analysis for the classification of hypervascular liver lesions: generation of a predictive model on the basis of quantitative spatial frequency measurements--a work in progress. J Comput Assist Tomo, 2015, 39(3):383-395.
[16] 吴虹霖, 雷丽程, 杨茂江, 等. 支持向量机CT引导下肺穿刺活检气胸预测模型的研究. 介入放射学杂志, 2018, 27(6):572-577.
[17] 范小波, 董莹, 王义云. CT三维最小类内散度多分类支持向量机在肺结节识别中的应用. 检验医学与临床, 2015, 12(7):940-941.
[18] 徐亚楠, 赵伟, 李铭, 等. 基于支持向量机的肺CT图像三维磨玻璃结节的提取和识别. 中国医学物理学杂志, 2019, 36(4):425-430.
[19] 张加辉, 陈峰, 薛星, 等. 基于支持向量机的MRI影像组学方法鉴别不同病理分型原发性肝癌的价值. 中华放射学杂志, 2018, 52(5):333-337.