分享:
分享到微信朋友圈
X
临床研究
R2*图纹理分析预测肝细胞癌肝切除术后早期复发的价值
许岂豪 赵莹 王悦 林涛 任雪 宋清伟 郭妍 李昕 吴艇帆 刘爱连

Cite this article as: Xu QH, Zhao Y, Wang Y, et al. Value of texture analysis based on R2* map for predicting early recurrence of HCC after hepatectomy[J]. Chin J Magn Reson Imaging, 2022, 13(12): 87-92.本文引用格式:许岂豪, 赵莹, 王悦, 等. R2*图纹理分析预测肝细胞癌肝切除术后早期复发的价值[J]. 磁共振成像, 2022, 13(12): 87-92. DOI:10.12015/issn.1674-8034.2022.12.015.


[摘要] 目的 探讨基于增强T2*加权血管成像(enhanced T2* weighted angiography, ESWAN)R2*图纹理分析预测肝细胞癌(hepatocellular carcinoma, HCC)肝切除术后早期复发的价值。材料与方法 回顾性分析2011年11月至2020年5月期间于我院接受肝切除术且病理证实的81例HCC患者病例。根据肝切除术后2年内是否出现增强CT、MRI或手术病理证实肝内新发HCC病灶或肝外转移,将HCC病例分为早期复发组(n=43)和非早期复发组(n=38)。所有患者于术前1个月内行上腹部1.5 T或3.0 T MRI T1WI、T2WI及ESWAN序列扫描。使用Functool软件对ESWAN图像进行后处理,获得R2*图。在R2*图上沿肿瘤边缘勾画肿瘤所有层面,然后使用Artificial Intelligence Kit软件提取107个纹理特征,包括一阶特征、形状特征、灰度共生矩阵(gray level co-occurrence matrix, GLCM)、灰度依赖矩阵(gray level dependence matrix, GLDM)、灰度大小矩阵(gray level size zone matrix, GLSZM)、灰度游程矩阵(gray level run length matrix, GLRLM)及邻域灰度差矩阵(neighbouring gray tone difference matrix, NGTDM)。采用组内相关系数(intra-class correlation coefficient, ICC)、Spearman相关系数、梯度提升决策树(gradient boosting decision tree, GBDT)进行特征降维。建立logistic回归模型,绘制受试者工作特征(receiver operating characteristic, ROC)曲线预测复发效能,计算ROC曲线下面积(area under the curve, AUC)、准确率、敏感度及特异度。使用校准曲线、Hosmer-Lemeshow(H-L)检验评价模型拟合度。进行临床决策曲线分析(decision curve analysis, DCA)评价临床获益度。结果 经筛选得到13个最优纹理特征,包括6个一阶特征(能量、峰度、最大值、中位数、偏度和总能量),1个GLCM特征(反差校正),1个GLDM特征(大依赖低灰度优势),1个GLRLM特征(运行熵),2个GLSZM特征(尺寸区不均匀性和尺寸区非均匀性归一化),1个NGTDM特征(忙碌值),1个形状特征[最大2D直径(切面)]。Logistic回归模型预测HCC肝切除术后早期复发的AUC、准确率、敏感度及特异度为0.830(95% CI:0.740~0.920)、79.00%(95% CI:78.60%~79.40%)、83.70%(95% CI:72.70%~94.80%)及73.70%(95% CI:59.70%~87.70%)。校准曲线图显示模型预测早期复发概率与真实早期复发概率之间有很好的一致性。经H-L检验显示该模型预测校准曲线与理想模型曲线之间差异无统计学意义(P=0.493)。DCA表明R2*图纹理分析预测HCC肝切除术后早期复发的临床净获益较高。结论 基于ESWAN序列的R2*图在肿瘤氧含量水平差异的基础上结合纹理分析的方法对HCC肝切除术后早期复发有一定的预测价值。
[Abstract] Objective To investigate the feasibility of predicting early postoperative recurrence of hepatocellular carcinoma (HCC) based on R2* map texture analysis of enhanced T2* weighted angiography (ESWAN) sequence.Materials and Methods A retrospective analysis was performed of all 81 cases of patients who underwent hepatectomy and were pathologically confirmed HCC between November 2011 and May 2020. According to whether there were enhanced computed tomography or MRI or surgical pathology confirmed new intrahepatic HCC lesions or extrahepatic metastases within 2 years after hepatectomy, HCC patients were divided into the early recurrence group (n=43) and the non-early recurrence group (n=38). All patients underwent 1.5 T or 3.0 T MRI scan of upper abdomen within 1 month before surgery, including T1WI, T2WI and ESWAN sequence. ESWAN image was postprocessed by Functool software (GE AW 4.6 workstation) to obtain R2* graph. Two radiologists with 3 and 7 years of MRI diagnosis experience respectively delineated all layers of the tumor along the tumor edge on R2* maps, and then extracted 107 texture features using Artificial Intelligence Kit software. It includes first-order features, shape features, gray level co-occurrence matrix (GLCM), gray level dependence matrix (GLDM), gray level size zone matrix (GLSZM), gray level run length matrix (GLRLM) and neighbouring gray tone difference matrix (NGTDM). Intra-class correlation coefficient (ICC), Spearman correlation test and gradient boosting decision tree (GBDT) were used for feature dimension reduction. Logistic regression model was established, receiver operating characteristic (ROC) curve was drawn to predict the efficacy of recurrence, and area under the curve (AUC), precision, sensitivity and specificity were calculated. Calibration curve and Hosmer-Lemeshow (H-L) were used to test the fit degree of the valence model. Clinical decision curve analysis (DCA) was performed to evaluate the clinical benefit.Results Thirteen optimal texture features were obtained, including six first-order features (nnergy, kurtosis, maximum, median, skewness and total energy), one GLCM feature (Idn), one GLDM feature (large dependence low gray level emphasis), one GLRLM feature (run entropy), two GLSZM features (size zone non uniformity and size zone non uniformity normalized), one NGTDM feature (busyness) and one shape feature (maximum 2D diameter, Slice). Logistic regression model was established to predict AUC, accuracy, sensitivity and specificity of early recurrence after hepatectomy for HCC were 0.830 (95% CI: 0.740-0.920), 79.00% (95% CI: 78.60%-79.40%), 83.70% (95% CI: 72.70%-94.80%) and 73.70% (95% CI: 59.70%-87.70%). The calibration curve showed that there was a good consistency between the predicted early recurrence probability of the model and the real early recurrence probability. H-L test showed that there was no significant difference between the predicted calibration curve of the model and the ideal model curve (P=0.493). DCA showed that R2* map texture analysis had a higher clinical net benefit in predicting early recurrence after hepatectomy for HCC.Conclusions R2* map based on ESWAN sequence combined with texture analysis has certain predictive value for early recurrence of HCC after hepatectomy based on the difference of tumor oxygen content level.
[关键词] 肝细胞癌;肝切除术;早期复发;R2*图;纹理分析;磁共振成像
[Keywords] hepatocellular carcinoma;hepatectomy;early recurrence;R2* map;texture analysis;magnetic resonance imaging

许岂豪 1   赵莹 1   王悦 1   林涛 1   任雪 1   宋清伟 1   郭妍 2   李昕 2   吴艇帆 2   刘爱连 1, 3*  

1 大连医科大学附属第一医院放射科,大连 116011

2 通用电气药业(上海)有限公司,上海 210000

3 大连市医学影像人工智能工程技术研究中心,大连 116011

刘爱连,E-mail:liuailian@dmu.edu.cn

作者利益冲突声明:全体作者均声明无利益冲突。


基金项目: 国家自然科学基金面上项目 61971091
收稿日期:2022-07-08
接受日期:2022-11-10
中图分类号:R445.2  R735.7 
文献标识码:A
DOI: 10.12015/issn.1674-8034.2022.12.015
本文引用格式:许岂豪, 赵莹, 王悦, 等. R2*图纹理分析预测肝细胞癌肝切除术后早期复发的价值[J]. 磁共振成像, 2022, 13(12): 87-92. DOI:10.12015/issn.1674-8034.2022.12.015.

       肝细胞癌(hepatocellular carcinoma, HCC)是全球第六大常见恶性肿瘤,第三大癌症死亡原因,我国HCC患者数量占亚洲地区的一半以上[1]。肝切除术是HCC的重要治疗手段,是肝功能良好的HCC患者的首选治疗方法[2]。然而,肝切除术后HCC复发率达40%~70%[3]。对于复发可能性高的HCC患者,术后可行新辅助治疗等方式预防肿瘤复发[4]。早期复发定义为肝切除术后2年内通过增强CT或MRI发现具有典型HCC影像学特征或病理学证实的肝内新病灶或肝外转移[5]。早期复发患者的总生存率往往低于非早期复发患者[6]。因此,术前准确预测HCC患者早期复发对最佳治疗方案的选择十分重要。MRI能够实现多序列、多参数成像。既往研究表明MRI常规影像表现与HCC患者早期复发有关[7],而功能MRI,包括扩散加权成像(diffusion-weighted imaging, DWI)、扩散峰度成像(diffusion kurtosis imaging, DKI)及体素内不相干运动(intravoxel incoherent motion, IVIM)-DWI等也有助于预测HCC肝切除术后早期复发[8, 9, 10]。同样属于功能MRI的增强T2*加权血管成像(enhanced T2-star weighted angiography, ESWAN)是基于磁敏感加权成像发展而来的重度T2*加权序列,经后处理可获得能反映局部组织氧含量变化的R2*[11]。目前尚未有研究使用R2*图预测HCC肝切除术后早期复发。通过勾画肿瘤所有层面并进行纹理分析提取纹理特征能够定量评估病灶内整体的异质性[12]。本研究旨在探讨基于ESWAN序列R2*图纹理分析预测HCC肝切除术后早期复发的价值。

1 材料与方法

1.1 研究对象

       本研究遵守《赫尔辛基宣言》,并经大连医科大学附属第一医院伦理委员会批准,免除受试者知情同意,批准文号:PJ-KS-KY-2019-167。回顾性分析2011年11月至2020年5月期间于大连医科大学附属第一医院行上腹部MRI检查且手术病理证实为HCC患者的临床及影像资料,纳入排除患者流程见图1。纳入标准:(1)接受部分肝切除术,术后病理确诊为HCC;(2)术前一个月内行上腹部1.5 T或3.0 T MRI常规平扫及ESWAN序列扫描;(3)患者MRI扫描前未进行其他抗肿瘤治疗。排除标准:(1)临床资料不完整(n=3);(2)图像质量差或呼吸运动伪影严重(n=1);(3)术后2年内失访(n=24)。根据患者2年内是否复发,将纳入患者病例分为早期复发组和非早期复发组。记录患者年龄、性别、有无肝炎病史、甲胎蛋白(alpha fetal protein, AFP)、谷丙转氨酶(alanine aminotransferase, ALT)、谷草转氨酶(aspartate transaminase, AST)、γ-谷氨酰转肽酶(γ-glutamyl transpeptidase, γ-GT)、总胆红素(total bilirubin, TBIL)及Child Pugh分级等临床基本资料。

图1  患者病例筛选流程图。HCC为肝细胞癌;ESWAN为增强T2*加权血管成像。
Fig. 1  Flow chart of patient case screening. HCC is hepatocellular carcinoma; ESWAN is enhanced T2-star weighted angiography.

1.2 检查方法

       采用1.5 T或3.0 T 磁共振扫描仪(Signa HDXT,GE healthcare,美国)行上腹部MRI平扫和ESWAN序列扫描,配体部相控阵线圈,患者仰卧位。所有患者在扫描前禁食禁水4~6 h。扫描序列及参数见表1

表1  MRI扫描序列及参数
Tab. 1  MRI scanning sequence and parameters

1.3 图像分割及特征提取

       使用GE AW 4.6工作站的Functool软件对所有患者的ESWAN序列进行图像后处理,获得DICOM格式R2*图,并导入ITK-SNAP软件(v.3.6.0,www.itksnap.org)进行肿瘤分割。参照T2WI图像,由具有3年和7年诊断经验的放射科医师分别在R2*图上逐层沿肿瘤边缘勾画感兴趣区(regions of interest, ROI),经融合获得肿瘤全域感兴趣容积(volume of interest, VOI)(图2)。使用A.K.软件(Artificial Intelligence Kit, v.3.2.5, GE Healthcare)提取各病灶的107个纹理特征,包括一阶特征、形状特征、灰度共生矩阵(gray level co-occurrence matrix, GLCM)、灰度依赖矩阵(gray level dependence matrix, GLDM)、灰度大小矩阵(gray level size zone matrix, GLSZM)、灰度游程矩阵(gray level run length matrix, GLRLM)及邻域灰度差矩阵(neighbouring gray tone difference matrix, NGTDM)。

图2  男,74岁,肝右后叶下段肝细胞癌(HCC)。2A:T2WI图像上表现为稍高信号为主的团块状混杂信号影,边界清楚(箭);2B:R2*图像;2C:R2*图上沿肿瘤边缘逐层勾画感兴趣区(ROI);2D:将每个层面的病灶ROI融合,得到感兴趣容积(VOI);2E:在R2*图上勾画全肿瘤ROI,并生成相应的直方图。
Fig. 2  Male, a 74 year old patient with hepatocellular carcinoma (HCC) in the lower segment of right posterior lobe of liver. 2A: T2WI image, it shows a clump of mixed signal shadow dominated by slightly high signal, with clear boundary (arrow); 2B: R2* map; 2C: Draw regions of interest (ROI) layer by layer along the tumor edge on R2* map; 2D: Fusion ROI of lesions at each level to obtain volume of interest (VOI); 2E: Draw the whole tumor ROI on the R2* map and generate the corresponding histogram.

1.4 统计学分析

       采用SPSS 26.0(IBM,美国)统计学软件分析患者临床基本资料,分类变量采用卡方检验分析,连续变量采用Shapiro-Wilks检验正态性,符合正态分布的变量用均值±标准差(x¯±s)表示,采用独立样本t检验比较,符合偏态分布的变量用中位数(25百分位数,75百分位数)表示,使用Mann-Whitney U检验比较。采用R语言(v.3.3.2,http://www.Rproject.org)筛选纹理特征,首先采用组内相关系数(intra-class correlation coefficient, ICC)选择观察者间ICC值>0.75的特征,然后采用Spearman相关性检验剔除特征间r>0.9的特征,最后采用梯度提升决策树(gradient boosting decision tree, GBDT)进行特征降维,通过计算特征的重要度来筛选最佳特征。通过多因素logistic回归将筛选后的纹理特征建立模型,绘制受试者工作特征(receiver operating characteristic, ROC)曲线评估诊断效能,计算ROC曲线下面积(area under the curve, AUC)、准确率、敏感度及特异度。使用校准曲线、Hosmer-Lemeshow(H-L)检验评价模型拟合度。使用临床决策曲线分析(decision curve analysis, DCA)评价模型的临床获益度。为了验证模型的普适性,我们对不同场强的MRI扫描仪进行分层分析,使用Delong检验比较1.5 T及3.0 T扫描仪亚组间模型的效能。P<0.05认为差异有统计学意义。

2 结果

2.1 患者临床基本资料

       根据纳排标准本研究纳入81例HCC患者病例,其中43例纳入早期复发组,38例纳入非早期复发组,临床基本资料见表2。两组间各指标差异均无统计学意义(P>0.05)。

表2  早期复发组与非早期复发组HCC患者的临床基本资料
Tab. 2  Basic clinical data of HCC patients in early recurrence group and non early recurrence group

2.2 纹理特征

       经筛选得到13个最优纹理特征,包括:6个一阶特征(能量、峰度、最大值、中位数、偏度和总能量)、1个GLCM特征(反差校正)、1个GLDM特征(大依赖低灰度优势)、1个GLRLM特征(运行熵)、2个GLSZM特征(尺寸区不均匀性和尺寸区非均匀性归一化)、1个NGTDM特征(忙碌值)、1个形状特征[最大2D直径(切面)](表3)。建立logistic回归模型预测HCC肝切除术后早期复发,得到方程Y=0.44+4.59×能量-1.28×峰度-0.73×最大值+1.00×中位数+1.93×偏度-1.01×总能量-0.51×反差校正-0.22×大依赖低灰度优势-1.79×运行熵-2.80×尺寸区不均匀性+0.77×尺寸区非均匀性归一化+0.21×忙碌值+1.82×最大2D直径(切面)。logistic回归模型预测HCC肝切除术后早期复发的ROC曲线见图3。该预测模型的AUC、准确率、敏感度及特异度为0.830(95% CI:0.740~0.920)、79.00%(95% CI:78.60%~79.40%)、83.70%(95% CI:72.70%~94.80%)及73.70%(95% CI:59.70%~87.70%)(P<0.05)。由校准曲线图(图4)可知,模型预测早期复发概率与真实早期复发概率之间有很好的一致性。经H-L检验分析显示,该模型预测校准曲线与理想模型曲线之间无显著差异(P=0.493),体现了较高的预测性能。DCA(图5)表明,R2*图纹理分析预测HCC肝切除术后早期复发的临床净获益较高。

图3  R2*图纹理分析预测HCC肝切除术后早期复发的ROC曲线。
图4  R2*图纹理分析预测HCC肝切除术后早期复发的校准曲线。横、纵坐标分别代表模型的预测概率、真实概率,对角实线及曲线分别表示理想模型及所构建模型的预测情况,当两条线吻合情况越好,代表所构建模型预测价值越高,更接近于理想模型。
图5  R2*图纹理分析预测HCC肝切除术后早期复发的决策曲线。参照“全部干预(treat all)”策略和“不干预(treat none)”策略的两个极端曲线,决策曲线离两个极端曲线越远,模型的临床决策净获益越高。HCC为肝细胞癌;ROC为受试者工作特征;DCA为决策曲线分析。
Fig. 3  ROC curve of R2* map texture analysis for predicting early recurrence of HCC after hepatectomy.
Fig. 4  Calibration curve of R2* map texture analysis for predicting early recurrence of HCC after hepatectomy. The horizontal and vertical coordinates respectively represent the prediction probability and real probability of the model. The diagonal solid line and curve respectively represent the prediction of the ideal model and the constructed model. When the two lines coincide better, it means that the higher the prediction value of the constructed model is, the closer it is to the ideal model.
Fig. 5  Decision curve of R2* map texture analysis for predicting early recurrence of HCC after hepatectomy. Referring to the two extreme curves of the "treat all" strategy and the "treat none" strategy, the farther the decision curve is from the two extreme curves, the higher the net benefit of clinical decision-making of the model. HCC: hepatocellular carcinoma; ROC: receiver operating characteristic; DCA: decision curve analysis.
表3  经筛选得出的13个最优纹理特征
Tab. 3  Thirteen optimal texture features selected

2.3 分层分析

       1.5 T和3.0 T MRI扫描仪亚组的AUC值分别为0.756(95% CI:0.597~0.914)及0.920(95% CI:0.835~0.999)。Delong检验显示,1.5 T和3.0 T MRI扫描仪亚组间AUC值差异无统计学意义(P=0.079);MRI扫描仪两个亚组与总体患者的AUC值间差异也无统计学意义(1.5 T亚组:P=0.428;3.0 T亚组:P=0.156)。

3 讨论

       HCC是最常见的原发性肝脏恶性肿瘤,其死亡率一直居高不下[13]。肝切除术是HCC的重要治疗手段,但即使采取根治性切除术,其复发率仍然很高[14],且早期复发HCC较晚期复发HCC预后更差。术后对高复发率HCC行预防性介入治疗或新辅助治疗可以改善患者预后[15, 16]。因此,术前准确评估HCC患者术后早期复发可能性对于患者最优治疗方案的选择十分重要。临床亟需一种术前无创预测HCC早期复发的方法。目前尚未有学者使用ESWAN序列的R2*图预测HCC肝切除术后早期复发。纹理分析能够将图像转化成纹理信息,可以提取出许多影像诊断医师肉眼难以辨识的纹理特征。本研究使用纹理分析方法对81名术后病理证实为HCC患者的R2*图进行回顾性分析,得到13个最优纹理特征,AUC值为0.830,显示出基于R2*图纹理分析对于HCC肝切除术后早期复发的准确预测的优势,为临床治疗方案的选择提供重要的影像学参考。通过分层分析进一步证明了该模型在不同MRI扫描仪的临床应用中具有一定的普适性。

3.1 纹理特征

       在纹理特征方面,传统的定量参数值的测量由人工放置ROI获得,主观因素影响较大,ROI放置位置不同、ROI放置在单个层面或几个层面上均对结果有影响,无法反映肿瘤整体的异质性。勾画肿瘤全域并通过纹理分析获取纹理参数能更客观、更全面地反映肿瘤整体的异质性。一阶特征可以反映所有体素的直方图特性;GLCM特征不仅反映灰度的分布特征,也反映具有同样灰度或者接近灰度的像素之间的位置分布特性;GLSZM特征可以量化图像中连续像素值的区域;GLDM特征可量化图像中的灰度依赖关系;GLRLM特征可量化图像中的像素值的分布;NGTDM特征量化了一个灰度值与其相邻距离内的平均灰度值之间的差异。本研究筛选出13个最优纹理特征,其中峰度、偏度、反差校正、大依赖低灰度优势、运行熵、尺寸区不均匀性、尺寸区非均匀性归一化及忙碌值均用于衡量组织的均匀性,我们推测可能是由于容易早期复发的HCC细胞增殖活跃,肿瘤中异常的新生血管增加,这种异常的新生血管构造紊乱,管壁更容易破裂,肿瘤更易发生出血、坏死,肿瘤异型性更明显,导致肿瘤内部信号更不均一。容易早期复发的HCC恶性程度更高、细胞增殖更活跃,导致肿瘤耗氧量增加,因此血液代谢产物如含铁血黄素、脱氧血红蛋白等顺磁性物质增加,造成肿瘤内R2*值增加[17],在R2*图上表现为早期复发组HCC与非早期复发组HCC之间图像纹理密集程度、规则度、平滑程度及灰度存在差异。体现了图像中具有相同较低灰度值的联通区域大小的大依赖低灰度优势及能量能够反映这些差异。13个纹理特征建立的回归模型更精确、更全面反映了R2*图纹理的强度、深浅度、走向、均匀度、空间排列关系与复杂程度、光滑粗糙程度等深层次肿瘤异质性信息,具有较高的预测效能。

3.2 ESWAN序列R2*图的价值

       MRI为术前无创预测HCC早期复发提供可能。既往研究通过肿瘤边缘是否规则、动脉期瘤周强化及是否存在卫星结节等影像征象来预测HCC早期复发[7],但影像医师对于影像征象评估的主观性较强。有研究认为功能MRI的定量参数,如DWI的表观扩散系数值、DKI的平均表观峰度系数值及IVIM-DWI的真性扩散系数值能有效预测HCC早期复发[7,10]。ESWAN序列属于功能MRI的一种,其应用磁敏感技术,基于血氧水平依赖效应和不同组织之间的磁敏感性差异进行成像,具有高分辨、薄层采集的优点,一次扫描可获得多个后处理图像及相应的定量参数,已在神经系统疾病的诊断和鉴别诊断中普遍应用[18, 19]。目前,ESWAN序列逐渐应用于腹部肿瘤的研究中,如评估子宫内膜癌病理分级[20]、前列腺癌鉴别诊断[17]以及卵巢囊肿病因鉴别[21]。也有学者应用ESWAN序列评价肝纤维化严重程度[22]、鉴别肝脏良恶性肿瘤[23]及预测HCC病理分级[24]。ESWAN序列后处理得到R2*图,可以无创反映局部组织氧含量[25],局部组织氧合能力下降时R2*值增高。Hui等[5]发现T2WI和DWI图像的纹理特征参数能有效预测HCC肝切除术后早期复发,其AUC值为0.780~0.840,但上述研究仅基于肿瘤最大层面提取纹理特征,并没有提取肿瘤全域的特征,不能反映肿瘤整体的异质性。也有研究证明增强MRI动脉期和门静脉期图像纹理分析有助于预测HCC术后早期复发[26],其AUC值为0.830。与增强MRI序列相比,ESWAN序列能从血氧水平反映肿瘤的异质性,且无需注射对比剂,为真正意义上的无创性检查[27]。本研究通过R2*图纹理分析预测HCC肝切除术后早期复发的AUC值为0.830,说明在肿瘤氧含量水平差异的基础上结合纹理分析的方法来预测HCC肝切除术后早期复发具有一定优势。

3.3 本研究的局限性与展望

       首先,本研究为回顾性研究,样本量相对较少,未来会加大样本量进一步研究;其次,本研究采用手动分割肿瘤病灶,未来会尝试全自动或半自动分割方法简化流程;最后,本研究未纳入临床模型,未来会加大样本量,纳入临床模型建立影像组学模型进行全面分析。

       综上所述,本研究初步证明基于ESWAN序列的R2*图纹理分析可以从肿瘤氧含量水平多参数、全面、客观、无创地预测HCC肝切除术后早期复发,为临床医生的治疗决策提供重要参考。

[1]
Torimura T, Iwamoto H. Treatment and the prognosis of hepatocellular carcinoma in Asia[J]. Liver Int, 2022, 42(9): 2042-2054. DOI: 10.1111/liv.15130.
[2]
Mocan L. Multimodal therapy for hepatocellular carcinoma: the role of surgery[J]. Eur Rev Med Pharmacol Sci, 2021, 25(13): 4470-4477. DOI: 10.26355/eurrev_202107_26238.
[3]
Tampaki M, Papatheodoridis GV, Cholongitas E. Intrahepatic recurrence of hepatocellular carcinoma after resection: an update[J]. Clin J Gastroenterol, 2021, 14(3): 699-713. DOI: 10.1007/s12328-021-01394-7.
[4]
Liu DM, Song TQ. Changes in and challenges regarding the surgical treatment of hepatocellular carcinoma in China[J]. Biosci Trends, 2021, 15(3): 142-147. DOI: 10.5582/bst.2021.01083.
[5]
Hui T, Chuah T, Low H, et al. Predicting early recurrence of hepatocellular carcinoma with texture analysis of preoperative MRI: a radiomics study[J/OL]. Clin Radiol, 2018, 73(12): 1056.e11-e16 [2022-05-13]. https://www.clinicalradiologyonline.net/article/S0009-9260(18)30398-2/fulltext. DOI: 10.1016/j.crad.2018.07.109.
[6]
Cheng ZJ, Yang PH, Qu SP, et al. Risk factors and management for early and late intrahepatic recurrence of solitary hepatocellular carcinoma after curative resection[J]. HPB (Oxford), 2015, 17(5): 422-427. DOI: 10.1111/hpb.12367.
[7]
王乐瑶, 马霄虹, 赵心明. 磁共振成像在预测肝细胞癌术后早期复发中的研究[J]. 放射学实践, 2021, 36(6): 803-806. DOI: 10.13609/j.cnki.1000-0313.2021.06.021.
Wang LY, Ma XH, Zhao XM. Study of magnetic resonance imaging in predicting early postoperative recurrence of hepatocellular carcinoma[J]. Radiol Pract, 2021, 36(6): 803-806. DOI: 10.13609/j.cnki.1000-0313.2021.06.021.
[8]
Muhi A, Ichikawa T, Motosugi U, et al. Diffusion-weighted imaging of hepatocellular carcinoma for predicting early recurrence and survival after hepatectomy[J]. Hepatol Int, 2013, 7(2): 662-668. DOI: 10.1007/s12072-012-9383-2.
[9]
Cao LK, Chen J, Duan T, et al. Diffusion kurtosis imaging (DKI) of hepatocellular carcinoma: correlation with microvascular invasion and histologic grade[J]. Quant Imaging Med Surg, 2019, 9(4): 590-602. DOI: 10.21037/qims.2019.02.14.
[10]
李洁, 杨君梅, 乔健, 等. 体素内不相干运动扩散加权成像对肝细胞癌根治性切除术后早期复发的预测价值[J]. 中华肝胆外科杂志, 2019, 25(6): 422-425. DOI: 10.3760/cma.j.issn.1007-8118.2019.06.006.
Li J, Yang JM, Qiao J, et al. Evaluation of intravoxel incoherent motion-diffusion weighted imaging in predicting early recurrence of hepatocellular carcinoma after curative hepatectomy[J]. Chin J Hepatobiliary Surg, 2019, 25(6): 422-425. DOI: 10.3760/cma.j.issn.1007-8118.2019.06.006.
[11]
Han X, Sun MY, Wang MY, et al. The enhanced T2 star weighted angiography (ESWAN) value for differentiating borderline from malignant epithelial ovarian tumors[J]. Eur J Radiol, 2019, 118: 187-193. DOI: 10.1016/j.ejrad.2019.07.011.
[12]
Xiong H, He XJ, Guo DJ. Value of MRI texture analysis for predicting high-grade prostate cancer[J]. Clin Imaging, 2021, 72: 168-174. DOI: 10.1016/j.clinimag.2020.10.028.
[13]
Baecker A, Liu X, La Vecchia C, et al. Worldwide incidence of hepatocellular carcinoma cases attributable to major risk factors[J]. Eur J Cancer Prev, 2018, 27(3): 205-212. DOI: 10.1097/CEJ.0000000000000428.
[14]
Zheng BH, Liu LZ, Zhang ZZ, et al. Radiomics score: a potential prognostic imaging feature for postoperative survival of solitary HCC patients[J/OL]. BMC Cancer, 2018, 18(1): 1148 [2022-05-16]. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6249916. DOI: 10.1186/s12885-018-5024-z.
[15]
Pinato DJ, Fessas P, Sapisochin G, et al. Perspectives on the neoadjuvant use of immunotherapy in hepatocellular carcinoma[J]. Hepatology, 2021, 74(1): 483-490. DOI: 10.1002/hep.31697.
[16]
Qi WQ, Zhang Q, Xu Y, et al. Peg-interferon and nucleos(t)ide analogue combination at inception of antiviral therapy improves both anti-HBV efficacy and long-term survival among HBV DNA-positive hepatocellular carcinoma patients after hepatectomy/ablation[J]. J Viral Hepat, 2020, 27(4): 387-396. DOI: 10.1111/jvh.13236.
[17]
Xin JY, Gao SS, Liu JG, et al. The value of ESWAN in diagnosis and differential diagnosis of prostate cancer: preliminary study[J]. Magn Reson Imaging, 2017, 44: 26-31. DOI: 10.1016/j.mri.2017.08.002.
[18]
Ji SQ, Zhang S, Mao ZJ, et al. Quantitative assessment of iron deposition in Parkinson's disease using enhanced T2 star-weighted angiography[J]. Neurol India, 2016, 64(3): 428-435. DOI: 10.4103/0028-3886.181532.
[19]
Chen XQ, Niu JP, Peng RQ, et al. The early diagnosis of Parkinson's disease through combined biomarkers[J]. Acta Neurol Scand, 2019, 140(4): 268-273. DOI: 10.1111/ane.13140.
[20]
田士峰, 刘爱连, 李烨, 等. R2*值预估子宫内膜样腺癌病理分级的价值[J]. 中国医学影像学杂志, 2016, 24(11): 864-867. DOI: 10.3969/j.issn.1005-5185.2016.11.016.
Tian SF, Liu AL, Li Y, et al. R2* value in predicting the pathologic grade of endometrial adenocarcinoma[J]. Chin J Med Imaging, 2016, 24(11): 864-867. DOI: 10.3969/j.issn.1005-5185.2016.11.016.
[21]
Li Y, Song QW, Sun MY, et al. Use of enhanced T2 star-weighted angiography (ESWAN) and R2* values to distinguish ovarian cysts due to endometriosis from other causes[J]. Abdom Imaging, 2015, 40(6): 1733-1741. DOI: 10.1007/s00261-014-0314-7.
[22]
王丹, 陈天武, 刘启榆, 等. 肝脏ESWAN与血清铁蛋白定量评价肝纤维化的实验研究[J]. 磁共振成像, 2017, 8(12): 917-922. DOI: 10.12015/issn.1674-8034.2017.12.008.
Wang D, Chen TW, Liu QY, et al. Liver ESWAN and serum ferritin for quantitative evaluation of liver fibrosis: an experimental study[J]. Chin J Magn Reson Imaging, 2017, 8(12): 917-922. DOI: 10.12015/issn.1674-8034.2017.12.008.
[23]
王升. ESWAN序列在肝细胞癌与肝血管瘤鉴别中的应用价值初探[D]. 大连: 大连医科大学, 2013.
Wang S. Using ESWAN to differentiate hepatocellular carcinomas from cavernous hemangiomas: a preliminary research[D]. Dalian: Dalian Medical University, 2013.
[24]
田士峰, 刘爱连, 郭维亚, 等. R2*值预估肝细胞癌病理分级的可行性[J]. 中国医学影像技术, 2016, 32(11): 1693-1696. DOI: 10.13929/j.1003-3289.2016.11.017.
Tian SF, Liu AL, Guo WY, et al. Feasibility of R2* value in predicting pathological grade of hepatocellular carcinoma[J]. Chin J Med Imaging Technol, 2016, 32(11): 1693-1696. DOI: 10.13929/j.1003-3289.2016.11.017.
[25]
Wang YC, Shen YQ, Hu XM, et al. Application of R2* and apparent diffusion coefficient in estimating tumor grade and T category of bladder cancer[J]. AJR Am J Roentgenol, 2020, 214(2): 383-389. DOI: 10.2214/AJR.19.21668.
[26]
Ahn SJ, Kim JH, Park SJ, et al. Hepatocellular carcinoma: preoperative gadoxetic acid-enhanced MR imaging can predict early recurrence after curative resection using image features and texture analysis[J]. Abdom Radiol (NY), 2019, 44(2): 539-548. DOI: 10.1007/s00261-018-1768-9.
[27]
Du J, Li K, Wang W, et al. Qualitative and quantitative diagnosis of intramuscular hemangioma subtypes: Diagnostic performance comparison of ESWAN and conventional MRI[J/OL]. Acta Radiologica, 2021: 2841851211065145 [2022-05-27]. https://journals.sagepub.com/doi/10.1177/02841851211065145. DOI: 10.1177/02841851211065145.

上一篇 鼻咽癌患者放疗期间的灰质体积变化:基于体素的形态学分析
下一篇 基于术前增强MRI影像组学分析的列线图模型预测肝细胞癌切除术后复发风险的价值
  
诚聘英才 | 广告合作 | 免责声明 | 版权声明
联系电话:010-67113815
京ICP备19028836号-2