Skip to main content
Journal of Central South University Medical Sciences logoLink to Journal of Central South University Medical Sciences
. 2022 Aug 28;47(8):1049–1057. [Article in Chinese] doi: 10.11817/j.issn.1672-7347.2022.220027

感兴趣区范围对CT影像组学模型预测肝细胞癌微血管侵犯的影响

Influence of different region of interest sizes on CT-based radiomics model for microvascular invasion prediction in hepatocellular carcinoma

ZHAO Huafei 1,2,1, FENG Zhichao 1, LI Huiling 1, YAO Shanhu 1, ZHENG Wei 1,, RONG Pengfei 1,
Editor: 陈 丽文
PMCID: PMC10950113  PMID: 36097772

Abstract

Objective

Microvascular invasion (MVI) is an important predictor of postoperative recurrence or poor outcomes of hepatocellular carcinoma (HCC). Radiomics is able to predict MVI in HCC preoperatively. This study aims to investigate the influence of different region of interest (ROI) sizes on CT-based radiomics model for MVI prediction in HCC.

Methods

Patients with HCC with or without MVI confirmed by pathology and those who underwent preoperative plain or enhanced abdominal CT scans in the Third Xiangya Hospital of Central South University from January 2010 to December 2020 were retrospectively and consecutively included. According to the ratio of 7 to 3, the patients were randomly assigned into a training set and a validation set. Clinical data were collected from medical records, and radiomics features were extracted from the arterial phase (AP) and portal venous phase (PVP) of preoperatively acquired CT in all patients. Six different ROI sizes were employed. The original ROI (OROI) was manually delineated along the visible borders of the tumor layer-by-layer. The OROI was expanded out by 1-5 mm. The OROI was combined with 5 different peritumoral regions to generate the other 5 ROIs, named Plus1-Plus5. Feature extraction, dimension reduction, and model development were conducted in 6 different ROIs separately. Supporter vector machine (SVM) was used for model construction. Model performance was assessed via receiver operating characteristic (ROC) curve.

Results

A total of 172 HCC patients were included, in which 83 (48.3%) were MVI positive, and 89 (51.7%) were MVI negative. Three hundred and ninety-six features based on AP or PVP images were extracted from each ROI. After feature selection and dimension reduction, 4, 5, 15, 11, 6, and 3 features of OROI, Plus1, Plus2, Plus 3, Plus4, and Plus5 were selected for model construction, respectively. In the training set, the sensitivity, specificity, and area under the curve (AUC) of OROI were 0.759, 0.806, and 0.855, respectively. The AUC values of Plus2 (0.979) and Plus3 (0.954) were higher than that of OROI. The AUC values of Plus1 (0.802), Plus4 (0.792), and Plus5 (0.774) were not significantly different from those of OROI. In the validation set, the sensitivity, specificity, and AUC value of OROI were 0.640, 0.630, and 0.664, respectively. The AUC value of Plus3 was 0.903, which was higher than that of OROI. The AUC values of Plus1 (0.679), Plus2 (0.536), Plus4 (0.708), and Plus5 (0.757) were not significantly different from that of OROI (P>0.05).

Conclusion

The size of ROI significantly inflluences on the performance of CT-based radiomics model for MVI prediction in HCC. Including appropriate area around the tumor into ROI could improve the predictive performance of the model, and 3 mm might be appropriate distance.

Keywords: hepatocellular carcinoma, computed tomography, radiomics, region of interest, microvascular invasion


肝细胞癌(hepatocellular carcinoma,HCC)是肝癌的主要病理类型,约占肝癌总数的75%[1]。手术切除和肝移植是有可能治愈HCC的主要方法,然而HCC术后易早期复发和转移[2]。微血管侵犯(microvascular invasion,MVI)是HCC术后复发和预后不良的重要预测因子[3]。MVI是指在显微镜下在内皮细胞衬附的血管腔内见到癌细胞的巢团[4],只能在术后确诊。术前有效预测HCC有无MVI对制订治疗方案、判断预后有重要价值。肿瘤的体积、肿瘤边缘不光整、动脉期瘤周强化、肝胆期瘤周低信号以及多灶性等影像特征能提示HCC存在MVI[5-6]。但是,常规的影像检查联合人工阅片的方式受医师主观判断及工作经验的影响,可靠性不佳[7]。临床需要更加客观、定量的方法对MVI进行术前评估。

影像组学能从影像中高通量地提取海量信息,定量并客观地描述肿瘤的异质性,达到对肿瘤进行分类和预测的目的。肿瘤周围由实质构成的瘤周区域能反映肿瘤微环境的状态,对于临床评估肿瘤侵袭行为具有重要价值[8]。已有数项研究发现,影像组学模型能在术前无创、有效地预测HCC有无MVI,但感兴趣区(region of interest,ROI)的选择仍存在争议:Ma等[9]沿肿瘤边缘勾画ROI;Xu等[10]认为沿肿瘤边缘勾画ROI对MVI具有最佳预测效能;汪禾青等[11]将瘤内及瘤周3 mm的区域作为一个整体进行纹理分析;Zhang等[12]及Feng等[13]将瘤内及瘤周1 cm范围的联合区域作为ROI;Zheng等[14]将肿瘤边缘分别向内收缩和向外扩展5像素,两者之间围成一个环形的ROI。但是,预测HCC有无MVI的最佳ROI范围仍不清楚。本研究通过选用6个不同的ROI范围分别建模,以探索影像组学模型预测HCC有无MVI的最佳ROI范围。

1. 对象与方法

1.1. 对象

本研究经中南大学湘雅三医院伦理委员会批准(审批号:2020-S494),并豁免患者知情同意。在中南大学湘雅三医院病历系统中回顾性连续收集2010年1月至2020年12月的HCC患者,患者的纳入标准:1)经肿瘤切除术或肝移植术后病理证实为HCC;2)术前1个月内进行过腹部CT平扫或增强扫描;3)术前未进行过其他抗肿瘤治疗。排除标准:1)合并其他类型的肿瘤;2)无法确定MVI情况;3)CT图像不符合分析要求。收集患者的临床资料如姓名、性别、年龄、MVI情况、有无病毒性肝炎、有无肝硬化的影像学表现、Child-Pugh分级、巴塞罗那临床肝癌(Barcelona Clinic Liver Cancer,BCLC)分期、甲胎蛋白(alpha-fetoprotein,AFP)及肝肾功能等。所有的患者按照7꞉3的比例随机分为训练组和验证组。

1.2. MVI的病理诊断

所有病理标本皆来自肝癌切除术后或因肝癌行肝移植术后患者。石蜡标本的取材依据《原发性肝癌规范化病理诊断指南》[15]进行操作。石蜡切片厚度为4 nm。将切片进行HE染色,步骤为:脱蜡、水化、脱水、透明、封片。然后在显微镜下对MVI进行病理分级:M0,未观察到MVI;M1,≤5个MVI,并且发生于瘤周1 cm以内;M2,>5个MVI,或MVI距离肿瘤超过1 cm。M0级为MVI阴性,M1及M2级为MVI阳性。

1.3. CT图像的获取

图像来自两台不同的CT扫描仪。扫描参数如下:1)Philips Brilliance 64排CT,管电压为120 kV,管电流为240 mA,旋转时间为0.5 s,矩阵为512×512,视野为350 mm×350 mm,层厚为5 mm;2)GE Revolution 256排CT,管电压为120 kV,管电流为135~240 mA,自动剂量调制,旋转时间为0.5 s,矩阵为512×512,视野为330 mm×330 mm,层厚为 5 mm。在平扫完成后,用非离子型造影剂(碘海醇,300 mg/mL;GE医疗)以60~110 mL(1.5 mL/kg)的剂量和3.0 mL/s的平均注射速率给药。使用团注跟踪技术:在腹主动脉衰减值达到100 HU后15 s采集动脉期图像;在动脉期后30 s和180 s分别采集门静脉期和延迟期的图像。所有纳入患者的平扫及增强CT图像均从放射科影像存储和传输系统(picture archiving and communication system,PACS)导出并保存为DICOM格式用于后续处理。由3位放射科医师对CT图像特征进行评估,包括病灶横截面最大直径、肿瘤边界是否光整、有无瘤内坏死、有无瘤内出血、有无瘤内动脉、肿瘤有无假包膜、有无“快进快出”强化模式(动脉期明显强化,门脉期及延迟期迅速廓清呈相对低密度)。当CT图像特征难以判断时,由3名医师经过讨论确定。最终由1名放射科副主任医师对全部图像特征进行审核。

1.4. ROI的分割

3名有5年工作经验的放射科医师经过培训后进行ROI勾画,3名医师对患者的临床信息及MVI情况皆不知情。将患者的图像随机分配给以上3名医师。患者的动脉期及门脉期图像用于ROI勾画,平扫及延迟期图像用于帮助确定肿瘤位置及大小。原始感兴趣区(original region of interest,OROI)的分割在ITK-SNAP软件(Version 3.6.0,www.itksnap.org)中进行,先在横断面沿着肿瘤边缘手动逐层勾画病灶从而形成肿瘤的三维ROI,勾画时应尽量避开临近的大血管。而后将原始图像及OROI一同导入A.K.(Artificial Intelligence Kit,Version 3.2.2.R,GE医疗)软件,利用软件的“dilation”功能将OROI向外扩展1~5 mm。OROI分别与瘤周1~5 mm区域联合形成另外5个ROI,分别命名为Plus1~Plus5(图1)。对于扩展后超出肝脏边缘的部分,利用“erase”功能沿着肝脏边缘将其擦除以保证ROI位于肝内。扩展后的ROI若包含大血管,由人工手动去除。对于有多个病灶的患者,选择最大病灶进行分割。当病灶边界难以界定时,由3名医师经过讨论确定其范围。最终由1名放射科副主任医师对全部ROI进行审核。

图1.

图1

感兴趣区勾画示意图

Figure 1 Sketch map of region of interest delineation

A and B: Lesions are in arterial phase and portal venous phase (arrows). C: Region of interest is manually delineated along the visible borders of the lesion, named original region of interest. D to H: On the base of original region of interest of the tumor, different regions of interest sizes with 1-5 mm distance to tumor surface have been automatically reconstructed, named Plus1 (D), Plus2 (E), Plus3 (F), Plus4 (G), and Plus5 (H).

1.5. 图像预处理及特征提取

所有图像的体素以1 mm×1 mm×1 mm的相同尺寸重新采样。采用“Gaussian smooth filter”算法降噪。体素强度离散范围限制为64。从每一个ROI提取6种特征,包括42个直方图(histogram)特征、144个灰度共生矩阵(gray-level cooccurrence matrix,GLCM)特征、180个灰度游程矩阵(gray-level run-length matrix,GLRLM)特征、11个灰度区域大小矩阵(gray-level size zone matrix,GLSZM)特征、9个形状因子(form factor)特征和10个Haralick特征。从动脉期和门静脉期图像中各提取396个特征,总共从每名患者中提取792个特征。图像预处理和特征提取使用A.K.软件完成。

1.6. 特征筛选及建模

首先使用零-均值(Z-score)进行数据标准化以减少数据之间的差异,Kolmogorov-Smirnov检验进行方差齐性检验。然后使用t检验(方差齐)及Mann-Whitney U检验(方差不齐)筛选出MVI阳性组和MVI阴性组有显著差异的特征(P<0.05)。将选择的特征使用最小绝对收缩和选择算子(least absolute shrinkage and selection operator,LASSO)回归算法进行降维。采用10折交叉验证方法对最优正则化参数λ进行识别,以最大限度地减少平均交叉验证误差。选择取得最大曲线下面积(area under curve,AUC)值的λ作为参数,将LASSO回归模型中为非零系数的图像特征作为最具预测性的特征。将选择的特征采用支持向量机(supporter vector machine,SVM)算法建模。采用灵敏度、特异度和AUC值评估模型的性能。将Plus1~Plus5分别与OROI所构建的模型进行比较。特征选择和建立模型使用R语言(R 4.0,http://www.Rproject.org)完成。

1.7. 统计学处理

患者的一般资料使用均数±标准差( x¯ ±s)或频数表示。分类变量的比较使用χ2检验或Fisher确切概率法。连续性变量采用Kolmogorov-Smirnov检验是否符合正态分布,然后使用独立样本t检验(符合正态分布)或Mann-Whitney U检验(不符合正态分布)进行比较。以上资料的统计采用SPSS 25.0完成。R语言的“DMwR”“psych”“pROC”“glmnet”“rpart”“rpart.plot”及“e1071”用于影像组学特征处理及建模。不同模型间敏感度及特异度的比较使用SPSS软件的配对χ2检验(McNemar检验)。AUC的比较采用MedCalc 15.8软件的Delong检验。P<0.05为差异有统计学意义。

2. 结 果

2.1. 临床资料

共纳入172名患者,其中83例为MVI阳性,89例为MVI阴性。阳性组患者较阴性组年龄更小,肿瘤直径更大(均P<0.05,表1)。按7꞉3的比例随机分组后,训练组包括120名患者,其中MVI阳性58例,阴性62例;验证组包括52名患者,其中MVI阳性25例,阴性27例。训练组与验证组的一般资料差异无统计学意义(均P>0.05,表2)。

表1.

微血管侵犯阳性组与阴性组的一般资料

Table 1 Clinical characteristics of patients with or without microvascular invasion

组别 n 年龄/岁 性别/例 AFP检测例数 肿瘤直径/cm
<400 ng/mL ≥400 ng/mL
MVI阳性 83 53±11 77 6 73 10 7.25±3.63
MVI阴性 89 57±10 76 13 83 6 5.14±2.66
统计值 -2.596* 2.379† 1.433† 4.322*
P 0.010 0.123 0.231 <0.001

*t值;†χ2值;MVI:微血管侵犯;AFP:甲胎蛋白。

表2.

训练组与验证组的一般资料

Table 2 Clinical characteristics in the training and validation cohorts

组别 n MVI/例 年龄/岁 性别/例 AFP检测例数 肿瘤直径/cm
阳性 阴性 <400 ng/mL ≥400 ng/mL
训练 120 58 62 55±10 106 14 110 10 6.37±3.49
验证 52 25 27 55±12 47 5 48 4 5.65±2.90
统计值 0.001* 0.237† 0.155* 1.300†
P 0.975 0.813 0.693 1.000‡ 0.195

2值;†t值;‡Fisher确切概率法;MVI:微血管侵犯;AFP:甲胎蛋白。

2.2. 特征提取及选择

OROI、Plus1~Plus5分别有4、5、15、11、6及3个特征被选择用于建模(表3)。

表3.

不同感兴趣区所选择的特征

Table 3 Features selected from different regions of interest

感兴趣区 特征数 特征
OROI 4 GLCMEntropy_AllDirection_offset1_SD, HighGreyLevelRunEmphasis_AllDirection_offset7_SD, SurfaceArea_p, IntensityVariability_p
Plus1 5 LongRunLowGreyLevelEmphasis_angle135_offset1, LowGreyLevelRunEmphasis_angle135_offset1, ShortRunLowGreyLevelEmphasis_angle135_offset1, SurfaceArea_p, IntensityVariability_p
Plus2 15 RelativeDeviation, kurtosis, ClusterShade_angle0_offset7, GLCMEntropy_AllDirection_offset1_SD, GreyLevelNonuniformity_angle90_offset1,

LongRunLowGreyLevelEmphasis_angle135_offset4,

LongRunLowGreyLevelEmphasis_angle90_offset4, RunLengthNonuniformity_angle135_offset7,

ShortRunHighGreyLevelEmphasis_angle0_offset1, Compactness2, LowIntensitySmallAreaEmphasis, GLCMEntropy_AllDirection_offset4_SD_p, LowGreyLevelRunEmphasis_angle45_offset1_p, SurfaceArea_p, IntensityVariability_p

Plus3 11 GLCMEntropy_AllDirection_offset1_SD, GreyLevelNonuniformity_angle90_offset1, LongRunLowGreyLevelEmphasis_AllDirection_offset1, LongRunLowGreyLevelEmphasis_angle45_offset4, ShortRunEmphasis_AllDirection_offset4_SD, GLCMEntropy_AllDirection_offset4_SD_p, LowGreyLevelRunEmphasis_angle135_offset1_p, LowGreyLevelRunEmphasis_angle90_offset1_p, IntensityVariability_p, ShortRunLowGreyLevelEmphasis_angle90_offset1_p, SurfaceArea_p
Plus4 6 LongRunLowGreyLevelEmphasis_angle135_offset1, LowGreyLevelRunEmphasis_angle135_offset1, ShortRunLowGreyLevelEmphasis_angle135_offset1, Maximum3DDiameter_p, SurfaceArea_p, IntensityVariability_p
Plus5 3 Maximum3DDiameter_p, SurfaceArea_p, IntensityVariability_p

门脉期特征后缀为“_p”,动脉期特征无后缀“_p”。

2.3. 各ROI所建立模型在训练组和验证组中的预测能力

各ROI所建立的模型在训练组和验证组中的预测能力比较见表4表5。各ROI所建立的模型在训练组和验证组的ROC曲线见图2

表4.

不同模型在训练组中的预测能力

Table 4 Predictive performance of different models in the training cohort

模型 灵敏度 P* 特异度 P* AUC Z P
OROI 0.759 0.806 0.855
Plus1 0.448 <0.001 0.952 0.004 0.802 1.210 0.230
Plus2 0.948 0.001 0.935 0.008 0.979 4.730 <0.001
Plus3 0.828 0.125 0.935 0.008 0.954 3.634 <0.001
Plus4 0.483 <0.001 0.968 0.002 0.792 1.515 0.130
Plus5 0.483 <0.001 0.952 0.004 0.774 1.756 0.079

*McNemar检验,无统计值;AUC:曲线下面积。

表5.

不同模型在验证组中的预测能力

Table 5 Predictive performance of different models in the validation cohort

模型 灵敏度 P* 特异度 P* AUC Z P
OROI 0.640 0.630 0.664
Plus1 0.480 0.125 0.889 0.016 0.679 0.871 0.384
Plus2 0.440 0.500 0.630 1.000 0.536 1.818 0.069
Plus3 0.880 0.031 0.926 0.008 0.903 4.291 <0.001
Plus4 0.440 0.063 0.963 0.004 0.708 0.795 0.427
Plus5 0.440 0.063 0.926 0.008 0.757 1.081 0.280

*McNemar检验,无统计值;AUC:曲线下面积。

图2.

图2

各模型在训练组和验证组预测肝细胞癌有无微血管侵犯的ROC曲线

Figure 2 ROC curves of each model for microvascular invasion prediction in hepatocellular carcinoma in the training and validation cohorts

A: In the training cohort; B: In the validation cohort. ROC: Receiver operating characteristic; OROI: Original region of interest; AUC: Area under the curve.

在训练组中,Plus2的灵敏度高于OROI(P=0.001),Plus1、Plus4及Plus5的灵敏度均低于OROI(均P<0.001),Plus3的灵敏度与OROI差异无统计学意义(P>0.05)。所有ROI的特异度均高于OROI(均P<0.05)。Plus2及Plus3的AUC高于OROI(均P<0.001),其余ROI的AUC与OROI差异无统计学意义(均 P>0.05)。

在验证组中,Plus3的灵敏度高于OROI(P=0.031),其余ROI的灵敏度与OROI差异无统计学意义(均P>0.05)。Plus1(P=0.016)、Plus3(P=0.008)、Plus4 (P=0.004)及Plus5(P=0.008)的特异度高于OROI,Plus2的特异度与OROI差异无统计学意义(P=1.000)。Plus3的AUC显著高于OROI(P<0.001),其余ROI的AUC与OROI差异无统计学意义(均P>0.05)。

3. 讨 论

本研究采用6个不同的ROI范围分别建模预测HCC有无MVI,以探讨ROI范围对模型预测效能的影响。结果显示将肿瘤外3 mm内的肝组织包含在ROI内可提高模型的预测效能。

本研究联合动脉期和门脉期数据建模的原因如下:首先,MVI主要见于门静脉的小分支,也可见于肝动脉的小分支,并且CT增强扫描的数据能反映肿瘤血供上的异质性。其次,既往的相关研究在扫描期相的选择上存在一定的争议。MA等[9]认为门脉期的影像组学特征在预测效能上优于动脉期及延迟期,而Zhang等[12]选择延迟期的影像组学特征建模。本研究不关心期相对预测效能的影响,因此联合了动脉期和门脉期特征共同建模,这也与Peng等[16]及Meng等[17]的研究方法一致。

在本研究6个不同的ROI构建的预测模型中,某些影像组学特征出现频率较高,表明这些特征在预测HCC是否发生MVI中发挥更重要的作用。例如,“SurfaceArea”属于形状因子特征,反映肿瘤的大小,“SurfaceArea”数值越大说明肿瘤越大。本研究结果与既往研究[18]结果一致,即HCC体积越大,发生MVI的概率越高。“IntensityVariability”反映灰度变化范围。“GLCMEntropy”属于灰度共生矩阵特征,共生矩阵特征反映具有某灰度值的像素对的空间关系,“GLCMEntropy”数值越大,说明图像的复杂程度越高。“GreyLevelNonuniformity”属于灰度游程矩阵特征,游程矩阵特征反映具有某灰度值的像素于既定方向上连续出现的频数,“GreyLevelNonuniformity”数值越大,说明灰度越不均匀。恶性程度越高的HCC细胞间的差异越大,越容易发生缺血坏死,且更容易突破包膜向周围浸润生长,这些都导致肿瘤内部及瘤周异质性增大,在影像上则表现为图像灰度不均匀性增大,图像复杂程度增高。本研究结果与既往研究[19]结果一致,即HCC恶性程度越高,发生MVI的可能性越大。

本研究对HCC的影像组学特征提取是基于三维空间的三维图像。与基于肿瘤最大层面的二维图像特征相比,三维图像特征更加完整,能更加全面地反映肿瘤的异质性[20]。考虑到肿瘤的实际边界大于影像图像上的肿瘤边界,且MVI主要发生在瘤周区域[21-23],本研究在OROI的基础上向外扩展1~5 mm,确定6个不同的ROI范围。结果显示,将肿瘤边界外3 mm的肝组织纳入研究范围的Plus3具有最佳的预测效能,在训练组和验证组的AUC分别达0.954及0.903,显著高于OROI的AUC。由肿瘤周围实质构成的肿瘤微环境包含丰富的生物学信息[24],影像组学方法能捕捉裸眼无法识别的肿瘤病理生理及微环境的细微变化,推测这是Plus3模型预测效能较OROI提高的病理生理基础。此外,OROI、Plus1及Plus2在训练组中表现出很好的预测效能,但是在验证组中表现不佳,说明模型出现了明显的过拟合现象,考虑与本研究的病例数偏少有关。当将肿瘤边界外 4 mm及5 mm的肝组织纳入研究范围时,Plus4及Plus5用于建模的特征数量较OROI减少,这可能是由于一些有预测价值的特征受到正常肝组织的干扰变成了冗余特征所造成。

本研究存在以下局限性:1)为单中心回顾性研究,样本量较小,结果有待多中心、大样本研究进一步证实。2)患者的CT图像来自两台不同的机器,不同的扫描参数可能会对图像产生一定的影响,进而影响最终的结果。3)未做组间和组内一致性分析,这可能导致特征选择的不稳定。4)仅将瘤内和瘤周区域进行整体分割,而未对瘤内和瘤周区域进行单独分割,再联合两个独立区域的特征建模,在今后的研究中需要进一步比较两种分割方式对模型预测效能的影响。

综上所述,在基于增强CT的影像组学预测HCC有无MVI的模型中,ROI范围对模型的预测能力有显著影响,将肿瘤外一定范围的肝组织包含在ROI内可以提高模型的预测效能,向瘤外扩展3 mm可能为最佳扩展范围。

基金资助

湖南省自然科学基金(2021JJ40951)。

This work was supported by the Natural Science Foundation of Hunan Province, China (2021JJ40951).

利益冲突声明

作者声称无任何利益冲突。

作者贡献

赵华飞 研究设计,数据收集,数据处理和论文撰写;冯智超 研究设计;李慧玲、姚山虎 数据收集;郑薇 论文撰写和审校;容鹏飞 研究设计与指导。所有作者阅读并同意最终的文本。

原文网址

http://xbyxb.csu.edu.cn/xbwk/fileup/PDF/2022081049.pdf

参考文献

  • 1. McGlynn KA, Petrick JL, El-Serag HB. Epidemiology of hepatocellular carcinoma[J]. Hepatology, 2021, 73(Suppl 1): 4-13. 10.1002/hep.31288. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2. Colombo M. Hepatocellular carcinoma[J]. J Hepatol, 1992, 15(1/2): 225-236. 10.1016/0168-8278(92)90041-M. [DOI] [PubMed] [Google Scholar]
  • 3. Schlichtemeier SM, Pang TC, Williams NE, et al. A pre-operative clinical model to predict microvascular invasion and long-term outcome after resection of hepatocellular cancer: the Australian experience[J]. Eur J Surg Oncol, 2016, 42(10): 1576-1583. 10.1016/j.ejso.2016.05.032. [DOI] [PubMed] [Google Scholar]
  • 4. Gouw ASH, Balabaud C, Kusano H, et al. Markers for microvascular invasion in hepatocellular carcinoma: where do we stand? [J]. Liver Transpl, 2011, 17(Suppl 2): S72-S80. 10.1002/lt.22368. [DOI] [PubMed] [Google Scholar]
  • 5. Lee S, Kim SH, Lee JE, et al. Preoperative gadoxetic acid-enhanced MRI for predicting microvascular invasion in patients with single hepatocellular carcinoma[J]. J Hepatol, 2017, 67(3): 526-534. 10.1016/j.jhep.2017.04.024. [DOI] [PubMed] [Google Scholar]
  • 6. Hong SB, Choi SH, Kim SY, et al. MRI features for predicting microvascular invasion of hepatocellular carcinoma: a systematic review and meta-analysis[J]. Liver Cancer, 2021, 10(2): 94-106. 10.1159/000513704. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7. Min JH, Lee MW, Park HS, et al. Interobserver variability and diagnostic performance of gadoxetic acid-enhanced MRI for predicting microvascular invasion in hepatocellular carcinoma[J]. Radiology, 2020, 297(3): 573-581. 10.1148/radiol.2020201940. [DOI] [PubMed] [Google Scholar]
  • 8. Kim N, Kim HK, Lee K, et al. Single-cell RNA sequencing demonstrates the molecular and cellular reprogramming of metastatic lung adenocarcinoma[J]. Nat Commun, 2020, 11(1): 2285. 10.1038/s41467-020-16164-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9. Ma XH, Wei JW, Gu DS, et al. Preoperative radiomics nomogram for microvascular invasion prediction in hepatocellular carcinoma using contrast-enhanced CT[J]. Eur Radiol, 2019, 29(7): 3595-3605. 10.1007/s00330-018-5985-y. [DOI] [PubMed] [Google Scholar]
  • 10. Xu X, Zhang HL, Liu QP, et al. Radiomic analysis of contrast-enhanced CT predicts microvascular invasion and outcome in hepatocellular carcinoma[J]. J Hepatol, 2019, 70(6): 1133-1144. 10.1016/j.jhep.2019.02.023. [DOI] [PubMed] [Google Scholar]
  • 11. 汪禾青, 曾蒙苏, 饶圣祥, 等. 常规MRI图像影像组学评估肝细胞癌微血管侵犯的价值[J]. 中华放射学杂志, 2019, 53(4): 292-298. 10.3760/cma.j.issn.1005?1201.2019.04.010. [DOI] [Google Scholar]; WANG Heqing, ZENG Mengsu, RAO Shengxiang, et al. Radiomic features to predict microvascular invasion in hepatocellular carcinoma based on conventional MRI: preliminary findings[J]. Chinese Journal of Radiology, 2019, 53(4): 292-298. 10.3760/cma.j.issn.1005?1201.2019.04.010. [DOI] [Google Scholar]
  • 12. Zhang XM, Ruan SJ, Xiao WB, et al. Contrast-enhanced CT radiomics for preoperative evaluation of microvascular invasion in hepatocellular carcinoma: a two-center study[J/OL]. Clin Transl Med, 2020, 10(2): e111 [2022-01-11]. 10.1002/ctm2.111. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13. Feng ST, Jia YM, Liao B, et al. Preoperative prediction of microvascular invasion in hepatocellular cancer: a radiomics model using Gd-EOB-DTPA-enhanced MRI[J]. Eur Radiol, 2019, 29(9): 4648-4659. 10.1007/s00330-018-5935-8. [DOI] [PubMed] [Google Scholar]
  • 14. Zheng J, Chakraborty J, Chapman WC, et al. Preoperative prediction of microvascular invasion in hepatocellular carcinoma using quantitative image analysis[J/OL]. J Am Coll Surg, 2017, 225(6): 778-788. e1 [2022-01-11]. 10.1016/j.jamcollsurg.2017.09.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15. 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)[J]. 中华肝脏病杂志, 2015, 23(5): 321-327. [Google Scholar]; Chinese Society of Liver Cancer, Chinese Anti-Cancer Association, Liver Cancer Study Group, Chinese Society of Hepatology, Chinese Medical Association, Chinese Society of Pathology, Chinese Anti-Cancer Association, et al. Evidence-based practice guidelines for standardized pathological diagnosis of primary liver cancer in China: 2015 update[J]. Chinese Journal of Hepatology, 2015, 23(5): 321-327. 10.3760/cma.j.issn.1007-3418.2015.05.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16. Peng J, Zhang J, Zhang QF, et al. A radiomics nomogram for preoperative prediction of microvascular invasion risk in hepatitis B virus-related hepatocellular carcinoma[J]. Diagn Interv Radiol, 2018, 24(3): 121-127. 10.5152/dir.2018.17467. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17. Meng XP, Tang TY, Ding ZM, et al. Preoperative microvascular invasion prediction to assist in surgical plan for single hepatocellular carcinoma: better together with radiomics[J]. Ann Surg Oncol, 2022, 29(5): 2960-2970. 10.1245/s10434-022-11346-1. [DOI] [PubMed] [Google Scholar]
  • 18. Lei ZQ, Li J, Wu D, et al. Nomogram for preoperative estimation of microvascular invasion risk in hepatitis B virus-related hepatocellular carcinoma within the Milan criteria[J]. JAMA Surg, 2016, 151(4): 356-363. 10.1001/jamasurg.2015.4257. [DOI] [PubMed] [Google Scholar]
  • 19. Erstad DJ, Tanabe KK. Prognostic and therapeutic implications of microvascular invasion in hepatocellular carcinoma[J]. Ann Surg Oncol, 2019, 26(5): 1474-1493. 10.1245/s10434-019-07227-9. [DOI] [PubMed] [Google Scholar]
  • 20. 邓娇, 谭一雄, 顾潜彪, 等. CT影像组学对原发性胃淋巴瘤与Borrmann Ⅳ型胃癌的鉴别诊断价值[J]. 中南大学学报(医学版), 2019, 44(3): 257-263. [DOI] [PubMed] [Google Scholar]; DENG Jiao, TAN Yixiong, GU Qianbiao, et al. Application of CT-based radiomics in differentiating primary gastric lymphoma from Borrmann type Ⅳ gastric cancer[J]. Journal of Central South University. Medical Science, 2019, 44(3): 257-263. [DOI] [PubMed] [Google Scholar]
  • 21. Kelsey CR, Schefter T, Nash SR, et al. Retrospective clinicopathologic correlation of gross tumor size of hepatocellular carcinoma: implications for stereotactic body radiotherapy[J]. Am J Clin Oncol, 2005, 28(6): 576-580. 10.1097/01.coc.0000184657.65679.6f. [DOI] [PubMed] [Google Scholar]
  • 22. Wang MH, Ji Y, Zeng ZC, et al. Impact factors for microinvasion in patients with hepatocellular carcinoma: possible application to the definition of clinical tumor volume[J]. Int J Radiat Oncol Biol Phys, 2010, 76(2): 467-476. 10.1016/j.ijrobp.2009.01.057. [DOI] [PubMed] [Google Scholar]
  • 23. Chen HY, Ma XM, Ye M, et al. CT-pathologic correlation in primary hepatocellular carcinoma: an implication for target delineation[J]. J Radiat Res, 2013, 54(5): 938-942. 10.1093/jrr/rrt030. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24. Gu L, Mooney DJ. Biomaterials and emerging anticancer therapeutics: engineering the microenvironment[J]. Nat Rev Cancer, 2016, 16(1): 56-66. 10.1038/nrc.2015.3. [DOI] [PMC free article] [PubMed] [Google Scholar]

Articles from Journal of Central South University Medical Sciences are provided here courtesy of Central South University

RESOURCES