分享:
分享到微信朋友圈
X
临床研究
HRT2WI联合DWI影像组学对直肠癌固有肌层突破的诊断价值
盛芳婷 田为中 冯泽萌

Cite this article as: SHENG F T, TIAN W Z, FENG Z M. The diagnostic value of radiomics based on HRT2WI and DWI in the breakthrough of the muscularis propria layer of rectal cancer[J]. Chin J Magn Reson Imaging, 2023, 14(4): 102-106, 131.本文引用格式:盛芳婷, 田为中, 冯泽萌. HRT2WI联合DWI影像组学对直肠癌固有肌层突破的诊断价值[J]. 磁共振成像, 2023, 14(4): 102-106, 131. DOI:10.12015/issn.1674-8034.2023.04.017.


[摘要] 目的 评估基于高分辨率T2加权成像(high-resolution T2-weighted imaging, HRT2WI)及扩散加权成像(diffusion-weighted imaging, DWI)的影像组学模型对于直肠癌是否突破固有肌层的诊断价值。材料与方法 回顾性分析2019年1月至2021年12月在南京医科大学附属泰州人民院经手术病理证实且术前行3.0 T磁共振HRT2WI序列及DWI(b值为800 s/mm2)序列扫描的直肠癌患者资料。根据病理T分期结果将T1和T2期患者归为未突破肌层组,T3和T4期患者归为突破肌层组。在图像上手动勾画病灶三维感兴趣区(volume of interest, VOI)后提取影像组学特征,之后采用独立样本t检验及支持向量机(support vector machine, SVM)线性核函数进行特征的选择及降维,选择出有价值的影像组学特征。将样本按7∶3的比例随机分为训练集与验证集进行机器学习,构建SVM分类器模型,获取训练集和验证集的受试者工作特征(receiver operating characteristic, ROC)曲线下面积(area under the curve, AUC)、敏感度、特异度及准确度,以评估不同模型对于直肠癌固有肌层突破的诊断效能。通过DeLong检验比较不同模型的AUC差异。结果 从每例患者肿瘤组织的HRT2WI及DWI图像中各提取出1142个影像组学特征,并分别通过独立样本t检验及SVM线性核函数筛选特征,根据HRT2WI影像组学特征构建的SVM模型验证组AUC值为0.894,敏感度为90.0%,特异度为70.6%;根据DWI影像组学特征构建的SVM模型验证组AUC值为0.774,敏感度为60.0%,特异度为76.5%;联合HRT2WI及DWI影像组学特征构建的SVM模型诊断效能明显更优,验证组AUC值为0.927,敏感度为80.0%,特异度为88.2%。DeLong检验显示联合模型与单独序列模型预测效能存在显著性差异(P<0.05)。结论 利用联合直肠HRT2WI及DWI影像组学特征构建影像组学模型对于直肠癌突破固有肌层有一定的诊断效能,可为临床个体化治疗提供帮助。
[Abstract] Objective To evaluate the diagnostic value of radiomics models based on high-resolution T2-weighted imaging (HRT2WI) and diffusion-weighted imaging (DWI) in the breakthrough of the muscularis propria of rectal cancer.Materials and Methods A retrospective analysis was performed on rectal cancer patients who underwent preoperative 3.0 T MRI scans including HRT2WI and DWI (b value of 800 s/mm2), and were confirmed by surgical pathology at Taizhou People's Hospital affiliated of Nanjing Medical University from January 2019 to December 2021. Patients with T1 and T2 staging were classified as the non-breakthrough group, and those with T3 and T4 staging were classified as the breakthrough group based on pathological staging. Radiomics features were extracted after manually delineating the volume of interest (VOI) on the lesion, and then independent sample t-tests and support vector machine (SVM) with a linear kernel were used for feature selection and dimensionality reduction, respectively, to select valuable radiomics features. The selected samples were randomly divided into training and validation sets at a ratio of 7∶3 for machine learning to build the SVM classifier model. The receiver operating characteristic (ROC) curve was used to evaluate the diagnostic performance of different models in terms of the area under the curve (AUC), sensitivity, specificity, and accuracy for detecting rectal cancer invasion beyond the muscularis propria. The DeLong test was used to compare the differences in AUC between different models.Results A total of 1142 radiomics features were extracted from the HRT2WI and DWI images of each patient's tumor tissue and screened by independent sample t-tests and SVM with a linear kernel. The SVM model constructed based on the radiomics features of HRT2WI images had a validation AUC value of 0.894, sensitivity of 90.0%, and specificity of 70.6%. The SVM model constructed based on the radiomics features of DWI images had a validation AUC value of 0.774, sensitivity of 60.0%, and specificity of 76.5%. The final predictive model combining HRT2WI and DWI had significantly better diagnostic performance than other models, with a validation AUC value of 0.927, sensitivity of 80.0%, and specificity of 88.2%. The DeLong test showed significant differences in predictive performance between the combined model and the single sequence models (P<0.05).Conclusions The radiomics model combining HRT2WI and DWI can effectively evaluate the breakthrough of the muscularis propria of rectal cancer, which may provide assistance for individualized clinical treatment.
[关键词] 直肠癌;磁共振成像;影像组学;固有肌层;诊断效能
[Keywords] rectal cancer;magnetic resonance imaging;radiomics;muscularis propria;diagnostic performan

盛芳婷 1   田为中 2*   冯泽萌 3  

1 大连医科大学研究生院,大连 116000

2 南京医科大学附属泰州人民医院影像科,泰州 225300

3 南京工业大学柔性电子(未来技术)学院,南京 210000

通信作者:田为中,E-mail:jstztwz@163.com

作者贡献声明:田为中设计本研究的方案,对稿件重要的内容进行了修改;盛芳婷起草和撰写稿件,获取、分析或解释本研究的数据;冯泽萌获取、分析或解释本研究的数据,对稿件重要的内容进行了修改;全体作者都同意发表最后的修改稿,同意对本研究的所有方面负责,确保本研究的准确性和诚信。


收稿日期:2022-11-15
接受日期:2023-04-07
中图分类号:R445.2  R735.37 
文献标识码:A
DOI: 10.12015/issn.1674-8034.2023.04.017
本文引用格式:盛芳婷, 田为中, 冯泽萌. HRT2WI联合DWI影像组学对直肠癌固有肌层突破的诊断价值[J]. 磁共振成像, 2023, 14(4): 102-106, 131. DOI:10.12015/issn.1674-8034.2023.04.017.

0 前言

       结直肠癌是我国最常见的恶性肿瘤之一。近些年来,我国结直肠癌的发病率和死亡率持续处于上升趋势[1]。因此,直肠癌的早期诊断与预后预测对保证我国人民的生命健康具有重要意义。由于早期直肠癌的隐匿发病,大多数患者在首次诊断时已经处于局部晚期阶段[2]。手术切除是早期直肠癌(T1~T2期)最常见的治疗方法[3],而局部晚期(T3~T4期)患者则推荐在术前进行新辅助放化疗(neoadjuvant chemoradiation, NCRT)[4, 5]。直肠癌是否突破固有肌层是影响患者预后的重要因素。由于肿瘤周围的炎症反应与突破肌层的表现类似,因此在影像学上不易区分[6]。影像组学是指应用大数据挖掘等技术,从医学图像中提取肉眼无法观察的高通量特征,建立与癌症对应的预测模型,提供准确的风险分层[7, 8, 9]。以往的研究表明,扩散加权成像(diffusion-weighted imaging, DWI)在直肠癌的定性评估中具有一定的作用[10]。高分辨率T2加权成像(high-resolution T2-weighted imaging, HRT2WI)是原发性直肠癌MRI评价的关键序列[11]。以往相关的研究大多是基于单一序列或只对图像进行了纹理分析[12, 13, 14],而缺乏基于多序列建立影像组学模型的研究。本研究拟基于直肠癌患者术前HRT2WI及DWI序列图像的影像组学特征构建支持向量机(support vector machine, SVM)模型,以术前评估直肠癌是否突破固有肌层,从而为患者提供适合的治疗方案,改善预后。

1 材料与方法

1.1 研究对象

       本研究遵守《赫尔辛基宣言》,通过南京医科大学附属泰州人民医院伦理委员会批准,免除受试者知情同意,批准文号:KY2022-148-01。回顾性分析2019年1月至2021年12月在南京医科大学附属泰州人民医院3.0 T MRI设备上进行直肠扫描并通过病理确诊为直肠癌的患者临床资料及影像学资料,所有资料均通过医院病历系统、影像归档和通信系统获得。纳入标准:(1)已确诊为直肠癌;(2)术前两周内行直肠MRI检查,扫描序列齐全;(3)病例资料完整。排除标准:(1)图像质量差,如明显的伪影导致病灶显示欠清,无法准确勾画三维感兴趣区(volume of interest, VOI);(2)未在我院接受直肠癌根治术;(3)患者手术前、MRI检查前后接受过任何的全身或局部治疗。

1.2 检查方法

       患者术前使用德国Siemens Skyra 3.0 T MRI仪进行检查,使用8通道体部相控阵线圈。检查前4 h禁食禁水以防止增强扫描时出现胃肠道反应。扫描前需排尿、排便,尽可能做灌肠准备。检查前应尽量避免其他直肠检查如直肠腔内超声、肠镜检查间隔进行,防止肠管激惹。考虑到直肠壁填充扩张可能会干扰肿瘤和直肠系膜筋膜之间的距离,因此不推荐使用。48 h内不行任何其他对比增强影像检查,嘱患者进行屏气呼气训练。患者取仰卧位,平静呼吸。扫描范围从结直肠末段至肛门。HRT2WI序列扫描参数:TR 8840 ms,TE 103 ms,FOV 220 mm×220 mm,矩阵240×240,层厚3 mm,层间距0.6 mm。DWI序列扫描参数:TR 7490 ms,TE 60 ms,FOV 252 mm×252 mm,矩阵144×144,层厚4 mm,层间距0.8 mm,b值为800 s/mm2

1.3 图像分析

       由两名分别具有3年及5年工作经验的腹部影像主治医师各在ITK SNAP软件上以盲法勾画VOI,沿着肿瘤的边缘逐层勾画,避开管腔内容物及周围脂肪组织。勾画结果如图1所示。根据第七届美国癌症联合委员会(American Joint Committee On Cancer, AJCC)的TNM分期标准,将T分期定义如下:T1期,肿瘤局限于黏膜下层;T2期,肿瘤延伸到固有肌层但不超过固有肌层;T3期,肿瘤突破固有肌层进入直肠系膜脂肪层;T4期,肿瘤侵犯周围结构或器官。

图1  三维感兴趣体积(VOI)勾画示意图。1A与1B分别从高分辨率T2WI及扩散加权成像(DWI)图像上逐层勾画肿瘤边缘,避开管腔内容物。
Fig. 1  Sketch of volumes of interest. 1A and 1B outline the tumor margins layer by layer from the patient's high-resolution T2WI and diffusion-weighted imaging images, avoiding the lumen contents.

1.4 病理诊断

       根据AJCC的TNM分期标准,对所有手术切除标本进行病理诊断并分期。根据病理结果,将T1~T2期归为未突破肌层组,T3~T4期归为突破肌层组。以病理结果为准进行模型的训练与验证。

1.5 影像组学特征的提取

       勾画VOI后,将DICOM图像和VOI输出,利用Pyradiomics软件进行影像组学特征提取。分别从每例患者肿瘤组织HRT2WI及DWI图像的VOI中提取了1142个特征,包括:(1)一阶统计(first order)特征;(2)形态(shape)特征;(3)二阶及高阶组学特征,包括灰度依赖矩阵(gray level dependence matrix, GLDM)特征、灰度共生矩阵(gray level co-occurence matrix, GLCM)特征、灰度游程矩阵(gray level run length matrix, GLRLM)特征、灰度大小区域矩阵(gray level size zone matrix, GLSZM)特征。对于除形状以外的特征进行小波变换及拉普拉斯高斯变换。采用组内相关系数(intra-class correlation coefficients, ICC)比较两位医师所勾画的特征数据,当ICC≥0.75时,则认为该影像组学特征具有一定的可靠性。

1.6 影像组学分析

       利用Python 3.7软件对于提取的影像组学特征进行分析。(1)数据的预处理:去除异常值,对输出值进行最大最小归一化,以统一所有特征的尺度;(2)特征筛选:利用独立样本t检验方法进行特征的筛选;(3)特征权重计算:采取相关性系数法提取权重,对提取出的特征进行权重计算,评估每个特征在分类任务中的重要性;(4)特征选择与降维:经过多次十折交叉验证后取平均值,去除冗余特征后,利用SVM的最大间隔特性,通过SVM核函数来对特征进行选择;(5)影像组学特征分析:选择特征权重较高(独立样本t检验中P值≤0.001)的组学特征进行比较分析,对比其在不同分组中的差异;(6)将特征降维所获得的影像组学特征与术后T分期构建SVM模型。随机选取70%病例作为训练集进行模型训练,并采用十折交叉验证方式进行验证以得出稳定的结果,利用剩余的30%病例作为验证集进行验证。使用受试者工作特征(receiver operating characteristic, ROC)曲线下面积(area under the curve, AUC)、敏感度、特异度及敏感度评估影像组学模型的预测能力。

1.7 统计学分析

       采用SPSS 26.0软件进行统计学分析。采用Kolmogorov-Smirnov方法对计量资料进行正态性检验,符合正态分布的计量资料以均值±标准差(x¯±s)表示,采用独立样本t检验比较数据间的差异性,统计量为t值。计数资料以例数表示,两组间比较采用卡方检验进行分析。P<0.05为差异有统计学意义。运用DeLong检验比较模型间的差异性。

2 结果

2.1 一般资料

       研究初始纳入140例患者病例,因图像存在伪影排除11例、未行根治性切除术排除6例。最终本次研究纳入123例患者病例,其中:男76名(61.8%),女47名(38.2%);年龄范围45~87岁,平均年龄67岁;高位直肠癌22例(17.9%),中位直肠癌56例(45.5%),低位直肠癌45例(36.6%);病理结果显示未突破肌层组(T1~T2期)57例(46.3%),突破肌层组(T3~T4期)66例(53.7%)。纳入对象被按照7∶3的比例随机分为训练组和验证组,训练组86例(T1~T2:40例,T3~T4:46例),验证组37例(T1~T2:17例,T3~T4:20例)。两组的性别、年龄及肿瘤位置差异均无统计学意义(P>0.05,表1)。

表1  患者临床资料
Tab. 1  Clinical characteristics of the patients

2.2 影像组学特征分析

       经过独立样本t检验分别从HRT2WI及DWI中筛选了41、83个特征,影像组学特征的显著性分析结果显示HRT2WI中提取的shape_Maximum2DdiameterSlice [最大2D直径(切片)]、shape_LeastAxisLength(最短轴长度)、shape_Maximum2DdiameterColumn [最大2D直径(列)]、glcm_JointEntropy(联合熵)、glcm_InverseVariance(逆方差)、gldm_SmallDependenceHighGrayLevelEmphasis(小依赖高灰度强调)等影像组学特征在突破肌层与未突破肌层组具有显著性差异;DWI中提取的shape_MinorAxisLength(第二大轴长度),shape_Maximum2DdiameterRow [最大2D直径(行)]、glcm_JointEntropy、gldm_DependenceEntropy(依赖熵)、glszm_LowGrayLevelZoneEmphasis(低灰度区域强调)等影像组学特征在突破肌层组与未突破肌层组中具有显著性差异(表2)。

表2  显著性影像组学特征分析
Tab. 2  Analysis of significant texture features extracted

2.3 联合模型建立及评估

       通过SVM分别建立了基于HRT2WI、DWI及联合二者影像组学特征的影像组学模型,三种模型在训练集中的AUC分别为0.904,0.880,0.943,在验证集中的AUC分别为0.894、0.774、0.927(表3)。验证集的DeLong检验结果显示,联合HRT2WI、DWI的影像组学模型与单序列模型预测效能差异具有统计学意义(P<0.05,图2)。

图2  不同模型验证组间受试者工作特征曲线(ROC)对比分析。HRT2WI表示基于高分辨T2WI图像的模型;DWI表示基于扩散加权成像图像的模型,HRT2WI+DWI表示二者联合模型。
Fig. 2  Comparative analysis for receiver operating characteristic curves of validation groups among different models. HRT2WI represents a model based on high-resolution T2WI images; DWI represents a model based on diffusion weighted imaging images, HRT2WI+DWI represents a joint model of the two.
表3  三种模型预测效能
Tab. 3  Prediction efficiency of the three models

3 讨论

       本研究通过对123例未突破肌层及突破肌层的直肠癌患者病例进行回顾性分析,联合HRT2WI及DWI序列构建影像组学模型,探究其对于评估直肠癌突破固有肌层的价值。在以往的研究中,有学者选择最大相关最小冗余(maximum relevance minimum redundancy, mRMR)算法和最小绝对值收缩与选择算子(least absolute shrinkage and selection operator, LASSO)算法进行组学特征降维筛选[15],但其更适用于大规模的数据集和高维特征。本研究采用的机器学习样本空间小,经过测试发现独立样本t检验获得的结果优于LASSO和mRMR等方法。本研究结果表明,未突破肌层组与突破肌层组的影像组学特征具有显著性差异,联合HRT2WI及DWI影像组学模型对于评估直肠癌是否突破固有肌层具有一定的诊断价值,弥补了以往单一序列评估直肠癌T分期的不足。该联合模型的准确度高于放射科医生肉眼观察评估直肠癌突破肌层的准确度,具有一定的客观性,能够帮助临床制订决策,改善患者预后。本研究创新地将不同序列的影像组学特征结合在一起,更加全面地分析基于不同序列的影像组学模型对评估肿瘤突破肌层的价值,为临床术前分期提供了新方法,对于患者的个性化治疗具有一定的价值。

3.1 联合HRT2WI及DWI影像组学的优势

       传统的HRT2WI及DWI序列对于直肠癌术前分期具有一定的局限性,这是因为直肠周围结缔组织增生性纤维化反应与肿瘤穿透直肠肌壁相似,模糊的肿瘤边界会导致分期存在困难[16]。影像组学可以挖掘肉眼无法识别的潜在信息,量化肿瘤的异质性以评估患者的病情进展。近年来,也有许多专家与学者们通过影像组学模型对直肠癌的分期及病理分级进行了大量的研究。萨莎等[17]基于CT的结直肠癌图像及临床资料建立了随机森林模型,结果得出基于影像组学模型预测T分期比传统的影像方法准确率更高(准确率达80.7%)。然而CT对于软组织的分辨率具有一定的局限性,其评估肿瘤的浸润肌层程度不如MRI准确。MA等[18]基于HRT2WI序列提取了不同TN分期的直肠癌患者影像组学特征,并分别建立了SVM、决策树(decision tree, DT)、随机森林(random forest, RF)等机器学习模型,发现基于HRT2WI的影像组学能有效地预测直肠癌的TN分期,但其只分析了单一的HRT2WI序列。DWI可以评估组织内水分子扩散情况。当肿瘤分化程度低时,水分子的运动受到阻碍,会导致DWI上肿瘤信号改变。因此联合HRT2WI及DWI序列能多方位评估肿瘤的信号表现,获取更多肿瘤病理图像特征。SVM是一种二分类的广义线性模型,具有较好的鲁棒性和泛化能力[19, 20, 21, 22]。相比于DT和逻辑回归模型,SVM能更有效地解决一系列的非线性问题,具有识别更深层次模式的能力[23]。本研究基于从HRT2WI及DWI中提取的影像组学特征,采用SVM分别建立HRT2WI、DWI及联合HRT2WI、DWI三种模型,用于无创性评估直肠癌患者是否突破固有肌层,三种模型在训练集和验证集中均取得较好效能,联合HRT2WI、DWI模型具有更好的预测效能。

3.2 HRT2WI及DWI影像组学特征分析

       本研究中不同分组的影像组学特征差异具有统计学意义。影像组学特征分析可以通过定量的方法提取图像的像素特征,通过病灶的灰度及规律性等反映肿瘤的微环境[24, 25, 26, 27]。影像组学特征可用于分析肿瘤内部的复杂性及异质性[28, 29, 30]。本研究从HRT2WI、DWI中提取的shape、GLDM、GLCM、GLSZM影像组学特征可分别反映肿瘤的形状、灰度的依赖性、灰度分布的均匀性等信息,在突破肌层及未突破肌层组中具有显著的差异。其中,GLCM特征具有重要的意义。本研究中的glcm_JointEntropy即联合熵,用来衡量邻域强度值的随机性或可变性。突破肌层组直肠癌的联合熵值比未突破组大,表明当肿瘤突破固有肌层时,其灰度强度分布越不均匀,提示晚期肿瘤的坏死及囊变区域较多。此前也有研究结果表明,基于TW2I及DWI的纹理分析对于评估直肠癌的术前分期具有一定的意义[31, 32, 33]

3.3 本研究的局限性

       本研究仍然存在一些不足。首先,本研究的病例数量较少,来源单一,缺乏来自于多中心的外部验证;其次,对于病灶的勾画虽已尽量避开管腔及坏死区域,然而仍然无法避免一定的误差;再次,本研究的序列存在一定的局限,只纳入了HRT2WI及DWI序列,未涉及到增强序列;最后,应尽可能地将临床生化指标与影像资料结合起来,以此来提高模型的预测效能。

4 结论

       综上所述,联合HRT2WI及DWI序列的影像组学模型对于评估直肠癌固有肌层突破具有良好的诊断效能,可以无创性地评估直肠癌术前分期,给患者提供个性化的治疗,帮助患者改善预后,辅助精准医疗。

[1]
中国结直肠癌诊疗规范(2020年版)专家组. 国家卫生健康委员会中国结直肠癌诊疗规范(2020年版)[J]. 中华胃肠外科杂志, 2020, 23(6): 521-540. DOI: 10.3760/cma.j.cn.441530-20200520-00289.
Chinese knotting Colorectal Cancer Diagnosis and Treatment Standards (2020 Edition) Expert Group. National health and wellness committee's diagnostic and therapeutic guidelines for colorectal cancer in China (2020 edition)[J]. Chin J Gastrointest Surg, 2020, 23(6): 521-540. DOI: 10.3760/cma.j.cn.441530-20200520-00289.
[2]
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(6): 394-424. DOI: 10.3322/caac.21492.
[3]
DESANTIS C E, LIN C C, MARIOTTO A B, et al. Cancer treatment and survivorship statistics, 2014[J]. CA Cancer J Clin, 2014, 64(4): 252-271. DOI: 10.3322/caac.21235.
[4]
GOLDENBERG B A, HOLLIDAY E B, HELEWA R M, et al. Rectal cancer in 2018: a primer for the gastroenterologist[J]. Am J Gastroenterol, 2018, 113(12): 1763-1771. DOI: 10.1038/s41395-018-0180-y.
[5]
SEBAG-MONTEFIORE D, STEPHENS R J, STEELE R, et al. Preoperative radiotherapy versus selective postoperative chemoradiotherapy in patients with rectal cancer (MRC CR07 and NCIC-CTG C016): a multicentre, randomised trial[J]. Lancet, 2009, 373(9666): 811-820. DOI: 10.1016/S0140-6736(09)60484-0.
[6]
KIM H, LIM J S, CHOI J Y, et al. Rectal cancer: comparison of accuracy of local-regional staging with two- and three-dimensional preoperative 3-T MR imaging[J]. Radiology, 2010, 254(2): 485-492. DOI: 10.1148/radiol.09090587.
[7]
STANZIONE A, VERDE F, ROMEO V, et al. Radiomics and machine learning applications in rectal cancer: current update and future perspectives[J]. World J Gastroenterol, 2021, 27(32): 5306-5321. DOI: 10.3748/wjg.v27.i32.5306.
[8]
COPPOLA F, GIANNINI V, GABELLONI M, et al. Radiomics and magnetic resonance imaging of rectal cancer: from engineering to clinical practice[J/OL]. Diagnostics (Basel), 2021, 11(5): 756 [2022-09-29]. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8146913. DOI: 10.3390/diagnostics11050756.
[9]
CHETAN M R, GLEESON F V. Radiomics in predicting treatment response in non-small-cell lung cancer: current status, challenges and future perspectives[J]. Eur Radiol, 2021, 31(2): 1049-1058. DOI: 10.1007/s00330-020-07141-9.
[10]
SCHURINK N W, LAMBREGTS D M J, BEETS-TAN R G H. Diffusion-weighted imaging in rectal cancer: current applications and future perspectives[J/OL]. Br J Radiol, 2019, 92(1096): 20180655 [2022-09-02]. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6540856. DOI: 10.1259/bjr.20180655.
[11]
SOFIC A, HUSIC-SELIMOVIC A, EFENDIC A, et al. MRI evaluation of extramural venous invasion (EMVI) with rectal carcinoma using high resolution T2 and combination of high resolution T2 and contrast enhanced T1 weighted imaging[J]. Acta Inform Med, 2021, 29(2): 113-117. DOI: 10.5455/aim.2021.29.113-117.
[12]
王进, 李智慧, 沈浮, 等. 基于高分辨T2WI的影像组学对直肠癌术前分期的应用价值[J]. 放射学实践, 2019, 34(11): 1251-1254. DOI: 10.13609/j.cnki.1000-0313.2019.11.016.
WANG J, LI Z H, SHEN F, et al. Application value of high-resolution T2WI-based imaging in preoperative staging of rectal cancer[J]. Radiol Pract, 2019, 34(11): 1251-1254. DOI: 10.13609/j.cnki.1000-0313.2019.11.016.
[13]
LIN X, ZHAO S, JIANG H J, et al. A radiomics-based nomogram for preoperative T staging prediction of rectal cancer[J]. Abdom Radiol (NY), 2021, 46(10): 4525-4535. DOI: 10.1007/s00261-021-03137-1.
[14]
YOU J, YIN J D. Performances of whole tumor texture analysis based on MRI: predicting preoperative T stage of rectal carcinomas[J/OL]. Front Oncol, 2021, 11: 678441 [2022-10-30]. https://www.ncbi.nlm.nih.gov/pmc/artic-les/PMC8369414. DOI: 10.3389/fonc.2021.678441.
[15]
HUANG Y H, WEI L H, HU Y L, et al. Multi-parametric MRI-based radiomics models for predicting molecular subtype and androgen receptor expression in breast cancer[J/OL]. Front Oncol, 2021, 11: 706733 [2022-10-30]. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8416497/. DOI: 10.3389/fonc.2021.706733.
[16]
LIU L H, LIU Y H, XU L, et al. Application of texture analysis based on apparent diffusion coefficient maps in discriminating different stages of rectal cancer[J]. J Magn Reson Imaging, 2017, 45(6): 1798-1808. DOI: 10.1002/jmri.25460.
[17]
萨莎, 李晶, 李晓东, 等. 基于CT图像及临床资料的随机森林模型对结直肠癌术前T分期的诊断价值[J]. 中华放射学杂志, 2017, 51(12): 933-938. DOI: 10.3760/cma.j.issn.1005-1201.2017.12.009.
SA S, LI J, LI X D, et al. Application of random forest model based on CT images and clinical data in preoperative T staging of colorectal cancer[J]. Chin J Radiol, 2017, 51(12): 933-938. DOI: 10.3760/cma.j.issn.1005-1201.2017.12.009.
[18]
MA X L, SHEN F, JIA Y, et al. MRI-based radiomics of rectal cancer: preoperative assessment of the pathological features[J/OL]. BMC Med Imaging, 2019, 19(1): 86 [2022-09-03]. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6864926. DOI: 10.1186/s12880-019-0392-7.
[19]
凌人男, 杨若峰, 易芹芹, 等. 基于神经网络模型预测宫颈癌淋巴结转移[J]. 磁共振成像, 2021, 12(10): 16-21. DOI: 10.12015/issn.1674-8034.2021.10.004.
LING R N, YANG R F, YI Q Q, et al. Prediction of lymph node metastasis of cervical cancer based on neural network model[J]. Chin J Magn Reson Imaging, 2021, 12(10): 16-21. DOI: 10.12015/issn.1674-8034.2021.10.004.
[20]
FELISAZ P F, COLELLI G, BALLANTE E, et al. Texture analysis and machine learning to predict water T2 and fat fraction from non-quantitative MRI of thigh muscles in Facioscapulohumeral muscular dystrophy[J/OL]. Eur J Radiol, 2021, 134: 109460 [2022-11-03]. https://www.sciencedirect.com/science/article/pii/S0720048X20306501. DOI: 10.1016/j.ejrad.2020.109460.
[21]
MENG X L, WEI Q P, MENG L, et al. Feature fusion and detection in alzheimer's disease using a novel genetic multi-kernel SVM based on MRI imaging and gene data[J/OL]. Genes, 2022, 13(5): 837 [2022-09-13]. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9140721. DOI: 10.3390/genes13050837.
[22]
SIAKALLIS L, SUDRE C H, MULHOLLAND P, et al. Longitudinal structural and perfusion MRI enhanced by machine learning outperforms standalone modalities and radiological expertise in high-grade glioma surveillance[J]. Neuroradiology, 2021, 63(12): 2047-2056. DOI: 10.1007/s00234-021-02719-6.
[23]
CUI Y F, LIU H H, REN J L, et al. Development and validation of a MRI-based radiomics signature for prediction of KRAS mutation in rectal cancer[J]. Eur Radiol, 2020, 30(4): 1948-1958. DOI: 10.1007/s00330-019-06572-3.
[24]
薛珂, 丁莹莹, 李振辉, 等. 利用磁共振成像动态增强纹理特征预测不同分子亚型乳腺癌[J]. 实用放射学杂志, 2020, 36(8): 1235-1239. DOI: 10.3969/j.issn.1002-1671.2020.08.015.
XUE K, DING Y Y, LI Z H, et al. Dynamic contrast-enhanced MRI texture analysis for distinguishing different molecular subtypes of breast cancer[J]. J Pract Radiol, 2020, 36(8): 1235-1239. DOI: 10.3969/j.issn.1002-1671.2020.08.015.
[25]
XIONG H, HE X J, GUO D J. 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.
[26]
EUN N L, KANG D, SON E J, et al. Texture analysis using machine learning-based 3-T magnetic resonance imaging for predicting recurrence in breast cancer patients treated with neoadjuvant chemotherapy[J]. Eur Radiol, 2021, 31(9): 6916-6928. DOI: 10.1007/s00330-021-07816-x.
[27]
FAN T W, MALHI H, VARGHESE B, et al. Computed tomography-based texture analysis of bladder cancer: differentiating urothelial carcinoma from micropapillary carcinoma[J]. Abdom Radiol (NY), 2019, 44(1): 201-208. DOI: 10.1007/s00261-018-1694-x.
[28]
MENG J, LIU S L, ZHU L J, et al. Texture Analysis as Imaging Biomarker for recurrence in advanced cervical cancer treated with CCRT[J/OL]. Sci Rep, 2018, 8(1): 11399 [2022-09-03]. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6065361. DOI: 10.1038/s41598-018-29838-0.
[29]
XING P Y, CHEN L G, YANG Q S, et al. Differentiating prostate cancer from benign prostatic hyperplasia using whole-lesion histogram and texture analysis of diffusion- and T2-weighted imaging[J/OL]. Cancer Imaging, 2021, 21(1): 54 [2022-09-03]. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8477463. DOI: 10.1186/s40644-021-00423-5.
[30]
NKETIAH G A, ELSCHOT M, SCHEENEN T W, et al. Utility of T2-weighted MRI texture analysis in assessment of peripheral zone prostate cancer aggressiveness: a single-arm, multicenter study[J/OL]. Sci Rep, 2021, 11: 2085 [2022-09-03]. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7822867. DOI: 10.1038/s41598-021-81272-x.
[31]
YIN J D, SONG L R, LU H C, et al. Prediction of different stages of rectal cancer: texture analysis based on diffusion-weighted images and apparent diffusion coefficient maps[J]. World J Gastroenterol, 2020, 26(17): 2082-2096. DOI: 10.3748/wjg.v26.i17.2082.
[32]
HOU M, ZHOU L, SUN J H. Deep-learning-based 3D super-resolution MRI radiomics model: superior predictive performance in preoperative T-staging of rectal cancer[J]. Eur Radiol, 2023, 33(1): 1-10. DOI: 10.1007/s00330-022-08952-8.
[33]
文大光, 胡斯娴, 李真林, 等. 基于T2WI与RS-EPI DWI影像组学特征的自动化机器学习模型预测直肠癌术前T分期的价值[J]. 四川大学学报(医学版), 2021, 52(4): 698-705. DOI: 10.12182/20210460201.
WEN D G, HU S X, LI Z L, et al. Application of automated machine learning based on radiomics features of T2WI and RS-EPI DWI to predict preoperative T staging of rectal cancer[J]. J Sichuan Univ Med Sci Ed, 2021, 52(4): 698-705. DOI: 10.12182/20210460201.

上一篇 基于钆塞酸二钠增强MRI肝胆期图像影像组学模型评估肝癌微血管侵犯分级的价值
下一篇 IVIM定量参数及联合参数在宫颈癌诊断及临床分期中的应用价值
  
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