作者投稿
专家审稿
编辑办公
邮件订阅
RSS
Value of CT Texture Analysis for Volumetric Changes of Abdominal Aortic Aneurysm after Endovascular Repair
-
摘要:目的 探讨CT纹理分析对腹主动脉瘤腔内修补术(endovascular aneurysm repair, EVAR)后瘤体体积变化的评价意义。方法 回顾性收集2014年7月至2019年6月于北京协和医院放射科行腹盆部增强CT血管造影(computed tomography angiography, CTA)随访的EVAR术后患者的临床及影像学资料, 患者于术后第3、12个月分别接受两次腹盆部CTA检查, 以传统测量方法为分组依据, 根据动脉瘤体积变化情况将患者分为动脉瘤体积增大组和体积未增大组。采用分割软件, 手动勾勒所有动脉瘤术后第1次随访时轴位最大截面, 提取动脉瘤支架外血栓部分的纹理特征, 同时采用灰度共生矩阵(grey level co-occurrence matrix, GLCM)、灰度游程矩阵(grey level run length matrix, GLRLM)和灰度差分矩阵(grey level difference matrix, GLDM)3种方法, 分析其与动脉瘤体积增大的关系。结果 共70例符合纳入和排除标准的患者入选本研究, 其中29例(41.4%, 29/70)EVAR术后腹主动脉瘤体积增大, 41例(58.6%, 41/70)瘤体未增大。GLCM、GLRLM和GLDM 3种方法预测动脉瘤体积增大的曲线下面积分别为0.892、0.888和0.800, GLCM的预测效能最佳, 灵敏度和特异度分别为86.2%和85.4%, GLDM的曲线下面积最小, 灵敏度和特异度分别为86.2%和68.3%。结论 CT纹理分析能有效预测EVAR术后动脉瘤的体积变化。Abstract:Objective The aim of this study was to investigate the value of CT texture analysis for volumetric changes of abdominal aortic aneurysm (AAA) after endovascular aneurysm repair (EVAR).Methods Clinical and imaging data of patients with AAA undergoing CT angiography (CTA) at the 3rd and 12th months after EVAR in the Department of Radiology, Peking Union Medical College Hospital between July 2014 and June 2019 were retrospectively collected. Using the traditional measurement as the grouping basis, patients were divided into increased and no-increased groups according to the volume changes during the follow-up. Segmentation software was used to manually outline the maximum axial sections of all aneurysms and extract the texture features of the thrombus. The grey level co-occurrence matrix (GLCM), the grey level run length matrix(GLRLM), and the grey level difference matrix(GLDM) were calculated to analyzed their relationship with the enlargement of the aneurysm volume.Results A total of 70 patients meeting the inclusive and exclusive criteria were enrolled in this study, 29 cases (41.4%, 29/70) in the increased group and 41 (58.6%, 41/70) in the non-increased group. The areas under the ROC curve of GLCM, GLRLM, and GLDM for the diagnosis of increased volumes were 0.892, 0.888, and 0.800, respectively. The best prediction efficiency was shown in GLCM with the sensitivity and specificity of 86.2% and 85.4% respectively. GLDM had the worst pridiction efficiency with a sensitivity of 86.2% and a specificity of 68.3%.Conclusion CT texture analysis might predict the volumetric change of AAA after EVAR.
-
Keywords:
- texture analysis /
- abdominal aortic aneurysm /
- volume /
- endovascular aneurysm repair
-
腹主动脉瘤的定义是动脉管壁永久性局限性扩张超过正常血管直径的50%或主动脉直径>3 cm[1]，该病在50岁以上男性发病率(4%~8%)高于同龄女性(1%~1.3%)[2]。1991年Parodi等[3]首次报道了腹主动脉瘤腔内修补术(endovascular aneurysm repair，EVAR)，即在血管腔内置入人工覆膜支架，重建血流通道，避免瘤体受到腹主动脉高压血流冲击发生破裂。与传统开放手术相比，EVAR微创、安全，可显著降低围手术期死亡率[4-6]，内漏是其常见并发症，发生率达16%~33%[7-9]，但发生内漏的动脉瘤，其破裂风险不一定增加[10-11]，故以内漏评估EVAR的疗效并不准确。
动脉瘤体积增大是动脉瘤破裂最重要的预测指标[12-13]，但传统动脉瘤体积测量过程繁琐、耗时，而CT纹理分析是近年广泛应用的一种图像后处理技术，通过量化分析图像像素灰度值的局部特征、变化规律及分布模式，定量鉴别特定区域的异质性[14]，已广泛应用于淋巴瘤、食管癌和结直肠癌等肿瘤的分型分析、治疗前评估及疗效预测[15-19]。EVAR术后CT信号的异质性提示血管壁发生结构改变[20-21]，可能与动脉瘤体积增大直接相关。
目前鲜有将CT纹理分析应用于动脉瘤预后监测的报道，本研究通过对EVAR术后动脉瘤进行CT纹理分析，探究其与动脉瘤体积变化的关系，以期为临床提供一种更准确、简便的风险分层方法。
1. 资料与方法
1.1 研究对象
回顾性收集2014年7月至2019年6月于北京协和医院放射科行腹盆部增强CT血管造影(computed tomography angiography，CTA)随访的EVAR术后患者的临床及影像学资料。
纳入标准：(1)肾下型腹主动脉瘤EVAR术后；(2)术后第3和12个月在本院规律随访，并完成两次腹盆CTA者。
排除标准：(1)合并其他腹盆腔脏器严重疾病，影响腹主动脉瘤轮廓准确识别者；(2)合并主动脉夹层或大动脉炎等其他血管疾病者。
本研究通过中国医学科学院北京协和医院伦理审查委员会审查(审批号：S-K1016)。
1.2 CT血管造影响
采用高压注射器由患者右肘正中静脉注射造影剂碘普罗胺(370 mg/ml，上海博莱科信谊药业有限责任公司)90 ml，流速4.0 ml/s。患者取仰卧位，双手上举过头，扫描范围自膈肌水平至耻骨联合水平。采用第一代双源CT(德国西门子)智能触发扫描，触发层面为腹腔干水平腹主动脉，触发阈值100 HU，触发后立刻扫描动脉期图像，25 s后采集门脉期图像。扫描参数：管电压120 kV，管电流200 mA，机架旋转时间330 ms，准直器2.0 mm×32.0 mm×0.6 mm，螺距0.8 mm。
1.3 传统动脉瘤体积测量及分组
将增强动脉期图像导入飞利浦重建工作站，利用后处理软件测量数据。使用画笔工具，以逐节方式手动勾勒主动脉外轮廓，体积测量范围从较低一侧肾动脉水平到两侧髂总动脉分叉水平[10, 22-25]，包括给定解剖区域内的正常主动脉及瘤体边缘的任何钙化和附壁血栓，排除所有瘤体分支血管。勾勒完成后计算机自动计算体积大小。
术后两次随访的动脉瘤体积差大于2%定义为瘤体体积增大，反之定义为瘤体未增大[10, 26]。根据动脉瘤体积变化情况，将入组患者分为动脉瘤体积增大组和体积未增大组。
1.4 纹理分析
取术后第1次随访时动脉期图像中动脉瘤囊轴位最大层面，使用分割软件MATLAB手动勾勒动脉瘤的轴位最大截面，将勾勒好的图像交第三方公司进行纹理分析。采用灰度矩阵提取感兴趣区(region of interest，ROI)血栓部分的纹理特征，生成ROI内的纹理特征值后，输入由神经网络生成的分类器中，得到最终分类结果。此过程中第三方公司并不知晓患者体积分组情况。
采用3种灰度矩阵提取纹理特征，分别为灰度共生矩阵(grey level co-occurrence matrix，GLCM)、灰度游程矩阵(grey level run length matrix，GLRLM)和灰度差分矩阵(grey level difference matrix，GLDM)。这3种矩阵可获得像素对或像素组间灰度值的二阶或更高阶统计关系，描述图像内容的纹理特征。GLCM描述像素距离和角度的二阶联合条件概率密度函数，体现像素灰度值的空间关系；GLRLM提取高阶纹理信息的二维矩阵，为与GLCM可比，计算4个方向的GLRLM矩阵，并将灰色级数保持在16；GLDM基于两个具有特殊关系像素点的出现概率，特殊关系指这两个像素点间的位移差和灰度值差固定，同GLCM一样，计算动脉瘤内4个方向的平均分布。
1.5 统计学处理
采用SPSS22.0软件进行统计分析。符合正态分布的计量资料以均数±标准差表示，组间比较采用配对样本t检验；不符合正态分布的计量资料以中位数(四分位数)表示，采用Wilcoxon秩和检验。受试者工作特征曲线(receiver operation characteristic curve，ROC)下面积(area under curve，AUC)用于评估纹理分析效果。P＜0.05为差异具有统计学意义。
2. 结果
2.1 一般临床资料
共70例符合纳入和排除标准的腹主动脉瘤患者入选本研究(图 1)，男性62例，女性8例，平均年龄(68.8±8.6)岁，60岁以上患者61例(87.1%，61/70)，既往史中有高血压、冠状动脉粥样硬化性心脏病、吸烟、2型糖尿病者分别为44例(62.9%，44/70)、38例(54.3%，38/70)、35例(50.0%，35/70)、14例(20.0%，14/70)。
2.2 腹主动脉瘤体积变化情况
70例患者中，29例(41.4%，29/70)出现EVAR术后腹主动脉瘤体积增大，41例(58.6%，41/70)瘤体未增大(表 1)。
表 1 腹主动脉瘤腔内修补术后动脉瘤体积的变化[M(P25, P75), cm3]分组 术后3个月 术后12个月 Z值 P值 体积增大组(n=29) 114.6(88.7，226.2) 123.4(91.8，241.5) -4.703 ＜0.01 体积未增大组(n=41) 127.0(97.4，196.9) 114.2(78.7，183.5) -5.417 ＜0.01 2.3 纹理特征分析结果
三种灰度矩阵纹理分析技术对EVAR术后动脉瘤体积增大的诊断效能各有侧重(表 2)，其中GLCM的AUC最大(0.892)，其准确度与GLRLM相当(0.859)，灵敏度与GLDM相当(0.862)，但特异度略低于GLRLM。GLDM的AUC最小(0.800)，准确度(0.788)和特异度(0.683)也最低。
表 2 3种灰度矩阵纹理分析技术对腹主动脉瘤腔内修补术后动脉瘤体积变化的诊断效能纹理分析技术 曲线下面积 准确度 灵敏度 特异度 GLCM 0.892 0.859 0.862 0.854 GLDM 0.800 0.788 0.862 0.683 GLRLM 0.888 0.859 0.828 0.902 GLCM：灰度共生矩阵；GLRLM：灰度游程矩阵；GLDM：灰度差分矩阵 3. 讨论
图像纹理特征分析是对图像像素灰度值的局部特征分析法，包括统计分析、结构分析、模型分析和频谱分析4种，其中统计分析最常用[14]，通过分析纹理的统计属性来描述纹理，提供纹理的平滑、稀疏等分布特性。本研究正是基于此种纹理分析方法，观察GLCM、GLDM和GLRLM对评估EVAR术后动脉瘤体积变化的预测价值，发现GLCM的诊断效能最优(AUC=0.892)。
纹理分析的有效性已在诸多领域得到证实，如鉴别病变性质、治疗前评估、预测疗效等，研究多集中在肿瘤学方面，如通过乳腺X线纹理分析鉴别肿瘤良恶性[27]，通过正电子发射断层显像/计算机体层成像(positron emission tomography/computed tomography, PET/CT)纹理分析将纹理参数与PET/CT最大标准摄取值结合，提高诊断肺癌的灵敏度[28]，原发性结肠癌CT纹理特征与患者5年总体生存率相关[29]，采用GLCM法预测宫颈癌疗效的准确度可达75%[30]。
CT纹理分析技术对血管疾病的辅助诊断亦具备可行性。研究显示，标准灰阶中位数技术较斑块纹理分析技术能更有效预测动脉内膜切除术后微栓塞的程度[31]。Kotze等[32]曾提出腹主动脉瘤CT信号异质性与瘤体扩张相关，联合PET/CT评估腹主动脉瘤的代谢活性，发现中等纹理的峰度与动脉瘤扩展显著相关，该研究是基于一阶统计的CT纹理参数，本研究则以二阶统计量为出发点，通过分析比较GLCM、GLRLM和GLDM的纹理特征，发现GLCM对EVAR术后动脉瘤体积增长的预测效能最佳，其次为GLRLM，最差为GLDM，此结果与García等[33]的结果一致，不同的是本研究是采用体积变化评价动脉瘤预后，García等则以动脉瘤轴位最大径作为预后分类指标。实际上，体积测量在反映动脉瘤真实增长方面优于直径测量[34]，因为体积从三维角度评估瘤体形态变化，而直径仅反应瘤体某一截面的变化。
EVAR术后随访中，放射科医师阅片时往往关注有无内漏及瘤体轴位最大径，易受主观影响，且评估不够全面。本研究中4例患者在第1次随访时出现内漏，但第2次随访时内漏消失，因此随访中发现内漏不一定需要立即干预。相较而言，纹理分析技术能更客观准确地描绘生物组织微观结构改变，获取肉眼无法辨识的细微信息，对预测动脉瘤体增大的最优灵敏度可达86.2%，特异度达85.4%。
本研究存在部分局限性：(1)为回顾性研究，入组例数较少，术后随访时间较短，存在一定的选择偏倚；(2)由于软件技术限制，在分析动脉瘤纹理特征时仅提取了瘤体最大截面的二维纹理信息，未能从三维角度全面评估；(3)GLCM、GLRLM及GLDM纹理特征参数的整合过程由人工神经网络算法完成，由于该资料为内部封存，提取困难，致使结果部分的阐述尚欠明确。
综上，CT纹理分析能有效预测EVAR术后动脉瘤的体积变化。目前该技术的研究尚处初级阶段，未来还需开展大样本、多中心、前瞻性研究进一步验证本文结论，推动纹理分析成为常规工具，为临床诊疗提供更精准的参考。
利益冲突 无 -
表 1 腹主动脉瘤腔内修补术后动脉瘤体积的变化[M(P25, P75), cm3]
分组 术后3个月 术后12个月 Z值 P值 体积增大组(n=29) 114.6(88.7，226.2) 123.4(91.8，241.5) -4.703 ＜0.01 体积未增大组(n=41) 127.0(97.4，196.9) 114.2(78.7，183.5) -5.417 ＜0.01 
下载: 导出CSV
表 2 3种灰度矩阵纹理分析技术对腹主动脉瘤腔内修补术后动脉瘤体积变化的诊断效能
纹理分析技术 曲线下面积 准确度 灵敏度 特异度 GLCM 0.892 0.859 0.862 0.854 GLDM 0.800 0.788 0.862 0.683 GLRLM 0.888 0.859 0.828 0.902 GLCM：灰度共生矩阵；GLRLM：灰度游程矩阵；GLDM：灰度差分矩阵
下载: 导出CSV
-
[1] Nowicka M, Kowalczyk A, Rusak G, et al. Evaluation the aortic aneurysm remodeling after a successful stentgraft implantation[J]. Pol J Radiol, 2016, 486-490.
[2] Kent KC, Zwolak RM, Egorova NN, et al. Analysis of risk factors for abdominal aortic aneurysm in a cohort of more than 3 million individuals[J]. J Vasc Surg, 2010, 52:539-548. DOI: 10.1016/j.jvs.2010.05.090
[3] Parodi JC, Palmaz JC, Barone HD. Transfemoral intralu-minal graft implantation for abdominal aortic aneurysms[J]. Ann Vase Surg, 1991, 5:491-499. https://www.sciencedirect.com/science/article/abs/pii/S0890509606615410
[4] Daye D, Walker TG. Complications of endovascular aneur-ysm repair of the thoracic and abdominal aorta:evaluation and management[J]. Cardiovasc Diagn Ther, 2018, (Suppl 1):S138-S156. http://europepmc.org/abstract/MED/29850426
[5] IMPROVE Trial Investigators, Powell JT, Sweeting MJ, et al. Endovascular or open repair strategy for ruptured abdominal aortic aneurysm:30 day outcomes from IMPROVE randomised trial[J]. BMJ, 2014, 348:f7661. DOI: 10.1136/bmj.f7661
[6] Takagi H, Umemoto T; ALICE (All-Literature Investigation of Cardiovascular Evidence Group). A meta-analysis of adjusted observational studies and randomized controlled trials of endovascular versus open surgical repair for ruptured abdominal aortic aneurysm[J]. Int Angiol, 2016, 35:534-545. https://www.minervamedica.it/en/journals/international-angiology/article.php?cod=R34Y2016N06A0534
[7] Javor D, Wressnegger A, Unterhumer S, et al. Endoleak detection using single-acquisition split-bolus dual-energy computer tomography (DECT)[J]. Eur Radiol, 2017, 27:1622-1630. DOI: 10.1007/s00330-016-4480-6
[8] Hobo R, Buth J. Secondary interventions following endovascular abdominal aortic aneurysm repair using current endografts. A EUROSTAR report[J]. J Vasc Surg, 2006, 43:896-902. DOI: 10.1016/j.jvs.2006.01.010
[9] Cannavale A, Lucatelli P, Corona M, et al. Evolving concepts and management of endoleaks after endovascular aneurysm repair: where do we stand in 2019?[J]. Clin Radiol, 2019, pii: S0009-9260(19)30641-5.
[10] Bley TA, Chase PJ, Reeder SB, et al. Endovascular abdominal aortic aneurysm repair:nonenhanced volumetric CT for follow-up[J]. Radiology, 2009, 253:253-262. DOI: 10.1148/radiol.2531082093
[11] Leemans EL, Willems TP, van der Laan MJ, et al. Biomechanical Indices for Rupture Risk Estimation in Abdominal Aortic Aneurysms[J]. J Endovasc Ther, 2017, 24:254-261. DOI: 10.1177/1526602816680088
[12] Bernhard VM, Mitchell RS, Matsumura JS, et al. Ruptured abdominal aortic aneurysm after endovascular repair[J]. J Vasc Surg, 2002, 35:1155-1162. DOI: 10.1067/mva.2002.123758
[13] Demehri S, Signorelli J, Kumamaru KK, et al. Volumetric quantification of type Ⅱ endoleaks:an indicator for aneurysm sac growth following endovascular abdominal aortic aneurysm repair[J]. Radiology, 2015, 271:282-290. http://europepmc.org/abstract/MED/24475801
[14] Lubner MG, Smith AD, Sandrasegaran K, et al. CT Texture Analysis:Definitions, Applications, Biologic Correlates, and Challenges[J]. Radiographics, 2017, 37:1483-1503. DOI: 10.1148/rg.2017170056
[15] Gao N, Tian S, Li X, et al. Three-Dimensional Texture Feature Analysis of Pulmonary Nodules in CT Images:Lung Cancer Predictive Models Based on Support Vector Machine Classifier[J]. J Digit Imaging, 2019, doi:10.1007/s10278-019-00238-8.[Epub ahead of print].
[16] Reinert CP, Federmann B, Hofmann J, et al. Computed tomography textural analysis for the differentiation of chronic lymphocytic leukemia and diffuse large B cell lymphoma of Richter syndrome[J]. Eur Radiol, 2019, 29:6911-6921. DOI: 10.1007/s00330-019-06291-9
[17] Seidler M, Forghani B, Reinhold C, et al. Dual-Energy CT Texture Analysis With Machine Learning for the Evaluation and Characterization of Cervical Lymphadenopathy[J]. Comput Struct Biotechnol, 2019, 17:1009-1015. DOI: 10.1016/j.csbj.2019.07.004
[18] Beukinga RJ, Hulshoff JB, Van Dijk LV, et al. Predicting response to neoadjuvant chemoradiotherapy in esophageal cancer with textural features derived from pre-treatment 18F-FDG PET/CT imaging[J]. J Nucl Med, 2017, 58:723-729. DOI: 10.2967/jnumed.116.180299
[19] Taguchi N, Oda S, Yokota Y, et al. CT texture analysis for the prediction of KRAS mutation status in colorectal cancer via a machine learning approach[J]. Eur J Radiol, 2019, 118:38-43. DOI: 10.1016/j.ejrad.2019.06.028
[20] Bezy-Wendling J, Kretowski M, Rolland Y, et al. Toward a better un-derstanding of texture in vascular CT scan simulated images[J]. IEEE Trans Biomed Eng, 2001, 48:120-124. DOI: 10.1109/10.900272
[21] Ganeshan B, Miles KA, Young RC, et al. In search of biologic correlates for liver texture on portal-phase CT[J]. Acad Radiol, 2007; 14:1058-1068. DOI: 10.1016/j.acra.2007.05.023
[22] Bargellini I, Cioni R, Petruzzi P, et al. Endovascular repair of abdominal aortic aneurysms:analysis of aneurysm volumetric changes at mid-term follow-up[J]. Cardiovasc Intervent Radiol, 2005, 28:426-433. DOI: 10.1007/s00270-004-0171-9
[23] Boos J, Brook OR, Fang J, et al. What Is the Optimal Abdominal Aortic Aneurysm Sac Measurement on CT Images during Follow-up after Endovascular Repair?[J]. Radiology, 2017, 285:1032-1041. DOI: 10.1148/radiol.2017161424
[24] Baumueller S, Nguyen TD, Goetti RP, et al. Maximum diameter measurements of aortic aneurysms on axial CT images after endovascular aneurysm repair:sufficient for follow-up?[J]. Cardiovasc Intervent Radiol, 2011, 34:1182-1189. DOI: 10.1007/s00270-010-9992-x
[25] Schnitzbauer M, Güntner O, Wohlgemuth WA, et al. CT after Endovascular Repair of Abdominal Aortic Aneurysms:Diagnostic Accuracy of Diameter Measurements for the Detection of Aneurysm Sac Enlargement[J]. J Vasc Interv Radiol, 2018, 29:178-187. DOI: 10.1016/j.jvir.2017.09.012
[26] Caldwell DP, Pulfer KA, Jaggi GR, et al. Aortic aneurysm volume calculation:effect of operator experience[J]. Abdom Imaging, 2005, 30:259-262. DOI: 10.1007/s00261-004-0280-6
[27] Li Z, Yu L, Wang X, et al. Diagnostic Performance of Mammographic Texture Analysis in the Differential Diagnosis of Benign and Malignant Breast Tumors[J]. Clin Breast Cancer, 2018, 18:e621-e627. DOI: 10.1016/j.clbc.2017.11.004
[28] 王长梅, 管一晖, 张文强, 等. PET/CT显像联合融合图像纹理特征分析在肺癌鉴别诊断中的应用价值探讨[J].中国医学计算机成像杂志, 2010, 16:147-151. http://doi.med.wanfangdata.com.cn/qk/zgyxjsjcx201002012 [29] Ng F, Ganeshan B, Kozarski R, et al. Assessment of Primary Colorectal Cancer Heterogeneity by Using Whole-Tumor Texture Analysis:Contrast-enhanced CT Texture as a Biomarker of 5-year Survival[J]. Radiology, 2013, 266:177-184. DOI: 10.1148/radiol.12120254
[30] Torheim T, Malinen E, Kvaal K, et al. Classification of dynamic contrast enhanced MR images of cervical cancers using texture analysis and support vector machines[J]. IEEE Trans Med Imaging, 2014, 33:1648-1656. DOI: 10.1109/TMI.2014.2321024
[31] Madycki G, Staszkiewicz W, Gabrusiewicz A. Carotid Plaque Texture Analysis Can Predict the Incidence of Silent Brain Infarcts Among Patients Undergoing Carotid Endarterectomy[J]. Eur J Vasc Endovasc Surg, 2006, 31:373-380. DOI: 10.1016/j.ejvs.2005.10.010
[32] Kotze CW, Rudd JH, Ganeshan B, et al. CT signal heterogeneity of abdominal aortic aneurysm as a possible predictive biomarker for expansion[J]. Atherosclerosis, 2014, 233:510-517. DOI: 10.1016/j.atherosclerosis.2014.01.001
[33] García G, Maiora J, Tapia A, et al. Evaluation of texture for classification of abdominal aortic aneurysm after endovascular repair[J]. J Digit Imaging, 2012, 25:369-376. DOI: 10.1007/s10278-011-9417-7
[34] Kontopodis N, Metaxa E, Papaharilaou Y, et al. Value of volume measurements in evaluating abdominal aortic aneurysms growth rate and need for surgical treatment[J]. Eur J Radiol, 2014, 83:1051-1056. DOI: 10.1016/j.ejrad.2014.03.018
-
期刊类型引用(1)
1. 陈震，刘豆. 彩色多普勒超声在腹主动脉瘤诊断中的应用价值. 医药论坛杂志. 2022(17): 111-113 . 
百度学术
其他类型引用(1)
计量
- 文章访问数:
- HTML全文浏览量:
- PDF下载量:
- 被引次数: 2
