首页 | 本学科首页   官方微博 | 高级检索  
     

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
引用本文:邓娇,谭一雄,顾潜彪,容鹏飞,王维,刘晟. 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: 10.11817/j.issn.1672-7347.2019.03.005
作者姓名:邓娇  谭一雄  顾潜彪  容鹏飞  王维  刘晟
作者单位:中南大学湘雅三医院放射科,长沙,410013;中南大学湘雅三医院放射科,长沙,410013;中南大学湘雅三医院放射科,长沙,410013;中南大学湘雅三医院放射科,长沙,410013;中南大学湘雅三医院放射科,长沙,410013;中南大学湘雅三医院放射科,长沙,410013
摘    要:目的:探讨基于CT图像的影像组学标签鉴别原发性胃淋巴瘤与Borrmann IV型胃癌的可行性。方法:回顾性分析2009年1月至2017年4月在中南大学湘雅三医院经病理证实为原发性胃淋巴瘤或Borrmann IV型胃癌且术前进行过腹部CT增强扫描的71例患者,其中原发性胃淋巴瘤患者28例,Borrmann IV型胃癌患者43例。采用基于Matlab 2017a软件的特征提取算法提取影像组学特征,并利用logistic回归模型进行特征筛选以建立CT影像组学标签。通过纳入胃周脂肪浸润、胃壁柔软性、腹部淋巴结及周围脏器转移、腹水、黏膜白线征和病灶厚度等征象,构建CT征象诊断模型。应用受试者工作特征(receiver operating characteristic,ROC)曲线评估影像组学标签和CT征象诊断模型的分类性能。结果:从每个患者CT扫描的肿瘤区域中提取32个三维特征,通过降维发现2个特征是最重要的鉴别诊断因子并建立了影像组学标签;CT征象诊断模型由腹水、胃周脂肪浸润、胃壁柔软性及黏膜白线征组成。影像组学标签和CT征象诊断模型的曲线下面积(area under curve,AUC)分别为0.964和0.867;准确性分别为94.4%和80.2%;敏感性分别为93.0%和74.4%;特异性分别为96.4%和89.3%。经Delong检验,影像组学标签的诊断效能高于CT征象诊断模型(P<0.001)。结论:基于CT图像的影像组学标签能够较准确地鉴别Borrmann IV型胃癌与原发性胃淋巴瘤,为临床辅助诊断提供有利的手段。

关 键 词:BorrmannⅣ型胃癌  原发性胃淋巴瘤  影像组学  CT征象

Application of CT-based radiomics in differentiating primary gastric lymphoma from Borrmann type IV gastric cancer
DENG Jiao,TAN Yixiong,GU Qianbiao,RONG Pengfei,WANG Wei,LIU Sheng. Application of CT-based radiomics in differentiating primary gastric lymphoma from Borrmann type IV gastric cancer[J]. Journal of Central South University. Medical sciences, 2019, 44(3): 257-263. DOI: 10.11817/j.issn.1672-7347.2019.03.005
Authors:DENG Jiao  TAN Yixiong  GU Qianbiao  RONG Pengfei  WANG Wei  LIU Sheng
Affiliation:Department of Radiology, Third Xiangya Hospital, Central South University, Changsha 410013, China
Abstract:Objective: To explore the feasibility of CT-based image radiomics signature in identification of primary gastric lymphoma and Borrmann type IV gastric cancer.Methods: A retrospective analysis of 71 patients with primary gastric lymphoma or Borrmann type IV gastric cancer confirmed by pathology in our Hospital from January 2009 to April 2017 was performed. There were 28 patients with primary gastric lymphoma and 43 patients with Borrmann type IV gastric cancer. The feature extraction algorithm based on Matlab 2017a software was used to extract the features of image, and the logistic regression model was used to screen the features to establish radiomics signature. The CT sign diagnosis model was established, which included the periplasmic fat infiltration, softness of the stomach wall, abdominal lymph node and peripheral organ metastasis, ascites, mucosal white line sign and lesion thickness. The classification of the two models was evaluated by receiver operating characteristic curve.Results: A total of 32 3D features were extracted from CT image for each patients. Two features were found to be the most important differential diagnosis factors, and the radiomics signature was established. The CT sign diagnosis model consisted of ascites, periplasmic fat infiltration, stomach wall softness and mucosal white line. For the radiomics signature and the CT subjective finding model, the AUCs were 0.964 and 0.867 with the accuracy at 94.4% and 80.2%, the sensitivity at 93.0% and 74.4%, the specificity at 96.4% and 89.3%, respectively. After Delong test, the diagnostic efficacy of the radiomics signature was higher than the CT sign diagnosis model (P<0.001).Conclusion: CT-based image radiomics signature can accurately identify primary gastric lymphoma and Borrmann type IV gastric cancer, and can potentially provide important assistance in clinical diagnosis for the two diseases.
Keywords:Borrmann type IV gastric cancer  primary gastric lymphoma  radiomics  CT signs  
本文献已被 万方数据 等数据库收录!
点击此处可从《中南大学学报(医学版)》浏览原始摘要信息
点击此处可从《中南大学学报(医学版)》下载全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号