發(fā)布時間：2018-04-20 19:26

  本文選題：列線圖 + 影像學��； 參考：《浙江大學》2017年碩士論文

【摘要】：背景:大多數(shù)胰腺癌患者診斷時已處于較晚期階段,往往伴隨有胰周鄰近血管侵犯。然而,傳統(tǒng)的計算機斷層掃描(CT)評價標準主要關(guān)注于血管是否受到侵犯,而較少關(guān)注患者長期生存預后。因此,我們希望通過建立一個新的有效的基于影像學的評估模型,來預測胰腺癌手術(shù)聯(lián)合靜脈重建患者的預后情況。方法:回顧性收集2012年1月至2016年6月4個大型醫(yī)學中心的266例接受聯(lián)合靜脈切除重建的胰腺癌手術(shù)患者的臨床資料。將266例患者按3:1比例隨機分為建模組(201例)和驗證組(65例),人口統(tǒng)計學因素和影像學因素納入到單因素和多因素Cox回歸分析。應用C-index和校正曲線來評判最終的列線圖模型,并借助65例驗證組患者數(shù)據(jù)資料進行驗證。我們將血管異常在CT的表現(xiàn)分為4種分型,并納入到新的預測模型。通過卡方檢驗分析血管異常在影像學和病理學表現(xiàn)上的相關(guān)性。結(jié)果:年齡、腫瘤與血管接觸長度和血管異常程度是患者預后生存的獨立危險因素(P值分別為0.002,0.016,0.001)。新建立的列線圖模型有著較好的預后預測能力,該模型預測建模組和驗證組的總生存期的C-index分別為0.811和0.800。將CT上血管異常程度的7個特征表現(xiàn)(無異常、輕度變形、淚滴征、集束征、狹窄1/2、閉塞和栓子)根據(jù)不同的預后分為1型(無異常)、2型(輕度變形或淚滴征)、3型(集束征或狹窄1/2)和4型(閉塞或栓子)。血管異常的影像學表現(xiàn)和病理學結(jié)果的相關(guān)性程度較高。結(jié)論:新建立的列線圖模型能夠準確地預測聯(lián)合胰腺癌手術(shù)聯(lián)合靜脈切除重建患者術(shù)后的總生存期。
[Abstract]:Background: most patients with pancreatic cancer are in advanced stage of diagnosis, often accompanied by peripancreatic vascular invasion. However, the traditional evaluation criteria of computed tomography (CT) focus on whether the blood vessels are invaded or not, but less on the long-term survival and prognosis of the patients. Therefore, we hope to establish a new and effective image-based evaluation model to predict the prognosis of patients with pancreatic cancer combined with venous reconstruction. Methods: the clinical data of 266 patients with pancreatic cancer underwent combined venectomy and reconstruction from January 2012 to June 2016 were retrospectively collected. Two hundred and six patients were randomly divided into modeling group (201 cases) and validation group (65 cases) according to 3:1 ratio. Demographic and imaging factors were included in univariate and multivariate Cox regression analysis. C-index and calibration curves were used to evaluate the final line graph model, and 65 patient data of the validation group were used to verify the model. We divided the CT findings of vascular abnormalities into four types and incorporated them into a new predictive model. The correlation between imaging and pathological findings of vascular abnormalities was analyzed by chi-square test. Results: age, tumor and vascular contact length and degree of vascular abnormality were independent risk factors for survival of patients with prognosis (P = 0.002, 0.016, 0.001, respectively). The new line graph model has good prognostic prediction ability. The C-index of the total survival time of the modeling group and the validation group is 0.811 and 0.800, respectively. Seven features (no abnormality, slight deformation, tear drop sign, cluster sign) of the abnormal blood vessels on CT were presented. According to different prognosis, stenosis 1 / 2, occlusion and embolus were divided into 1 type (no abnormal type 2) (mild deformation or tear sign) and 4 type (occlusion or embolus) (cluster sign or stenosis 1 / 2). There is a high correlation between the imaging findings of vascular abnormalities and pathological results. Conclusion: the new line graph model can accurately predict the total survival time of patients with pancreatic cancer combined with venous resection and reconstruction.
【學位授予單位】：浙江大學
【學位級別】：碩士
【學位授予年份】：2017
【分類號】：R735.9
，

本文編號：1779110