腹腔镜胰腺手术后胰瘘的防治体会

目的 探讨腹腔镜胰腺切除术后胰瘘(POPF)发生的原因及其处理.方法 我院2008年2月至2012年3月间施行腹腔镜胰腺手术25例,其中腔镜辅助下胰十二指肠切除术( LAPD) 10例,完全腹腔镜下胰体尾切除术(LDP)15例,术中及术后采取相应的胰瘘防治措施.结果 本组资料中发生POPF5例.胰瘘发生率为20％,其中胰瘘A级2例,胰瘘B级1例,胰瘘C级2例.5例患者均保守治疗而痊愈.结论 腹腔镜胰腺手术POPF发病率仍然较高.术中、术后对患者的恰当处理是减少POPF发生的关键.     

钆塞酸二钠增强MRI影像组学对肝血管瘤腹腔镜肝切除术后疗效的预测价值

目的 探讨钆塞酸二钠（Gd-EOB-DTPA）增强MRI影像组学对肝血管瘤腹腔镜肝切除术疗效的预测价值。方法 选取2018年2月至2020年10月行腹腔镜肝切除术治疗的肝血管瘤患者131例作为研究对象。采用计算机产生随机数法将131例肝血管瘤患者以7:3的比例分为训练集（92例）和验证集（39例）,分别用于构建和验证预测模型效能。根据术后是否出现并发症将训练集患者分为预后不良组和预后良好组,比较两组患者的临床特征资料。使用最小绝对收缩和选择算子（the least absolute shrinkage and selection operator,LASSO）回归5折交叉验证方法筛选训练集临床和影像组学特征,得到最优特征子集。使用机器学习算法构建预测模型,并使用受试者工作特征（receiver operating characteristics,ROC）曲线评价各模型对肝血管瘤术后疗效的预测价值。结果 训练集中预后良好64例,预后不良28例,两组在慢性疾病、肝血管瘤最大径、血管瘤数、血管瘤供血动脉支数、肝切除大小、术中出血量方面差异均有统计学意义（P<0.05）。依据LASSO回归中的最佳λ取值筛选出14个非零系数特征构成的最优特征子集,其中包括12个组学特征和2个临床特征。利用训练集筛选出的最优特征子集构建Logistic回归、CatBoost、XGBoost和LightGBM模型的AUC分别为0.820、0.722、0.784、0.693。验证集构建的Logistic回归、CatBoost、XGBoost和LightGBM模型的ROC曲线下面积（AUC）分别为0.814、0.706、0.785、0.684。4种机器学习模型的预测效果良好,其中Logistic回归的预测效果优于其他3种。结论 血管瘤最大径和血管瘤数目,结合Gd-EOB-DTPA增强MRI影像组学,以此构建的机器学习模型可用于预测肝血管瘤腹腔镜肝切除术后疗效,其中Logistic回归算法构建的模型更加精准。     

胰十二指肠切除术后胰瘘的防治

目的: 探讨胰十二指肠切除术后胰瘘的防治经验.方法: 回顾分析1994年1月至2003年12月60例胰十二指肠切除术,均作Child术式重建.胰肠吻合采用传统双层套入法吻合,胰管予支架管内引流.结果: 共发生胰瘘5例,占8.3%.手术开展早期明显高于开展后期,残胰质地软者发生率高.全组无手术死亡.结论: 胰十二指肠切除术后胰瘘的防治在于技术的提高及胰肠吻合的处理.     

胰十二指肠切除术并发症的预防

目的: 探讨胰十二指肠切除术后并发症的预防方法. 方法: 全部病例均在多功能手术解剖器的辅助下,行胰十二指肠切除术、捆绑式胰肠吻合术,同时注重精细操作和围手术期处理. 结果: 16例病人中除1例术后出现应激性渍疡外,无其他严重并发症发生. 结论: 通过改进手术技巧和方法,可有效地预防胰十二指肠切除术后胰瘘、胆瘘、腹腔内出血等并发症的发生.     

胰十二指肠切除术后早期并发症的防治

目的 探讨胰十二指肠切除术后早期并发症的预防和治疗. 方法 回顾性分析60例胰十二指肠切除术后并发症的发生情况及治疗方法. 结果 术后出现胰漏、胆漏、出血、感染等早期并发症16例,发生率为26.7%(16/60),手术死亡率为3.3%(2/60). 结论 积极预防、细致操作和合理诊治是降低胰十二指肠切除术后早期并发症的主要措施.     

改良胰空肠吻合预防胰十二指肠切除术后胰瘘

目的: 探讨胰十二指肠切除术中细胰管与空肠黏膜吻合技术的改进.方法: 回顾1994~2004年间接受胰十二指肠切除手术且胰管直径≤2 mm的恶性肿瘤患者61例临床资料,其中29例采取Child胰肠端端套入式吻合法(A组),32例行胰管与空肠黏膜二点或三点吻合法(B组),比较两组病人胰瘘的发生率、病死率、住院时间.结果:A组胰瘘的发生率为24.1%,B组为3.1%,差异有显著性意义(P<0.05).A组病死率为13.8%, B组病死率为0%, 差异有显著性意义 (P<0.05). A, B两组的住院天数分别为:(28.7±6.6)d,(19.4±4.6)d,差异有非常显著性意义(P<0.01).结论: 胰十二指肠切除术中对细胰管(直径≤2 mm)与空肠黏膜二点或三点吻合法能有效防治术后胰瘘,操作简单,是胰十二指肠切除术中对细胰管较可靠的处理方式.     

胰十二指肠切除术后急诊再手术23例诊治体会

目的: 探讨胰十指肠切除术后再手术的原因、手术方式及价值。 临床资料:回顾性分析1990~1998年间胰十二指肠切除术后再手术23例的原 因、诊断及处理方式。本组再手术率9.9%,再手术原因依次是腹腔感染（43.5%）,上消化 道出血（21.7%）,腹腔内出血（17.4%）,切口裂开和胰胆瘘（13.4%,8.7%）。结论: 加强围手术期尤其是术中处理,重视再手术在胰十二指肠切除术后并 发症治疗中的价值是提高手术成功率的关键。     

FRS与a-FRS对不同胰腺手术术后胰瘘的预测价值分析

背景与目的:胰十二指肠切除术(PD)和胰体尾切除术(DP)是治疗胰腺肿瘤常见的手术方式。术后胰瘘(POPF)是胰腺术后最常见、最严重的并发症之一,若能准确预测POPF的发生将有重要的临床意义。胰瘘危险评分(FRS)和新的胰瘘危险评分(a-FRS)是运用最广的两个POPF预测模型,但这两个预测模型能否有效预测POPF仍需进一步验证。本研究比较FRS和a-FRS对PD与DP的POPF预测价值,以期为临床医师选择合适的预测模型提供理论依据和参考。方法:回顾性收集单中心2018—2019年间行胰腺手术的所有患者的临床资料,经排除标准筛选的入选病例作为研究对象。统计整体与不同手术类型(PD、DP)的POPF发生情况,用受试者工作特征曲线下面积(AUC)分析两种评分模型对整体及不同手术类型的POPF的预测效能。结果:排除不符合的病例后共纳入339例患者,其中193例行PD,146例行DP。全组POPF发生率为17.4%,PD组为18.1%,DP组为16.4%。FRS和a-FRS在全组中预测POPF的能力相似(AUC:0.67 vs.0.65,P=0.412),FRS在PD组中POPF预测价值优于a-FRS(AUC:0.74 vs.0.67,P=0.006),但对DP组的POPF无预测价值(AUC=0.57,95% CI=0.44～0.70,P=0.285),而a-FRS预测DP组POPF的能力好于FRS(AUC:0.66 vs.0.57,P=0.048)。此外,按FRS与a-FRS任何一个模型,POPF的发生率在全组、PD组、DP组的发生率均随着风险等级的上升而增加。FRS的预测因子中,术中失血量和主胰管直径在DP组和PD组间存在明显差异(均P0.05)。结论:FRS和a-FRS均可用于POPF的预测。FRS对PD的POPF预测价值大于a-FRS,但不适用于DP的POPF的预测,而a-FRS对DP的POPF的预测有帮助。术中失血量和主胰管直径是导致FRS对DP的POPF预测效能较低的原因。但由于研究的局限性,结论仍需进一步验证。     

大网膜包裹胰肠吻合口在胰十二指肠切除术中的应用体会

目的 探讨采用大网膜包裹胰肠吻合口方法预防胰十二指肠切除术后胰瘘的效果。方法 回顾性分析江门市中心医院肝胆胰脾外科自2018年1月至2020年12月78例行胰十二指肠切除术患者的临床资料,其中38例胰十二指肠切除联合大网膜包绕胰肠吻合口（包绕组）,40例仅行胰十二指肠切除术（未包绕组）,比较两组术前、术中、术后相关临床指标。结果 两组患者年龄、性别、术前白蛋白、术前总胆红素、胰腺质地、胰管直径和手术时间差异均无统计学意义（均P>0.05）。虽然术后两组的胰漏发生率差异无统计学意义,但包绕组术后胰漏分级均为A级胰漏,未见B、C级胰漏,而未包绕组A级胰漏2例,B级胰漏5例,C级胰漏1例,差异具有统计学意义（P=0.003）;术后包绕组出现2例胃排空延迟,而未包绕组出现1例胆漏,5例术后出血,3例胃排空延迟,除了1例术后腹腔出血需行介入治疗外,其余患者均予以保守治疗而治愈,两组差异无统计学意义（P=0.231）;术后住院时间（1.5±3.0天 vs. 17.4±6.8天,P<0.001）及住院费用（106442±16230 vs. 129831±35875元,P<0.001）包绕组较为包绕组明显减少（P<0.05）。结论 采用大网膜包绕胰肠吻合口方法可以减轻胰十二指肠切除术后胰瘘程度,促进术后胰漏愈合,未增加腹腔感染、腹腔脓肿、胃排空延迟等严重并发症发生,且可减少住院时间及住院费用,且方法简单、易操作。     

胰十二指肠切除术中空肠经十二指肠床消化道重建的临床应用

目的 评价空肠自十二指肠床向上引出行消化道重建方法在胰十二指肠切除术中应用的合理性.方法 回顾性分析我院2007 年7 月至2011 年1 月间实施胰十二指肠切除术中空肠经十二指肠床消化道重建方法的8 例患者的临床资料.结果 患者均顺利完成空肠经十二指肠床消化道重建,无手术死亡病例,无胃排空障碍、无腹腔感染及残留脓肿、肠梗阻等并发症.结论 空肠经十二指肠床消化道重建接近原始的消化道排列方式,符合生理状态,术后引流更通畅,腹腔感染及脓肿、肠梗阻发生率低,是胰十二指肠切除术中的一种安全的消化道重建方法.     

Correlation between Preoperative Imaging and Intraoperative Risk Assessment in the Prediction of Postoperative Pancreatic Fistula Following Pancreatoduodenectomy

Background

Prediction of postoperative pancreatic fistula (POPF) can be carried out with the intraoperative assessment of pancreatic consistency (PC) and via pancreatic duct width (iPDW). Preoperative computed tomography (CT) calculated pancreatic remnant volume (PRV) and duct width (rPDW) have also been shown to offer useful information about the risk of POPF.

Objective

The objective of this study was to determine the predictive value of the preoperative radiological features as compared with the intraoperative risk estimation for the subsequent development of POPF.

Method

All patients undergoing pancreatoduodenectomy between September 2007 and March 2012 at the Karolinska University Hospital Stockholm were included. PRV and rPDW were determined on preoperative CT and in parallel, intraoperative PC and iPDW of the remnant pancreas were independently assessed.

Results

A total of 296 consecutive pancreatoduodenectomies were included. POPF occurred in 45 patients (15.2 %). Of those with a preoperatively calculated PRV < 23.0 cm³, 2.8 % developed POPF compared with 25.7 % of those with a corresponding volume > 46.0 cm³. In patients with an rPDW > 7.0 mm, 4.1 % had a POPF as compared with 38.7 % for those with rPDW < 2.0 mm. The POPF risk estimates based on PRV and rPDW and the intraoperative risk assessments were found to be identical (p < 0.001). In the receiver operating characteristic analysis, area under the curve was 0.80 (95 % confidence interval [CI] 0.72–0.87) and 0.80 (95 % CI 0.72–0.88) for the CT-based and intraoperative risk prediction models, respectively.

Conclusions

Preoperative CT-based and intraoperative gland risk assessments offer comparable predictive information on the risk of POPF after pancreatoduodenectomy. These results imply that accurate POPF risk estimation can be carried out in the preoperative setting to opt for improved patient selection into relevant research protocols and the availability of surgical expertise and techniques.     

基于CT影像组学预测临床ⅠA期老年非小细胞肺癌淋巴结转移

目的探讨基于胸部增强CT影像组学模型预测临床ⅠA期非小细胞肺癌(NSCLC)老年患者淋巴结转移(LNM)的价值。方法回顾性分析361例临床ⅠA期NSCLC老年患者术前增强CT及临床资料。术后病理显示其中87例LNM(LNM组)、274例无LNM(无LNM组),比较2组临床及影像学表现差异。提取术前增强CT影像学特征,进行归一化和降维,采用最小绝对收缩选择算子(LASSO)法筛选最优影像组学特征,建立影像组学模型。按7∶3比例将患者分为训练组和测试组,于训练组中以10次交叉验证法获得最佳影像组学预测模型。根据临床ⅠA期NSCLC老年患者LNM影响因素建立LNM临床预测模型,以之预测训练组和测试组LNM,并以ROC曲线评价2种模型对于训练组和测试组的诊断效能。结果共于所有病灶中提取396个影像组学特征,经归一化后采用LASSO法获得5个最佳影像组学特征建立影像组学模型,并获得最佳影像组学模型,以之预测训练组和测试组LNM的AUC值分别为0.816和0.797,均高于临床模型(0.650和0.686,P均<0.05)。结论基于胸部增强CT的影像组学模型可用于预测临床ⅠA期NSCLC老年患者LNM。     

胰十二指肠切除和胰体尾切除术后胰瘘预测的研究

目的: 探讨2017年中华医学会外科学分会胰腺外科学组的术后胰瘘预测评分的应用价值。方法: 本研究回顾性分析2014年1月至2016年12月我院249胰十二指肠切除术（pancreaticoduodenectomy, PD）及156例胰体尾切除术（distal pancreatectomy, DP）病人的资料。对可能与术后胰瘘发生有关的因素进行分析。根据预测评分中胰腺质地、肿瘤病理、胰管直径以及术中出血量4项指标进行评分。同时应用受试者工作特征（receiver operating characteristic, ROC）曲线分析此评分对病人术后胰瘘发生的灵敏度和特异度。结果: 249例PD病人中31例（12.4%）发生术后胰瘘。单因素分析结果显示,胰腺质地、肿瘤病理、胰管直径以及术中失血量是术后胰瘘发生的危险因素。ROC曲线结果显示,曲线下面积为0.894,P<0.001,95%CI：0.839~0.949。预测评分的灵敏度和特异度分别为83.9%和77.1%。156例DP病人中44例（28.2%）发生术后胰瘘。单因素分析结果显示,胰管直径以及术中出血量是术后胰瘘发生的危险因素。ROC曲线结果显示,曲线下面积为0.567,P=0.190,95%CI：0.467~0.668。预测评分的灵敏度和特异度分别为84.1%和27.7%。结论: 2017年中华医学会外科学分会胰腺外科学组术后胰瘘预测评分能准确地预测PD术后胰瘘的发生,但对DP术后胰瘘的预测没有作用。DP术后胰瘘的预测有待进一步研究。     

Usefulness of Drain Amylase, Serum C-Reactive Protein Levels and Body Temperature to Predict Postoperative Pancreatic Fistula After Pancreaticoduodenectomy

Background

Postoperative pancreatic fistula (POPF) after pancreaticoduodenectomy (PD) is a worrisome and life-threatening complication. Recently, early drain removal has been recommended as a means of preventing POPF. The present study sought to determine how to distinguish clinical POPF from non-clinical POPF in the early postoperative period after PD to aid in early drain removal.

Methods

From March 2002 through December 2010, 176 patients underwent PD and were enrolled in this study to examine factors predictive of clinical POPF after PD. POPF was defined and classified according to the International Study Group of Pancreatic Surgery guideline, and grade B/C POPF was defined as clinical POPF.

Results

Grade A POPF occurred in 39 (22.2 %) patients, grade B in 19 (10.8 %) patients, and grade C in 11 (6.3 %) patients. Clinical POPF (grade B/C) occurred in 17.1 % of patients. Multivariate analysis revealed male gender and body mass index (BMI) ≥22.5 kg/m² to be the independent preoperative risk factors predictive of POPF. Receiver operating characteristic curves showed that the combination of drain amylase ≥750 IU/L, C-reactive protein (CRP) ≥20 mg/dL, and body temperature ≥37.5 °C on postoperative day 3 could effectively distinguish clinical POPF from non-clinical POPF. Sensitivity, specificity, and accuracy were 84.6, 98.2, and 95.7 %, respectively.

Conclusions

Male gender and BMI ≥22.5 were the independent preoperative predictive risk factors for POPF. We assume that when amylase is <750 IU/L, serum CRP is <20 mg/dL, and body temperature is <37.5 °C the drain can safely be removed, even if POPF is indicated.     

Perihematomal edema-based CT-radiomics model to predict functional outcome in patients with intracerebral hemorrhage

PurposeThe purpose of this study was to identify possible association between noncontrast computed tomography (NCCT)-based radiomics features of perihematomal edema (PHE) and poor functional outcome at 90 days after intracerebral hemorrhage (ICH) and to develop a NCCT-based radiomics-clinical nomogram to predict 90-day functional outcomes in patients with ICH.Materials and methodsIn this multicenter retrospective study, 107 radiomics features were extracted from 1098 NCCT examinations obtained in 1098 patients with ICH. There were 652 men and 446 women with a mean age of 60 ± 12 (SD) years (range: 23–95 years). After harmonized and univariable and multivariable screening, seven of these radiomics features were closely associated with the 90-day functional outcome of patients with ICH. The radiomics score (Rad-score) was calculated based on the seven radiomics features. A clinical-radiomics nomogram was developed and validated in three cohorts. The model performance was evaluated using area under the curve analysis and decision and calibration curves.ResultsOf the 1098 patients with ICH, 395 had a good outcome at 90 days. Hematoma hypodensity sign and intraventricular and subarachnoid hemorrhages were identified as risk factors for poor outcomes (P < 0.001). Age, Glasgow coma scale score, and Rad-score were independently associated with outcome. The clinical-radiomics nomogram showed good predictive performance with AUCs of 0.882 (95% CI: 0.859–0.905), 0.834 (95% CI: 0.776–0.891) and 0.905 (95% CI: 0.839–0.970) in the three cohorts and clinical applicability.ConclusionNCCT-based radiomics features from PHE are highly correlated with outcome. When combined with Rad-score, radiomics features from PHE can improve the predictive performance for 90-day poor outcome in patients with ICH.     

CT影像组学模型鉴别诊断肺淋巴瘤与肺浸润性黏液腺癌

目的观察CT影像组学模型鉴别诊断肺淋巴瘤与肺浸润性黏液腺癌(PIMA)的价值。方法回顾性分析经病理证实的34例肺淋巴瘤(淋巴瘤组)及64例PIMA患者(PIMA组),按7∶3比例将其随机分入训练集和验证集。于胸部CT纵隔窗图像上沿病灶边缘手动勾画三维容积感兴趣区(VOI),提取影像组学特征参数。以最小冗余最大相关、LASSO十折交叉验证进行特征降维,以多因素Logistic回归分别构建影像组学标签、影像学特征模型及二者融合的个体化预测模型(以列线图表示)。采用受试者工作特征(ROC)曲线评价各模型对肺淋巴瘤与PIMA的鉴别效能,以决策曲线分析(DCA)综合评价模型的临床效用价值。结果共获得15个影像组学参数用于构建影像组学标签;ROC曲线结果显示其鉴别训练集肺淋巴瘤与PIMA的AUC=0.84,验证集AUC=0.77。以空气支气管征、支气管扩张及胸腔积液构建影像学特征模型,训练集AUC=0.85,验证集AUC=0.81;融合列线图对训练集AUC=0.95,验证集AUC=0.92;列线图具有更高的临床效用价值。结论基于CT征象、影像组学标签构建的个体化预测模型可有效鉴别肺淋巴瘤与PIMA。     

Amylase level after pancreaticoduodenectomy in predicting postoperative pancreatic fistula

BackgroundIt is new clinical interest higher serum amylase level with pancreatitis after pancreaticoduodenectomy (PD) correlates with postoperative pancreatic fistula (POPF). Nevertheless, its evidence and study were scarce. We aimed to investigate correlation of serum amylase level immediate after PD and POPF occurrence.MethodsOf 163 patients who underwent PD at between January 2009 and December 2019, retrospective analysis was conducted to identify risk factors including serum amylase level immediate after PD for POPF occurrence.ResultsOverall incidence of POPF (25/163) was 15.3%. The patients occurred a POPF had significantly higher level of serum amylase on POD0 compared to in whom without a POPF (414 vs 253, p < 0.001). In univariate analysis, ASA classification, post pancreatectomy acute pancreatitis (POAP, serum amylase on POD0 >285IU/L) and Fistula Risk Grade were correlated with POPF occurrence. In multivariable analysis, Fistula risk grade and POAP were significantly associated with developing POPF.ConclusionIn patients with higher serum amylase (>285IU/L) on POD0 with higher fistula risk grade, comprehensive management to achieve mitigation of POPF is important.     

Pancreatico-duodenectomy and postoperative pancreatic fistula: risk factors and technical considerations in a specialized HPB center

Postoperative pancreatic fistula (POPF) is a common complication of pancreatic resection. Aim of this study is to identify variables related to the development of POPF, analyze their clinical significance and discuss our current approach to the pancreatico-jejunal anastomosis. A series of 129 patients undergoing pancreatico-duodenectomy (PD) have been analyzed. Patients were divided in two groups: group F, 26 patients who have developed POPF; group NF, 103 patients who have not developed POPF. Demographic, clinical and intraoperative data were compared. Seventy-six patients had an end-to-side (ES) pancreatico-jejuno anastomosis, 53 an end-to-end (EE) anastomosis. Fifteen patients developed grade A fistulas, seven grade B, and four grade C; two patients with grade C fistula died from septic shock. Preoperative bile duct lithiasis, diameter of the pancreatic duct and consistency of the pancreatic stump were significantly different between F and NF groups at multivariate analysis. POPF has been related to clinical and biological parameters: preoperative bile duct lithiasis and challenging pancreatico-jejunal anastomosis (with small pancreatic duct and friable pancreatic stump) are the most prominent according to our experience. As the incidence of POPF seems to be related to technically demanding surgery, we presently reserve the EE anastomosis to the cases in which a friable gland or a very small duct will make a direct anastomosis on the pancreatic duct unreliable. In case of grade C fistulas a total spleen-preserving pancreatectomy should be considered an adequate treatment to prevent the onset of a multi-organ failure or a septic shock if no other treatment seems suitable.     

基于CT的直肠癌新辅助化疗后病理完全缓解预测模型的初步探索--DACCA数据库的联合研究

目的初步探索基于CT影像组学特征建立的决策树模型对局部进展期直肠癌(LARC)患者行单纯新辅助化疗疗效的评估价值。方法回顾性分析四川大学华西医院肠癌数据库(DACCA)中2016年10月至2019年3月期间符合本研究纳入和排除标准的244例单纯新辅助化疗后行根治性手术的LARC患者的临床及CT检查的DICOM格式图像资料。利用ITK-SNAP软件选取肿瘤最大层面并对影像感兴趣区域进行勾画。使用计算机随机分配软件将200例患者纳入训练集,44例患者纳入测试集。利用MATLAB软件读取DICOM格式图像并提取和筛选影像组学特征,进而用降维后得到的影像组学特征进行机器学习并建模。通过绘制受试者操作特征曲线并计算曲线下面积(AUC)来评估模型预测单纯新辅助化疗后对病理完全缓解(pCR)的效能。结果根据术后病理肿瘤退缩分级(TRG)患者被分为pCR组(TRG0,28例)和非pCR组(TRG1~TRG3,216例)。最终获得13个影像组学特征即6个灰度特征(均值、方差、标准差、偏离度、峰态、能量)、3个纹理特征(对比度、相关性、同质化)及4个形状特征(边界长度、直径、面积、形状参数)。基于CT的决策树模型预测LARC患者新辅助化疗后效果的AUC值为0.772 [95%CI(0.656,0.888)],对于非pCR的预测准确度较高(97.2%),但对于pCR的预测准确度较低(57.1%)。结论在本次初步探索中,基于CT的决策树模型在判断LARC患者新辅助化疗后pCR上较同质研究的预测效能低,通过分析后将进一步从优化算法、继续拓展数据集、挖掘更多影像组学特征值等方面优化模型,从而最终实现pCR的精准预测。     

基于术前MRI影像组学特征预测肝癌经肝动脉化疗栓塞治疗的反应性

目的 探索术前MRI影像组学特征预测肝细胞癌（hepatocellular carcinoma,HCC）对经肝动脉化疗栓塞（transcatheter arterial chemoembolization,TACE）治疗反应性的可行性。方法 回顾性收集2016年3月至2018年4月经临床或病理确诊、并在温州医科大学附属第五医院接受TACE治疗的中晚期HCC患者86例,采用随机数字表法分为训练组（n=59）及验证组（n=27）,基于ITK-SNAP软件在术前T2WI图像勾画肿瘤感兴趣区,使用GE AI-Kit软件提取影像特征;经最小绝对收缩与选择算子（LASSO）回归筛选最优的特征子集,并构建TACE反应性预测模型;利用受试者工作特征（ROC）曲线、校准曲线及决策曲线分析评价模型的诊断效能和临床应用价值。结果 共筛选出8个影像学特征与TACE反应性相关,并成功构建预测模型。训练组中,该模型的ROC曲线下面积（AUC）为0.838（95%CI 0.737～0.939）,灵敏度为74.1%,特异度为84.4%;在验证组中,AUC为0.794（95%CI 0.604～0.985）,灵敏度为75.0%,特异度为80.0%。校准曲线显示该模型在训练组和验证组的预测概率与实际概率拟合较好;决策曲线分析显示该模型在0.1～1.0的阈值范围内具有较高的净收益。结论 术前MRI T2WI图像影像组学特征可用于预测HCC的TACE 反应性,基于MRI T2WI图像影像特征的预测模型特异度和灵敏度较高。