术前B型利钠肽水平预测冠状动脉旁路移植术后新发心房颤动准确性的系统评价

目的评价术前B型利钠肽(brain natriuretic peptide,BNP)水平在预测冠状动脉旁路移植术(coronary artery bypass grafting,CABG)后新发心房颤动(atrial fibrillation,AF)中的价值。方法计算机检索PubMed、EMbase、Cochrane图书馆、中国期刊全文数据库(CNKI)及维普(VIP)数据库,筛选有关CABG术前BNP与术后AF发生相关性的文献,检索时间从建库至2012年11月。根据诊断性试验准确性质量评价(QUADAS)标准评价纳入文献质量。采用RevMan 5.0进行异质性检验,Meta-Disc1.4进行Meta分析,对纳入文献进行加权定量合并,计算汇总敏感度、特异度、诊断比值比、阳性似然比和阴性似然比及其95%可信区间(95%CI),绘制汇总受试者工作特征曲线(SROC),并计算ROC曲线下面积(AUC)。结果检出相关文献236篇,根据纳入标准最终纳入5篇文献;共纳入患者802例,其中术后发生AF 228例,未发生AF 574例。纳入文献质量均较高。术前增加的BNP浓度与术后AF发生的诊断比值比为4.15[95%CI(2.90,5.95)];汇总敏感度为0.78[95%CI(0.72,0.83)],汇总特异度为0.58[95%CI(0.54,0.58)],汇总阳性似然比为1.91[95%CI(1.42,1.56)],汇总阴性似然比为0.42[95%CI(0.32,0.54)];AUC为0.79(Q值为0.72)。结论 CABG术前BNP水平升高与术后新发AF具有明显的相关性,是术后AF发生的强预测因子,可以在一定程度上预测术后AF的发生。     

超声引导下穿刺活检诊断甲状腺结节良恶性Meta分析

目的采用Meta分析评价超声引导下穿刺组织学活检对甲状腺结节良恶性的诊断价值。方法检索PubMed和中国国家知识基础设施(CNKI)、万方数据库中关于超声引导下穿刺活检诊断甲状腺结节的中、英文文献。应用诊断性试验准确性评价(QUADAS)工具评价文献质量。采用Stata 11.0统计软件进行数据分析,获得相关的敏感度、特异度、阳性似然比、阴性似然比、诊断优势比,绘制汇总受试者工作特性(SROC)曲线,计算曲线下面积(AUC)。结果共纳入文献14篇。合并敏感度、特异度分别为0.96[95%CI(0.93,0.98)]、0.98[95%CI(0.96,0.99)],合并阳性似然比、阴性似然比、诊断优势比分别为47.78[95%CI(26.53,86.05)]、0.04[95%CI(0.02,0.07)]、1 151.12[95%CI(487.95,2 715.62)]。SROC曲线的AUC为0.99[95%CI(0.98,1.00)]。结论超声引导下穿刺活检诊断甲状腺结节良恶性具有较高敏感度和特异度。     

三维子宫输卵管超声造影评估输卵管通畅性：Meta分析

目的采用Meta分析评价三维子宫输卵管超声造影(3D HyCoSy)评价输卵管通畅性的价值。方法检索1994—2013年国内外公开发表的有关3DHyCoSy的中外文文献,按照诊断性试验的纳入和排除标准筛选符合要求的文献,提取纳入研究的相关信息进行Meta分析。结果共纳入11项独立研究,其中3DHyCoSy诊断输卵管通畅性的敏感度范围为0.67~1.00,特异度为0.77~1.00,合并敏感度和特异度分别为0.95(95%CI 0.93~0.96)、0.92(95%CI0.89~0.94),合并阳性和阴性似然比分别为9.28(95%CI 5.91~14.58)、0.08(95%CI 0.05~0.13),SROC曲线下面积为0.98。结论 3DHyCoSy用于评价输卵管通畅性具有高度临床价值。     

超声造影判定浸润与非浸润性膀胱癌的Meta分析

目的采用Meta分析评价CEUS对浸润与非浸润膀胱癌的诊断价值。方法检索PubMed、EBSCO、万方、CBM等数据库收集国内外已发表的相关文献,通过Meta分析对资料进行汇总分析,并绘制总体ROC(SROC)曲线,计算曲线下面积(AUC)。结果符合纳入标准的文献4篇合并敏感度、特异度、阳性预测值、阴性预测值、诊断比数比及其95%可信区间分别为0.95(0.87,0.99)、0.93(0.86,0.97)、13.08(6.25,27.34)、0.06(0.03,0.16)、226.75(57.31,897.10)。SROC曲线AUC为0.98。结论通过CEUS可较准确地鉴别浸润与非浸润性膀胱癌。     

WB-MRI与骨扫描诊断恶性肿瘤骨转移：Meta分析

目的采用Meta分析评价WB-MRI与99 Tcm-亚甲基二磷酸盐骨扫描诊断恶性肿瘤骨转移的价值。方法检索PubMed、EMbase、Web of Science、Cochrane图书馆、中国期刊全文数据库、中国科技期刊数据库(维普)、万方数字化期刊全文数据库中关于比较WB-MRI和骨扫描诊断恶性肿瘤骨转移能力的中英文文献,采用诊断性试验质量评价方法评价纳入文献质量,以Meta-Disc 1.4软件进行Meta分析。结果共纳入9篇文献、447例患者。WB-MRI诊断恶性肿瘤骨转移的合并敏感度和特异度分别为0.91(95%CI 0.86~0.95)和0.95(95%CI 0.92~0.97),SROC曲线下面积为0.9797;骨扫描诊断恶性肿瘤骨转移的合并敏感度和特异度分别为0.80(95%CI 0.73~0.85)和0.89(95%CI 0.85~0.93),SROC曲线下面积为0.9183。结论与99 Tcm-亚甲基二磷酸盐骨扫描相比,WB-MRI诊断恶性肿瘤骨转移敏感度和特异度更高。     

尿液中LncRNA UCA1检测对膀胱癌诊断价值的Meta分析

目的:通过Meta分析文献中已经报道的尿液中长链非编码RNA尿路上皮癌抗原1(LncRNA UCA1)对膀胱癌的诊断价值,探讨尿液中LncRNA UCA1对膀胱癌的诊断价值。方法:采用计算机检索PubMed、中国知网、万方和维普等数据库,按照已定的纳入和排除标准筛选文献,利用Meta-DiSc软件分别计算LncRNA UCA1检查组和尿细胞学(UC)检查组灵敏度、特异度、似然比和诊断性比数比。绘制总受试者工作特征曲线(SROC),计算曲线下面积,估计总诊断精确度。结果:共检出62篇相关文献,5篇文献纳入本次Meta分析,共1 066例,通过合并文献得出,LncRNA UCA1对膀胱癌的诊断总灵敏度为0.82(95%CI:0.76~0.88),特异度为0.87(95%CI:0.78~0.97),SROC的曲线下面积为0.905;UC检查检测膀胱癌总灵敏度为0.61(95%CI:0.57~0.65),特异度为0.95(95%CI:0.93~0.97),SROC的曲线下面积为0.91。LncRNA UCA1检查在总体诊断效能上与UC检查差异无统计学意义(P0.05)。结论:目前研究证据表明,LncRNA UCA1对膀胱癌具有较好的诊断价值,其对膀胱癌的诊断灵敏度高于细胞学检查,但其特异度低于细胞学检查,所以尚不能证实LncRNA UCA1检测的诊断价值高于细胞学检查。     

平衡式稳态自由进动序列MRA诊断肾动脉狭窄:Meta分析

目的采用Meta分析方法评价平衡式稳态自由进动序列(B-SSFP)MRA对肾动脉狭窄(RAS)的诊断价值。方法计算机检索PubMed、EMabase、Cochrane Library数据库中关于B-SSFP诊断RAS的英文文献。以诊断准确性研究质量评价工具-2评价文献质量。采用Meta Disc 1.4和Stata 12.0统计分析软件,绘制汇总受试者工作特征(SROC)曲线,并计算AUC及合并敏感度、特异度、诊断比值比。结果最终纳入的12篇文献21组数据存在高度异质性,采用随机效应模型。Meta分析结果显示,B-SSFP诊断RAS的合并敏感度、特异度分别为0.85[95%CI(0.81,0.88),I~2=56.8%]和0.96[95%CI(0.95,0.97),I~2=85.3%],合并SROC曲线的AUC为0.964 4,合并诊断比值比为137.94[95%CI(84.75,224.50)]。回归分析结果显示异质性可能来源于发表年(P=0.038)。结论 B-SSFP对RAS诊断具有较高敏感度和特异度,有望成为肾功能不全患者的首选检查方法。     

磁共振成像对女性盆腔肿块诊断价值的Meta分析

目的应用Meta分析探讨磁共振成像(MRI)对女性盆腔肿块的诊断价值。方法计算机检索近20年来PubMed、Cochrane Database Systematic Reviews、EMbase、中国知网(CNKI)检索平台以及万方数据库中MRI对女性盆腔肿块诊断价值的病例对照研究文献。研究者对文献质量进行严格评价和资料提取,提取符合质量标准的文献中的诊断数据用Metadisc 1.40软件进行Meta分析,以术后病理诊断为金标准,分别合并MRI及B超诊断女性盆腔肿块的灵敏度、特异度、阴性似然比和阳性似然比,采用拟合受试者工作特征曲线(SROC)分析两者的诊断价值,并对二者的各项数据进行比较。结果 10个自身病例对照研究共649例患者729个肿块纳入研究。Meta分析结果显示:MRI诊断女性盆腔肿块的汇总敏感度(95%CI)和汇总特异度(95%CI)分别为〔89%(84%～92%),P=0.046 6〕和〔87%(83%～90%),P=0.000 2〕,汇总阳性似然比和汇总阴性似然比分别为6.25(P=0.008 5)和0.14(P=0.029 1),SROC曲线下面积(AUC)为0.941;B超的汇总敏感度(95%CI)和汇总特异度(95%CI)分别为〔87%(82%～91%),P=0.000 0〕和〔73%(69%～77%),P=0.000 0〕,汇总阳性似然比和汇总阴性似然比分别为3.07(P=0.000 0)和0.18(P=0.000 1),SROC曲线下的面积为0.897。MRI的汇总特异度、阳性似然比和SROC曲线下的面积明显高于B超。结论 MRI对女性盆腔肿块诊断的准确性明显优于B超,MRI应成为怀疑存在盆腔肿块女性首选的影像学检查手段。     

血清高尔基体蛋白73对乙型肝炎病毒相关性肝癌诊断价值的Meta分析

目的:探讨血清高尔基体蛋白73(GP73)对乙型肝炎病毒(HBV)相关性肝癌的诊断价值。方法:计算机系统性检索国内外公开发表的相关文献。采用Meta-Disc1.4软件汇总诊断准确性指标,计算综合受试者工作特征曲线(SROC)、曲线下面积(AUC)和Q*指数。采用Stata14.0软件计算验后概率,并绘制Fagan图来评估GP73的临床应用价值。结果:最终纳入14篇文献,总样本量为3953例。血清GP73诊断HBV相关性肝癌的总敏感度为0.70(95%CI=0.68~0.73),总特异度为0.75(95%CI=0.73~0.77),总阳性似然比为3.37(95%CI=2.75~4.13),总阴性似然比为0.36(95%CI=0.29~0.45),诊断比值比(DOR)为9.67(95%CI=6.72~13.91),AUC和Q*指数分别为0.826和0.759。当设定验前概率为50%时,阳性、阴性GP73的验后概率分别为78%和26%。结论:血清GP73对HBV相关性肝癌有一定的诊断意义,但其单独应用于确诊或排除该病的价值有限,尚需结合其他检查。     

3D子宫输卵管超声造影评估输卵管通畅性的meta分析(原稿号Y201310054)

目的　采用Meta分析评价三维子宫输卵管超声造影((3D HyCoSy) )对输卵管通畅性的诊断价值。方法　检索1994—2013年国内外公开发表的有关3D HyCoSy)的中外文文献,按照诊断性试验的纳入和排除标准筛选符合要求的文献,提取纳入研究的相关信息进行Meta分析。结果　共纳入11项独立研究,各研究中3D HyCoSy 诊断输卵管通畅性的敏感度范围为0.67~1.00,特异度为0.77～1.00。各研究的合并灵敏度和特异度分别为0.95（95%CI 0.93～0.96）、0.92（95%CI 0.89～0.94）;合并阳性和阴性似然比分别为9.28（95%CI 5.91～14.58）、0.08（95%CI 0.05～0.13）;SROC曲线下面积为0.98?.01。结论　3D HyCoSy对评价输卵管通畅性具有高度的临床诊断价值。     

Comparative effectiveness of coronary screening in heart valve surgery: Computed tomography versus conventional coronary angiography

Background

Although conventional coronary angiography (CAG) is considered the gold standard for coronary artery disease (CAD) screening in the setting of heart valve surgery, coronary artery computed tomography angiography (CCTA) has emerged as an alternative modality. This study was conducted to evaluate the clinical outcomes of CCTA compared with conventional CAG for CAD screening in patients undergoing heart valve surgery.

Twenty-year outcomes of coronary artery bypass grafting utilizing 3 in situ arterial grafts

ObjectiveThe added value of total arterial revascularization in coronary artery bypass grafting becomes particularly apparent when evaluating long-term results. We previously reported on our 10-year outcomes of total arterial revascularization using bilateral internal thoracic and gastroepiploic arteries as in situ grafts in patients with 3-vessel disease. This study aimed to increase the follow-up period to 20 years.MethodsWe updated clinical outcomes of 201 patients operated on between 1992 and 2002. At that time, the technique was primarily performed in patients with a longer life expectancy. Primary end points were overall survival and freedom from the composite of major adverse cardiac events. Secondary end points were the separate cardiac events.ResultsExtended follow-up included all patients. The median follow-up time was 19.2 years (interquartile range, 16.2-20.0). The respective 15- and 20-year Kaplan-Meier estimated survival probabilities were 73.9% (95% confidence interval [CI], 67.2%-79.5%) and 63.5% (95% CI, 55.7%-70.4%) for overall survival and 57.9% (95% CI, 50.7%-64.5%) and 47.9% (95% CI, 40.1%-55.3%) for freedom from major adverse cardiac events. The respective estimated cumulative incidences at 15 and 20 years were 7.0% (95% CI, 3.5%-10.6%) and 7.8% (95% CI, 4.0%-11.6%) for myocardial infarction, 8.6% (95% CI, 4.7%-12.5%) and 9.3% (95% CI, 5.2%-13.3%) for percutaneous reintervention, 7.0% (95% CI, 3.5%-10.5%) and 7.0% (95% CI, 3.5%-10.5%) for reoperation, 8.6% (95% CI, 4.7%-12.6%) and 12.9% (95% CI, 7.6%-18.2%) for cardiac death, and 10.8% (95% CI, 6.5%-15.2%) and 15.2% (95% CI, 9.8%-20.6%) for death from other causes.ConclusionsThe use of in situ bilateral internal thoracic and gastroepiploic arteries provides outstanding 15- and 20-year survival and cardiac event-free survival probabilities. Further studies are needed in older patients with more severe comorbidities. Nevertheless, the results from this and scarce other studies on 15- to 20-year outcomes of total arterial revascularization suggest that cardiac surgeons should embrace the application of total arterial grafting to further reduce the risks of long-term cardiac events, especially during the second decade after surgery.     

Impact of coronary calcification assessed by coronary CT angiography on treatment decision in patients with three-vessel CAD: insights from SYNTAX III trial

 Open in a separate windowOBJECTIVESThe aim of this study was to determine Syntax scores based on coronary computed tomography angiography (CCTA) and invasive coronary angiography (ICA) and to assess whether heavy coronary calcification significantly limits the CCTA evaluation and the impact of severe calcification on heart team’s treatment decision and procedural planning in patients with three-vessel coronary artery disease (CAD) with or without left main disease.METHODSSYNTAX III was a multicentre, international study that included patients with three-vessel CAD with or without left main disease. The heart teams were randomized to either assess coronary arteries with coronary CCTA or ICA. We stratified the patients based on the presence of at least 1 lesion with heavy calcification defined as arc of calcium >180° within the lesion using CCTA. Agreement on the anatomical SYNTAX score and treatment decision was compared between patients with and without heavy calcifications.RESULTSOverall, 222 patients with available CCTA and ICA were included in this trial subanalysis (104 with heavy calcification, 118 without heavy calcification). The mean difference in the anatomical SYNTAX score (CCTA derived—ICA derived) was lower in patients without heavy calcifications [mean (−1.96 SD; +1.96 SD) = 1.5 (−19.3; 22.4) vs 5.9 (−17.5; +29.3), P = 0.004]. The agreement on treatment decision did not differ between patients with (Cohen’s kappa 0.79) or without coronary calcifications (Cohen’s kappa 0.84). The agreement on the treatment planning did not differ between patients with (concordance 80.3%) or without coronary calcifications (concordance 82.8%).CONCLUSIONSAn overall good correlation between CCTA- and ICA-derived Syntax score was found. The presence of heavy coronary calcification moderately influenced the agreement between CCTA and ICA on the anatomical SYNTAX score. However, agreement on the treatment decision and planning was high and irrespective of the presence of calcified lesions.     

Evaluation of a deep learning model on coronary CT angiography for automatic stenosis detection

PurposeThe purpose of this study was to evaluate a deep-learning model (DLM) for classifying coronary arteries on coronary computed tomography -angiography (CCTA) using the Coronary Artery Disease-Reporting and Data System (CAD-RADS).Materials and methodsThe DLM was trained with 10,800 curved multiplanar reformatted (cMPR) CCTA images classified by an expert radiologist using the CAD-RADS. For each of the three main coronary arteries, nine cMPR images 40° apart acquired around each arterial circumference were then classified by the DLM using the highest probability. For the validation set composed of 159 arteries from 53 consecutive patients, the images were read by two senior and two junior readers; consensus of the two seniors was the reference standard. With the DLM, the majority vote for the nine images was used to classify each artery. Three groups (CAD-RADS 0, 1–2, or 3–4–5) and 2 groups CAD-RADS 0–1–2 or 3–4–5 (<50% vs. ≥50% stenosis) were used for comparisons with readers and consensus. Performance of the model and readers was compared to the consensus reading using the intraclass coefficient (ICC) and Cohen's kappa coefficient at the artery and patient levels.ResultsWith the three groups at the artery level, the ICC of the DLM was 0.82 (95% CI: 0.75–0.88) and not significantly different from those of 3/4 readers; accuracy was 81%. With the binary classification, Cohen kappa coefficient of the DLM was 0.85 (95% CI: 0.69–0.94) and not significantly different from that of 3/4 readers; accuracy was 96%. At the patient level, sensitivity and specificity were 93% and 97% respectively, and the negative predictive value was 97%.ConclusionThe DLM detected ≥50% stenoses with performances similar to those achieved by senior radiologists.     

Clinical Course of Pain and Function Following Total Knee Arthroplasty: A Systematic Review and Meta-Regression

BackgroundTotal knee arthroplasty (TKA) is widely considered a successful intervention for osteoarthritis and other degenerative knee diseases. This study addresses the need for a high-quality meta-analysis that outlines the clinical course of pain and function post-TKA.MethodsThe review included prospective cohort studies assessing pain or function of patients undergoing primary TKA at baseline (preoperatively) and at least 2 additional time points including one at least 12 months postoperatively. Two reviewers independently screened references, extracted data, and assessed risk of bias using the Quality in Prognosis Studies tool. The time course of recovery of pain and function was modeled using fractional polynomial meta-regression.ResultsIn total, 191 studies with 59,667 patients were included, most with low risk of bias. The variance-weighted mean pain score (/100, 0 = no pain) was 64.0 (95% confidence interval [CI] 60.2-67.7) preoperatively, 24.1 (95% CI 20.3-27.9) at 3 months, 20.4 (95% CI 16.7-24.0) at 6 months, and 16.9 (95%CI 13.6-20.3) at 12 months, and remained low (10.1; 95% CI 4.8-15.4) at 10 years postoperatively. The variance-weighted mean function score (/100, 0 = worst function) was 47.1 (95% CI 45.7-48.4) preoperatively, 72.8 (95% CI 71.3-74.4) at 3 months, 76.3 (95% CI 74.7-77.8) at 6 months, and 78.1 (95%CI 76.4-79.7) at 12 months. Function scores were good (79.7; 95% CI 77.9-81.5) at 10 years postoperatively.ConclusionPatients undergoing primary TKA can expect a large and rapid but incomplete recovery of pain and function in the first postoperative year. At 10 years, the gains in pain scores may still remain while there is an improvement in function.     

Sex modifies the association between HIV and coronary artery disease among older adults in Uganda

IntroductionLittle is known about the epidemiology of coronary artery disease (CAD) in sub‐Saharan Africa, where the majority of people living with HIV (PLHIV) live. We assessed the association of HIV with CAD and explored relationships with monocyte activation in sex‐stratified analyses of older PLHIV and people without HIV (PWOH) in Uganda.MethodsThe Ugandan Study of HIV effects on the Myocardium and Atherosclerosis (mUTIMA) follows 100 PLHIV on antiretroviral therapy (ART) and 100 age‐ and sex‐matched PWOH controls in Kampala, Uganda; all >45 years of age with >1 cardiovascular disease risk factor. At the year 2 exam (2017–2019), 189 participants had available coronary calcium score and 165 had coronary CT angiography (CCTA) for this analysis. A subset of participants (n = 107) had both CCTA and fresh whole blood flow cytometry for monocyte phenotyping.ResultsMedian age was 57.8 years and 63% were females. Overall, 88% had hypertension, 37% had diabetes and 4% were smokers. Atherosclerotic cardiovascular disease (ASCVD) risk was modestly higher for PWOH, but not statistically significant (median 10‐year ASCVD risk 7.2% for PLHIV vs. 8.6% for PWOH, p = 0.09). Median duration of ART was 12.7 years and 86% had suppressed viral load. Despite a high prevalence of risk factors, only 34/165 (21%, 95% CI 15–28%) had any coronary plaque. After adjustment for ASCVD risk score, HIV status was not associated with CAD (OR 0.55, 95% CI 0.23–1.30) but was associated with more severe CAD (segment severity score>3) among those with disease (OR 10.9, 95% CI 1.67–70.45). Females had a trend towards higher odds of CAD among PLHIV (OR 4.1, 95% CI 0.4–44.9), but a trend towards lower odds of CAD among PWOH (OR 0.30; 95% CI 0.07–1.3; HIV*sex interaction p = 0.019). CAD was positively correlated with classical monocytes (r = 0.3, p = 0.012) and negatively correlated with CX3CR1 expression (r = –0.31, p = 0.011) in PLHIV and negatively correlated with patrolling monocytes (r = –0.36, p = 0.031) and tissue factor expression (r = –0.39, p = 0.017) in PWOH.ConclusionsOur results suggest that HIV may be associated more with severity rather than the presence of CAD in Uganda. Sex differences in the HIV effect suggest that tailored CAD prevention strategies may be required in this setting.     

Performance of deep learning for differentiating pancreatic diseases on contrast-enhanced magnetic resonance imaging: A preliminary study

PurposeThe purpose of this study was to evaluate the ability of deep learning to differentiate pancreatic diseases on contrast-enhanced magnetic resonance (MR) images with the aid of generative adversarial network (GAN).Materials and MethodsA total of 504 patients who underwent T1-weighted contrast-enhanced MR examinations before any treatments were included in this retrospective study. First, the MRI examinations of 398 patients (215 men, 183 women; mean age, 59.14 ± 12.07 [SD] years [range: 16-85 years]) from one hospital were used as the training set. Then the MRI examinations of 50 (26 men, 24women; mean age, 58.58 ± 13.64 [SD] years [range: 24–85 years]) and 56 (30 men, 26 women; mean age, 59.13 ± 11.35 [SD] years [range: 26–80 years]) consecutive patients from two hospitals were separately collected as the internal and external validation sets. An InceptionV4 network was trained on the training set augmented by synthetic images from GANs. Classification performance of trained InceptionV4 network for every patch and every patient were made on both validation sets, respectively. The prediction agreement between convolutional neural network (CNN) and radiologist was measured by the Cohen's kappa coefficient.ResultsThe patch-level average accuracy and the micro-averaging area under receiver operating characteristic curve (AUC) of InceptionV4 network were 71.56% and 0.9204 (95% confidence interval [CI]: 0.9165–0.9308) for the internal validation set, and 79.46% and 0.9451 (95%CI: 0.9320–0.9523) for the external validation set, respectively. The patient-level average accuracy and the micro-averaging AUC of InceptionV4 network were 70.00% and 0.8250 (95%CI: 0.8147–0.8326) for the internal validation, 76.79% and 0.8646 (95%CI: 0.8489–0.8772) for the external validation set, respectively. Evaluated by human reader, the average accuracy and micro-averaging AUC for internal and external validation sets were 82.00% and 0.8950 (95%CI: 0.8817–0.9083), 83.93% and 0.9063 (95%CI: 0.8968–0.9212), respectively. The Cohen's kappa coefficients between InceptionV4 network and human reader for the internal and external invalidation sets were 0.8339 (95%CI: 0.6991–0.9447) and 0.8862 (95%CI: 0.7759–0.9738), respectively.ConclusionDeep learning using CNN and GAN had the potential to differentiate pancreatic diseases on contrast-enhanced MR images.     

MRI-based traditional radiomics and computer-vision nomogram for predicting lymphovascular space invasion in endometrial carcinoma

PurposeTo determine the capabilities of MRI-based traditional radiomics and computer-vision (CV) nomogram for predicting lymphovascular space invasion (LVSI) in patients with endometrial carcinoma (EC).Materials and methodsA total of 184 women (mean age, 52.9 ± 9.0 [SD] years; range, 28–82 years) with EC were retrospectively included. Traditional radiomics features and CV features were extracted from preoperative T2-weighted and dynamic contrast-enhanced MR images. Two models (Model 1, the radiomics model; Model 2, adding CV radiomics signature into the Model 1) were built. The performance of the models was evaluated by the area under the curve (AUC) of the receiver operator characteristic (ROC) in the training and test cohorts. A nomogram based on clinicopathological metrics and radiomics signatures was developed. The predictive performance of the nomogram was assessed by AUC of the ROC in the training and test cohorts.ResultsFor predicting LVSI, the AUC values of Model 1 in the training and test cohorts were 0.79 (95% confidence interval [CI]: 0.702–0.889; accuracy: 65.9%; sensitivity: 88.8%; specificity: 57.8%) and 0.75 (95% CI: 0.585–0.914; accuracy: 69.5%; sensitivity: 85.7%; specificity: 62.5%), respectively. The AUC values of Model 2 in the training and test cohorts were 0.93 (95% CI: 0.875–0.991; accuracy: 94.9%; sensitivity: 91.6%; specificity: 96.0%) and 0.81 (95% CI: 0.666–0.962; accuracy: 71.7%; sensitivity: 92.8%; specificity: 62.5%), respectively. The discriminative ability of Model 2 was significantly improved compared to Model 1 (Net Reclassification Improvement [NRI] = 0.21; P = 0.04). Based on histologic grade, FIGO stage, Rad-score and CV-score, AUC values of the nomogram to predict LVSI in the training and test cohorts were 0.98 (95% CI: 0.955–1; accuracy: 91.6%; sensitivity: 91.6%; specificity: 96.0%) and 0.92 (95% CI: 0.823–1; accuracy: 91.3%; sensitivity: 78.5%; specificity: 96.8%), respectively.ConclusionsMRI-based traditional radiomics and computer-vision nomogram are useful for preoperative risk stratification in patients with EC and may facilitate better clinical decision-making.