分享:
分享到微信朋友圈
X
临床研究
基于磁共振T2WI影像组学预测急性胰腺炎后糖尿病的价值
杜青林 黄小华 刘念 陈雨薇 钟舒婷 刘琢玉

Cite this article as: DU Q L, HUANG X H, LIU N, et al. The value of radiomics based on MRI T2WI in predicting the post-acute pancreatitis diabetes mellitus[J]. Chin J Magn Reson Imaging, 2023, 14(7): 67-72.本文引用格式:杜青林, 黄小华, 刘念, 等. 基于磁共振T2WI影像组学预测急性胰腺炎后糖尿病的价值[J]. 磁共振成像, 2023, 14(7): 67-72. DOI:10.12015/issn.1674-8034.2023.07.012.


[摘要] 目的 探究基于磁共振T2WI序列的影像组学模型在急性胰腺炎后糖尿病(post-acute pancreatitis diabetes mellitus, PPDM-A)中的预测价值。材料与方法 回顾性分析2016年1月至2019年12月在本院确诊为急性胰腺炎(acute pancreatitis, AP)的患者病例,经临床随访后得到PPDM-A组57例和AP后血糖正常组83例。将入组患者按7∶3的比例随机分为训练组98例(PPDM-A组40例,AP后血糖正常组58例)和验证组42例(PPDM-A组17例,AP后血糖正常组25例),并同期收集两组相关临床特征。利用3D Slicer软件勾画胰腺实质边缘并提取特征,采用方差阈值法、K最佳选择法(select K Best)、最小绝对收缩和选择算子(least absolute shrinkage and selection operator, LASSO)算法实现对影像组学特征的降维、筛选,利用随机森林方法建立预测PPDM-A的临床模型、影像组学模型及联合模型,通过受试者工作特征(receiver operating characteristic, ROC)曲线下面积(area under the curve, AUC)评估模型的预测性能,利用DeLong检验比较模型间的预测效能,绘制最优模型的校准曲线并采用Hosmer-Lemesow检验验证模型的拟合优度,采用决策曲线分析(decision curve analysis, DCA)评估各个模型的临床价值。结果 临床模型、影像组学模型及联合模型在训练组中的AUC分别为0.773 [95%置信区间(confidence interval, CI):0.679~0.866]、0.831(95% CI:0.750~0.912)和0.881(95% CI:0.816~0.946),在验证组中的AUC分别为0.664(95% CI:0.488~0.839)、0.821(95% CI:0.684~0.958)和0.899(95% CI:0.806~0.992),DeLong检验结果显示,联合模型与临床模型的预测效能在训练组和验证组中的差异均存在统计学意义(训练组:P=0.024、验证组:P=0.013);联合模型与影像组学模型在训练组中的差异存在统计学意义(P=0.047),在验证组中的差异无统计学意义(P=0.090)。Hosmer-Lemeshow检验显示联合模型校正良好(P=0.250)。DCA结果显示,影像组学模型、联合模型预测PPDM-A的临床净收益均优于临床模型,当阈值概率大于12%时,联合模型的临床净收益优于影像组学模型。结论 基于临床特征及T2WI影像组学特征构建的联合模型预测效能较好,可作为早期预测PPDM-A发生的方法。
[Abstract] Objective To evaluate the value of radiomics analysis based on T2WI sequence MRI in predicting post-acute pancreatitis diabetes mellitus (PPDM-A). Methods and Materials: Retrospective collection of patients diagnosed with acute pancreatitis (AP) in our hospital from January 2016 to December 2019, through clinical follow-up, they were divided into PPDM-A group (n=57) and normal blood glucose after AP group (n=83). The enrolled patients were randomly divided into the training group (n=98, 40 cases in the PPDM-A group, 58 cases in the normal blood glucose group after AP) and the validation group (n=42, 17 cases in the PPDM-A group, 25 cases in the normal blood glucose group after AP) at a ratio of 7∶3, and the relevant clinical characteristics of the two groups were collected at the same time. 3D Slicer software was used to delineate the edge of pancreatic parenchyma and extract features, variance threshold method, select K Best method and least absolute shrinkage and selection operator (LASSO) were used to extract the final radiomics features. Finally, random forest method was used to establish the clinical model, radiomics model and combined model for predicting PPDM-A. The receiver operating characteristic (ROC) curve was drawn to evaluate the predictive performance of the model, and DeLong test was used to compare the predictive efficiency of these models. The calibration curve of the best model was drawn and the Hosmer-Lemesow test was used to verify the goodness of fit of the model. The decision curve analysis (DCA) was used to evaluate the net clinical benefits of each model.Results The AUCs of the clinical model, radiomics model, and combined model in the training group were 0.773 [95% confidence interval (CI): 0.679-0.866], 0.831 (95% CI: 0.750-0.912) and 0.881 (95% CI: 0.816-0.946), respectively. In the validation group, they were 0.664 (95% CI: 0.488-0.839), 0.821 (95% CI: 0.684-0.958) and 0.899 (95% CI: 0.806-0.992), respectively. DeLong test results showed that the prediction performance of the combined model was statistically different from that of the clinical model in both the training and validation groups (training group: P=0.024, validation group: P=0.013). In the training group, there was a significant difference between the combined model and the radiomics model (P=0.047), but no significant difference was observed in the validation group (P=0.090). The Hosmer-Lemeshow test showed that the combined model was well calibrated (P=0.250). The DCA results indicated that both the radiomics model and the combined model had better net clinical benefits in predicting PPDM-A compared to the clinical model. Moreover, when the threshold probability was greater than 12%, the net clinical benefit of the combined model was superior to that of the radiomics model.Conclusions The combined model based on clinical features and T2WI radiomics features has a good predictive performance and can be used as an early method to predict the occurrence of PPDM-A.
[关键词] 影像组学;磁共振成像;急性胰腺炎;胰源性糖尿病;预测
[Keywords] radiomics;magnetic resonance imaging;acute pancreatitis;pancreatic diabetes mellitus;prediction

杜青林    黄小华 *   刘念    陈雨薇    钟舒婷    刘琢玉   

川北医学院附属医院放射科,南充 637000

通信作者:黄小华,E-mail:15082797553@163.com

作者贡献声明:黄小华设计本研究的思路方案,对稿件关键内容进行了修改和指正,获得了南充市市校科技战略合作基金项目支持;杜青林起草和撰写稿件,对文章数据进行收集,分析和处理;刘念、陈雨薇、钟舒婷、刘琢玉协助完成随访以及数据处理,对稿件进行修订和评议;全体作者都同意发表最后的修改稿,同意对本研究的所有方面负责,确保本研究的准确性和诚信。


基金项目: 南充市市校科技战略合作 20SXQT0303
收稿日期:2022-12-28
接受日期:2023-06-25
中图分类号:R445.2  R576  R587.1 
文献标识码:A
DOI: 10.12015/issn.1674-8034.2023.07.012
本文引用格式:杜青林, 黄小华, 刘念, 等. 基于磁共振T2WI影像组学预测急性胰腺炎后糖尿病的价值[J]. 磁共振成像, 2023, 14(7): 67-72. DOI:10.12015/issn.1674-8034.2023.07.012.

0 前言

       急性胰腺炎后糖尿病(post-acute pancreatitis diabetes mellitus, PPDM-A)是指既往无糖尿病史,在急性胰腺炎(acute pancreatitis, AP)发病三个月后出现的血糖异常增高,并达到糖尿病诊断标准,是AP治疗后备受关注的远期并发症之一,是胰源性糖尿病的重要组成部分[1, 2]。PPDM-A的发病率随时间延长而增加,第一年发病率约为7.2%,在第二年时约为11.2%[3]。研究表明与2型糖尿病相比,PPDM-A更易出现血糖控制不佳,具有更高的死亡风险,还被认为是胰腺癌发生的前兆[4, 5, 6]。同时,研究表明PPDM-A是AP后个体精神障碍的危险因素[7]。目前,PPDM-A发病的主要机制及相关危险因素的临床研究已取得进展[8, 9, 10],但尚缺乏稳定的预测模型实现PPDM-A的早期预测。影像组学可通过影像图像获取大量的潜在特征数据,实现相关预测模型的建立[11, 12],MRI相较于CT可得到更精准的结构、功能信息,且无电离辐射。因此,本研究基于MRI T2WI序列,利用影像组学构建PPDM-A预测模型,并联合相关临床特征实现PPDM-A的早期预测,以期帮助临床实现早期干预,降低PPDM-A的发生风险。

1 材料与方法

1.1 研究对象

       本研究遵守《赫尔辛基宣言》,经川北医学院附属医院伦理委员会批准,免除受试者知情同意书,批准文号:2022ER043-1。

       回顾性收集2016年1月至2019年12月在本院确诊的AP患者病例。纳入标准:(1)符合AP诊断标准[13];(2)入院后7 d内接受MRI上腹部检查。排除标准:(1)既往有糖尿病病史;(2)AP发病3个月内出现的应激性血糖升高;(3)慢性胰腺炎急性发作;(4)合并胰腺良、恶性肿瘤;(5)临床数据缺失、失访患者。所有入组患者通过病历查询、电话随访的方式,确定患者是否出现糖尿病临床症状并通过最近一次空腹血糖、糖化血红蛋白或随机血糖的结果来诊断是否为PPDM-A,随访时间2022年9月截止。最终共纳入140例病例,57例归入PPDM-A组,83例为AP后血糖正常组。将入组患者按7∶3的比例随机分为训练组98例(PPDM-A组40例,AP后血糖正常组58例)和验证组42例(PPDM-A组17例,AP后血糖正常组25例)。

1.2 检查方法

       采用美国GE公司Discovery 750 3.0 T MRI扫描仪和32通道体部相控阵列线圈行上腹部扫描。扫描序列为横断面T2加权单次激发快速自旋回波序列,扫描参数:FOV 34 cm×34 cm,矩阵320×256,层厚6 mm,间距1 mm,TR 6000 ms,TE 120 ms。

1.3 图像分割、感兴趣区的勾画及相关特征提取

       使用开源软件3D Slicer(V5.0.3,https://www.slicer.org)对感兴趣区(region of interest, ROI)进行分割及特征提取(图1)。一名具有6年工作经验的放射科医师,在T2WI序列图像上沿胰腺边缘逐层勾画ROI,注意区分血管、肠管及周围脂肪组织。利用3D Slicer软件中的“PyRadiomics”从ROI中提取常见的7个特征组:一阶特征(first order)、形状特征(shape)、灰度共生矩阵(gray level co-occurrence matrix, GLCM)、灰度游程长度矩阵(gray level run-length matrix, GLRLM)、灰度区域大小矩阵(gray level size zone matrix, GLSZM)、邻域灰度差矩阵(neighboring gray tone difference matrix, NGTDM)、灰度依赖矩阵(gray level dependence matrix, GLDM)。

图1  T2WI序列的胰腺实质感兴趣区(ROI)勾画示意图。男,49岁,急性胰腺炎,勾画时应沿胰腺边缘勾画,尽量避开血管和胆总管。1A:胰腺最大层面的横断位图像;1B:基于3D slicer软件的胰腺实质ROI勾画示意图。
Fig. 1  Schematic illustration depicting the contour of pancreatic parenchyma on T2WI sequence. Male, 49 years old, acute pancreatitis. During the delineation, it should be done along the edge of the pancreas, avoiding the blood vessels and common bile duct as much as possible. 1A: The cross-sectional images of the pancreas at its maximum dimension; 1B: Schematic diagram of ROI of pancreatic parenchyma based on 3D slicer software.

1.4 影像组学特征的选择及模型的构建和评估

       为确保所选特征具备良好的一致性和可重复性,由另一名5年工作经验的放射科医师随机抽取大约1/3的图像独立勾画ROI并提取相应特征,通过组间相关系数(inter-class correlation coefficient, ICC)进行观察者间的影像组学特征的一致性评价,ICC>0.75的特征表示拥有良好的一致性。针对ICC>0.75的影像组学特征进行Z-score标准化的处理,再连续采用方差阈值法、K最佳选择法(select K best)、最小绝对收缩和选择算子(least absolute shrinkage and selection operator, LASSO)算法完成特征筛选,其中LASSO算法采用十折交叉验证以确保影像组学特征的稳定性。最后利用两组间差异具有统计学意义的临床特征及降维后的影像组学特征采用随机森林的方法分别构建临床模型、影像组学模型以及二者的联合模型。通过展示各个模型的特异度、敏感度、准确度及曲线下面积(area under the curve, AUC)等参数直观体现模型的预测价值,并对效能最佳的模型使用校正曲线、Hosmer-Lemeshow检验验证模型的拟合情况,再利用DeLong检验对各模型之间的预测效能进行评估,最后采用决策曲线分析(decision curve analysis, DCA)评估各个模型的临床净收益。影像组学特征分析、上述三个模型的建立均在uAI Research Portal(V1.6,上海联影智能医疗科技有限公司,中国)完成。

1.5 统计学方法

       采用SPSS 26.0进行统计学分析。采用Kolmogorov-Smirnov法对计量资料进行正态性检验。符合正态分布的计量资料采用独立样本t检验进行两组间比较,以(x¯±s)表示,不符合正态分布的计量资料采用Mann-Whitney U检验,并以MQR)表示。计数资料以例数或构成比表示,两组间比较采用χ2检验。DeLong检验、Hosmer-Lemeshow检验及受试者工作特征(receiver operating characteristic, ROC)曲线、校正曲线和决策曲线的绘制使用R(4.2.1)和Python(3.9.13)软件完成。P<0.05为差异具有统计学意义。

2 结果

2.1 临床资料

       本研究纳入患者140例,PPDM-A组患者57例,年龄(46±12)岁,其中男36例、女21例;AP后血糖正常患者83例,年龄(47±12)岁,其中男55例、女28例。PPDM-A组和AP后血糖正常组在性别、年龄、住院时间、脂肪肝、胆道结石、胰腺坏死、脂肪酶、血清淀粉酶、血清胰淀粉酶和严重程度中差异均无统计学意义(P>0.05);在AP多次发作、高脂血症、局部并发症、入院血糖及甘油三酯中两组差异存在统计学意义(P<0.05),详见表1

表1  PPDM-A组和AP后血糖正常组的临床特征比较
Tab 1  Comparison of clinical characteristics between PPDM-A group and normoglycemia group after AP

2.2 影像组学特征选择的结果

       本研究对所得特征进行一致性评价,将ICC>0.75的1171个稳定特征保留(图2);利用方差阈值法,去除方差低于0.8的特征;采用select K best去除在两组间差异无统计学意义(P>0.05)的特征;最后使用LASSO算法选择-Log(Alpha)为2.02对应的特征,经过上述三步降维处理最终得到6个特征及相应系数(图3A3B表2)。

图2  两名观察者组间相关系数(ICC)示意图。ICC>0.75(红线以上)代表一致性较好的特征,经筛选后保留1171个稳定特征。
图3  最小绝对收缩和选择算子(LASSO)算法降维示意图。3A:LASSO特征选择图;3B:LASSO系数变化曲线图。纵轴为系数,横轴表示-Log(Alpha)值。当-Log(Alpha)=2.02,此时有6个系数不等于0的影像组学特征。
Fig. 2  The inter-class correlation coefficients of inter-observer. ICC>0.75 (above the red line) represented the features with good consistency, and 1171 stable features were retained after screening.
Fig. 3  The image of dimension reduction with least absolute shrinkage and selection operator (LASSO). 3A: LASSO feature selection plot; 3B: Curve of LASSO coefficient variation. The vertical axis is the coefficient, and the horizontal axis represents the number of the -Log (Alpha). When -Log(Alpha)=2.02, there were 6 radiomics features with coefficient not equal to 0.
表2  影像组学特征系数表
Tab. 2  Radiomics feature coefficient table

2.3 模型效能的评估

       临床模型训练组AUC为0.773 [95%置信区间(confidence interval, CI):0.679~0.866],验证组AUC为0.664(95% CI:0.488~0.839);影像组学模型训练组AUC为0.831(95% CI:0.750~0.912),验证组AUC为0.821(95% CI:0.684~0.958);联合模型训练组AUC为0.881(95% CI:0.816~0.946),验证组AUC为0.899(95% CI:0.806~0.992),详见表3图4

       DeLong检验结果显示,联合模型与临床模型的预测效能在训练组和验证组中差异均存在统计学意义(训练组:P=0.024、验证组:P=0.013);联合模型与影像组学模型在训练组中差异存在统计学意义(P=0.047),在验证组中差异无统计学意义(P=0.090);影像组学模型与联合模型在训练组、验证组中差异均无统计学意义(训练组:P=0.362、验证组:P=0.191)。

       绘制联合模型的校正曲线(图5),Hosmer-Lemeshow检验(P=0.250)验证了模型预测值与实际观察值之间存在一致性,具有较好的拟合情况。DCA结果显示,影像组学模型、联合模型预测PPDM-A的临床价值均优于临床模型,当阈值概率大于12%时,联合模型的临床净收益优于影像组学模型(图6)。

图4  三种模型的受试者工作特征(ROC)曲线及曲线下面积(AUC)值。4A:训练组;4B:验证组。
Fig. 4  The receiver operating characteristic (ROC) curve and area under the curve (AUCs) of the three models. 4A: Training group; 4B: Validation group.
图5  联合模型的校正曲线图。
图6  决策曲线图。
Fig. 5  Calibration plot for the combined model.
Fig. 6  Decision curve diagram.
表3  训练组和验证组中各模型预测PPDM-A的效能
Tab. 3  Performance of each model in predicting PPDM-A in the training and validation groups

3 讨论

       本研究通过电话随访等方式回顾性收集PPDM-A患者57例,AP后血糖正常患者83例,并基于临床特征、影像组学特征建立相应模型,寻找可实现早期预测PPDM-A的有效方法。研究发现基于临床特征及T2WI影像组学特征构建的联合模型具有较好的预测效能和临床价值,提示临床特征与影像组学特征的联合应用可实现对PPDM-A的早期预测,为临床医师的早期干预提供参考。

3.1 临床研究分析

       PPDM-A是胰源性糖尿病的重要组成部分,相较于其他病因而言,AP的发生最为常见,应得到广泛的关注[14, 15]。目前,关于PPDM-A的研究大多局限于临床研究,研究表明胰腺β细胞损伤、胰腺内脂肪沉积、铁代谢等与PPDM-A的发生发展密切相关[8,16, 17]。在多项研究中发现肥胖、胰腺坏死程度、AP的复发均为PPDM-A的危险因素[18, 19]。研究还发现体质量指数、空腹血糖和白介素-1β可作为PPDM-A的预测因子[20, 21]。故本研究参考上述研究纳入相应的临床特征构建临床模型,研究结果显示高脂血症、AP复发、局部并发症、入院血糖及甘油三酯含量在PPDM-A组和AP后血糖正常组中差异均存在统计学意义。高脂血症及甘油三酯的差异提示血脂的异常可能是PPDM-A的危险因素[22];局部并发症提示胰腺周围的水肿或坏死在PPDM-A中也值得关注;本研究发现PPDM-A组的血糖含量高于AP后血糖正常组,与既往研究一致[9,23],提示AP发作期间血糖的有效控制可能会降低PPDM-A的发生;本研究中AP反复发作在两组间具有差异,提示其可能是PPDM-A的诱因,既往研究也得到了类似结果[18]。AP严重程度是否影响PPDM-A的发生发展并未达成共识[24, 25],值得进一步深入研究。本研究中构建的临床模型预测效能一般,有待纳入更多相关临床指标进行进一步研究。

3.2 影像组学模型及联合模型研究分析

       PPDM-A与影像组学的相关研究鲜有报道,影像组学是一种可高通量地获取影像图像潜在特征信息的定量分析方法,已在预测AP复发、AP进展等方面取得进展,证明了该方法在胰腺疾病的可行性[26, 27, 28]。能否通过胰腺的影像组学特征来实现PPDM-A的早期预测值得研究。因此,本研究基于MRI T2WI序列及相关临床指标探究影像组学在早期预测PPDM-A的价值。本研究得到的影像组学特征为一阶特征中的偏度(Skewness)和峰度(Kurtosis),GLSZM中的Large Area Emphasis(LAE),GLCM中的Informational Measure of Correlation1(IMC1)及GLDM中的Large Dependence High Gray Level Emphasis(LDHGLE)。其中,LAE可度量长游程长度(long run-lengths),数值越大表示结构越粗糙;IMC1可量化纹理的复杂性;LDHGLE可衡量图像中具有较高灰度值的相关关系;偏度可衡量数值分布在平均值上的不对称性;峰度表示图像ROI中值分布的“峰值”,较高的峰度意味着分布的主要集中在尾部而不是平均值[29, 30]。提示PPDM-A的发展与胰腺结构的粗糙度,纹理的复杂性以及图像的灰度值等因素可能存在一定关系。本研究建立的影像组学模型效能良好,其预测PPDM-A发生的净收益优于临床模型,但模型敏感度欠佳。本研究将临床特征与影像组学特征共同构建联合模型,结果发现联合模型的预测效能及敏感度、准确度均有所提升,提示多元特征的联合应用有助于提高模型的稳定性,进一步优化模型。综上,本研究中的联合模型预测效能及临床净收益均优于临床模型与影像组学模型,可作为早期预测PPDM-A的有效手段。

3.3 本研究的局限性及展望

       本研究仍存在一些不足:(1)数据样本量较小,在患者选择上可能存在一定偏倚;(2)特征均来源于单中心、单一序列,模型敏感度欠佳;(3)临床特征纳入不全,临床模型预测效能欠佳。因此,期望今后能纳入更大样本的多序列数据,进一步提高数据的多样性及准确性,再基于多序列的MRI联合临床模型,进一步探讨临床模型、影像组学模型及联合模型的预测效能,为预测PPDM-A的发生发展寻求更为稳定可靠的预测方法。

4 结论

       综上所述,基于临床特征及T2WI影像组学特征构建的联合模型在早期预测PPDM-A的效能良好,具有一定价值,有助于临床医生在AP治疗后对患者进行风险评估,并通过有效干预进而降低PPDM-A的发生风险。

[1]
PETROV M S, YADAV D. Global epidemiology and holistic prevention of pancreatitis[J]. Nat Rev Gastroenterol Hepatol, 2019, 16(3): 175-184. DOI: 10.1038/s41575-018-0087-5.
[2]
钱苏皖, 童智慧. 重视急性胰腺炎后糖尿病的发生与发展[J]. 中华胰腺病杂志, 2022, 22(2): 87-92. DOI: 10.3760/cma.j.cn115667-20220126-00015.
QIAN S W, TONG Z H. Focus on the development and progression of post-acute pancreatitis diabetes mellitus[J]. Chin J Pancreatol, 2022, 22(2): 87-92. DOI: 10.3760/cma.j.cn115667-20220126-00015.
[3]
BHARMAL S H, CHO J, ALARCON RAMOS G C, et al. Trajectories of glycaemia following acute pancreatitis: a prospective longitudinal cohort study with 24 months follow-up[J].J Gastroenterol, 2020, 55(8): 775-788. DOI: 10.1007/s00535-020-01682-y.
[4]
WOODMANSEY C, MCGOVERN A P, MCCULLOUGH K A, et al. Incidence, demographics, and clinical characteristics of diabetes of the exocrine pancreas (type 3c): a retrospective cohort study[J]. Diabetes Care, 2017, 40(11): 1486-1493. DOI: 10.2337/dc17-0542.
[5]
CHO J, SCRAGG R, PETROV M S. Risk of mortality and hospitalization after post-pancreatitis diabetes mellitus vs type 2 diabetes mellitus: a population-based matched cohort study[J]. Am J Gastroenterol, 2019, 114(5): 804-812. DOI: 10.14309/ajg.0000000000000225.
[6]
PETROV M S. Post-pancreatitis diabetes mellitus and excess intra-pancreatic fat deposition as harbingers of pancreatic cancer[J]. World J Gastroenterol, 2021, 27(17): 1936-1942. DOI: 10.3748/wjg.v27.i17.1936.
[7]
CHO J, WALIA M, SCRAGG R, et al. Frequency and risk factors for mental disorders following pancreatitis: a nationwide cohort study[J]. Curr Med Res Opin, 2019, 35(7): 1157-1164. DOI: 10.1080/03007995.2018.1560748.
[8]
PETROV M S. Post-pancreatitis diabetes mellitus: investigational drugs in preclinical and clinical development and therapeutic implications[J]. Expert Opin Investig Drugs, 2021, 30(7): 737-747. DOI: 10.1080/13543784.2021.1931118.
[9]
刘慧, 李贤秋, 罗刚, 等. 初发急性胰腺炎后新发糖尿病的危险因素评估[J]. 临床肝胆病杂志, 2020, 36(12): 2771-2776. DOI: 10.3969/j.issn.1001-5256.2020.12.026.
LIU H, LI X Q, LUO G, et al. Risk factors for new-onset diabetes after incipient acute pancreatitis[J]. J Clin Hepatol, 2020, 36(12): 2771-2776. DOI: 10.3969/j.issn.1001-5256.2020.12.026.
[10]
CHO J, SCRAGG R, PANDOL S J, et al. Exocrine pancreatic dysfunction increases the risk of new-onset diabetes mellitus: results of a nationwide cohort study[J]. Clin Transl Sci, 2021, 14(1): 170-178. DOI: 10.1111/cts.12837.
[11]
MAYERHOEFER M E, MATERKA A, LANGS G, et al. Introduction to radiomics[J]. J Nucl Med, 2020, 61(4): 488-495. DOI: 10.2967/jnumed.118.222893.
[12]
LAFATA K J, WANG Y Q, KONKEL B, et al. Radiomics: a primer on high-throughput image phenotyping[J].Abdom Radiol, 2022, 47(9): 2986-3002. DOI: 10.1007/s00261-021-03254-x.
[13]
中华医学会外科学分会胰腺外科学组. 中国急性胰腺炎诊治指南(2021)[J]. 中华外科杂志, 2021, 59(7): 578-587. DOI: 10.3760/cma.j.cn112139-20210416-00172.
Pancreatic Surgery Group of Surgery Branch of Chinese Medical Association. Guidelines for diagnosis and treatment of acute pancreatitis in China (2021)[J]. Chin J Surg, 2021, 59(7): 578-587. DOI: 10.3760/cma.j.cn112139-20210416-00172.
[14]
余洁, 张化冰, 李玉秀. 胰源性糖尿病[J]. 中华糖尿病杂志, 2021, 13(1): 116-118. DOI: 10.3760/cma.j.cn115791-20201105-00656.
YU J, ZHANG H B, LI Y X. Pancreatogenic diabetes mellitus[J]. Chin J Diabetes Mellit, 2021, 13(1): 116-118. DOI: 10.3760/cma.j.cn115791-20201105-00656.
[15]
PETROV M S, BASINA M. DIAGNOSIS OF ENDOCRINE DISEASE: Diagnosing and classifying diabetes in diseases of the exocrine pancreas[J]. Eur J Endocrinol, 2021, 184(4): R151-R163. DOI: 10.1530/eje-20-0974.
[16]
KO J, SKUDDER-HILL L, CHO J, et al. The relationship between abdominal fat phenotypes and insulin resistance in non-obese individuals after acute pancreatitis[J/OL]. Nutrients, 2020, 12(9): 2883 [2022-12-27]. https://www.mdpi.com/2072-6643/12/9/2883. DOI: 10.3390/nu12092883.
[17]
CHAND S K, SINGH R G, PENDHARKAR S A, et al. Iron: a strong element in the pathogenesis of chronic hyperglycaemia after acute pancreatitis[J]. Biol Trace Elem Res, 2018, 183(1): 71-79. DOI: 10.1007/s12011-017-1131-y.
[18]
FIRKINS S A, HART P A, PAPACHRISTOU G I, et al. Identification of a risk profile for new-onset diabetes after acute pancreatitis[J]. Pancreas, 2021, 50(5): 696-703. DOI: 10.1097/MPA.0000000000001818.
[19]
MAN T, SEICEAN R, LUCACIU L, et al. Risk factors for new-onset diabetes mellitus following acute pancreatitis: a prospective study[J]. Eur Rev Med Pharmacol Sci, 2022, 26(16): 5745-5754. DOI: 10.26355/eurrev_202208_29511.
[20]
GUO S Y, YANG H Y, NING X Y, et al. Combination of body mass index and fasting blood glucose improved predictive value of new-onset prediabetes or diabetes after acute pancreatitis[J]. Pancreas, 2022, 51(4): 388-393. DOI: 10.1097/mpa.0000000000002025.
[21]
BHARMAL S H, KIMITA W, KO J, et al. Cytokine signature for predicting new-onset prediabetes after acute pancreatitis: a prospective longitudinal cohort study[J/OL]. Cytokine, 2022, 150: 155768 [2022-12-27]. https://www.sciencedirect.com/science/article/abs/pii/S1043466621003574?via%3Dihub. DOI: 10.1016/j.cyto.2021.155768.
[22]
LV Y Q, ZHANG J, YANG T, et al. Non-alcoholic fatty liver disease (NAFLD) is an independent risk factor for developing new-onset diabetes after acute pancreatitis: a multicenter retrospective cohort study in Chinese population[J/OL]. Front Endocrinol (Lausanne), 2022, 13: 903731 [2022-12-27]. https://www.frontiersin.org/articles/10.3389/fendo.2022.903731/full. DOI: 10.3389/fendo.2022.903731.
[23]
MA J H, YUAN Y J, LIN S H, et al. Nomogram for predicting diabetes mellitus after the first attack of acute pancreatitis[J]. Eur J Gastroenterol Hepatol, 2019, 31(3): 323-328. DOI: 10.1097/meg.0000000000001307.
[24]
HART P A, BRADLEY D, CONWELL D L, et al. Diabetes following acute pancreatitis[J]. Lancet Gastroenterol Hepatol, 2021, 6(8): 668-675. DOI: 10.1016/S2468-1253(21)00019-4.
[25]
RICHARDSON A, PARK W G. Acute pancreatitis and diabetes mellitus: a review[J]. Korean J Intern Med, 2021, 36(1): 15-24. DOI: 10.3904/kjim.2020.505.
[26]
胡云涛, 黄小华, 刘念, 等. 基于磁共振T2WI序列影像组学预测急性胰腺炎复发的价值[J]. 磁共振成像, 2021, 12(10): 12-15. DOI: 10.12015/issn.1674-8034.2021.10.003.
HU Y T, HUANG X H, LIU N, et al. The value of T2WI sequence-based radiomics in predicting recurrence of acute pancreatitis[J]. Chin J Magn Reson Imaging, 2021, 12(10): 12-15. DOI: 10.12015/issn.1674-8034.2021.10.003.
[27]
范海云, 陈基明, 陈亮亮, 等. CT影像组学预测早期急性胰腺炎进展的价值[J]. 中华放射学杂志, 2022, 56(7): 778-784. DOI: 10.3760/cma.j.cn112149-20210829-00615.
FAN H Y, CHEN J M, CHEN L L, et al. The value of CT based radiomics in predicting progression of early acute pancreatitis[J]. Chin J Radiol, 2022, 56(7): 778-784. DOI: 10.3760/cma.j.cn112149-20210829-00615.
[28]
TANG L L, MA L, CHEN Y Y, et al. Radiomics analysis of contrast-enhanced T1W MRI: predicting the recurrence of acute pancreatitis[J/OL]. Sci Rep, 2023, 13: 2762 [2022-12-27]. https://www.nature.com/articles/s41598-022-13650-y. DOI: 10.1038/s41598-022-13650-y.
[29]
ZWANENBURG A, VALLIÈRES M, ABDALAH M A, et al. The image biomarker standardization initiative: standardized quantitative radiomics for high-throughput image-based phenotyping[J]. Radiology, 2020, 295(2): 328-338. DOI: 10.1148/radiol.2020191145.
[30]
ABBASIAN ARDAKANI A, BUREAU N J, CIACCIO E J, et al. Interpretation of radiomics features-a pictorial review[J/OL]. Comput Methods Programs Biomed, 2022, 215: 106609 [2022-12-27]. https://www.sciencedirect.com/science/article/abs/pii/S0169260721006830?via%3Dihub. DOI: 10.1016/j.cmpb.2021.106609.

上一篇 BOLD-fMRI对2型糖尿病患者早期肾脏功能变化的研究
下一篇 DWI联合MRI平扫对中晚期卵巢癌PDS手术方案制订的应用价值研究
  
诚聘英才 | 广告合作 | 免责声明 | 版权声明
联系电话:010-67113815
京ICP备19028836号-2