影像组学是通过高通量计算,从医学影像中挖掘大量(200+)的定量影像特征,使用统计学/机器学习方法,筛选出最有价值的影像组学特征,用于临床决策支持系统,以提高诊断、预后和分子学水平预测的准确性,在肿瘤研究中越来越重要.磁共振成像(MRI)由于其空间高分辨率的优势,影像组学提取MRI图像特征,在中枢神经系统肿瘤得到了更广泛的研究.本文就MRI影像组学在脑胶质瘤临床评估中的研究应用,予以综述.

目的 分析终末期肾病（end stage renal diseases, ESRD）患者心血管不良事件（cardiovascular adverse events, CAE）的危险因素，构建风险预测模型。方法 回顾性分析武汉大学人民医院ESRD住院患者血液生化指标及超声心动图参数，按是否发生CAE进行分组。采用R语言分析CAE危险因素，建立风险预测模型并绘制列线图。使用校准曲线、C-index、R~2进行验证。结果 研究共纳入患者414例。其中CAE组233例、非CAE组181例，多因素logistic回归分析结果显示，年龄、糖尿病、血清磷酸盐、血红蛋白、左房前后径（left atrial diameter, LAD）、二尖瓣舒张早期流速与室间隔处二尖瓣环舒张早期运动速度比值（E/e’）同患者CAE风险独立相关，列线图模型C-index=0.868,R~2=0.517,校准曲线一致性良好。结论 年龄、糖尿病、血清磷酸盐、低血红蛋白、LAD及E/e’同CAE风险独立相关，列线图模型能够较好地预测患者CAE发生风险。

Objective:To investigate the application value of echocardiography (UCG) in risk assessment of cardiovascular adverse events (CAE) in patients with end stage renal diseases (ESRD) .Methods:Patients with ESRD hospitalized in Renmin Hospital of Wuhan University from April 2020 to June 2021 were selected. CAE was considered as the adverse outcome. They were followed up until July 2022 and divided into event group and non-event group according to the occurrence of outcome. The clinical data, blood biochemical parameters and echocardiographic parameters were analyzed by cohort study. Data analysis and drawing were performed in R language (version 4.1.3) . The variables of combined and uncombined echocardiography were screened by Lasso regression, and multivariate Cox regression modeling was included. Random forest was used to analyze the importance of variables, internal resampling method was used and calibration curves were drawn for model validation, and C-index and net reclassification index (NRI) were used to evaluate and compare model efficacy. Kaplan-Meier method was used to draw survival curves and compare the median survival times of different dialysis patients with peak tricuspid regurgitation velocity (TRV) ≥2.5 m/s as assessed by ultrasound.Results:Age, CAE history, serum phosphorus, hemoglobin, left atrial anteroposterior diameter (LAD) , mitral E peak velocity and ratio of mitral to myocardial early velocities (E/e’) , TRV were independently associated with adverse outcomes. Random forest showed that the importance of variables in order of importance were: E/e’, LAD, CAE history, serum phosphorus, hemoglobin, age, TRV. The C-index of unbound and bound ultrasound parameter models were 0.719 (95% CI: 0.690-0.748) and 0.753 (95% CI: 0.726-0.780) , respectively. The calibration curves have good consistency. NRI=0.168 for 1-year events and 0.277 for 2-year events; Kaplan-Meier survival curves showed that the median survival time of TRV≥2.5 m/s group (about 279 days) was lower than that of TRV< 2.5 m/s group (about 619 days) . In the TRV≥2.5 m/s group, the median survival time of peritoneal dialysis (about 316 days) was longer than that of hemodialysis (about 244 days) . Conclusion:E/e’, LAD and TRV are independent predictor of CAE. E/e’ and LAD are important UCG parameters for prognosis of ESRD. TRV can provide valuable information for the selection of dialysis mode. UCG has important clinical application value for risk assessment of CAE in ESRD.

目的 观察对比增强液体衰减反转恢复(CE-FLAIR)序列MRI影像组学模型判断成人弥漫性低级别胶质瘤(DLGG)1p/19q状态的价值.方法 纳入135例成人DLGG患者、含81例1p/19q共缺失,经分层抽样按7:3比例将其分为训练集(n=95)及验证集(n=40).基于训练集CE-FLAIR数据提取、筛选DLGG 1p/19q共缺失影像组学特征,构建支持向量机(SVM)、随机森林(RF)、极限梯度提升(XGBoost)、轻量梯度提升机(LightGBM)及逻辑回归(LR)模型;绘制受试者工作特征曲线,计算曲线下面积(AUC),评价影像组学模型判断训练集及验证集DLGG 1p/19q状态的价值,并以DeLong检验进行比较.结果 共提取851个影像组学特征,以Mann-Whitney U检验筛选出74个差异有统计学意义者,再经5折交叉验证的最小绝对收缩和选择算子选出12个与1p/19q状态显著相关者;以之构建的SVM、RF、XGBoost、LightGBM及LR模型评价训练集DLGG 1p/19q状态的AUC分别为0.89、0.97、0.97、0.96及0.85,评估验证集的AUC分别为0.86、0.92、0.93、0.92及0.78.验证集中,LR模型的AUC低于SVM、RF、XGBoost及LightGBM(Z=2.981、3.136、3.014、2.827,P均<0.05),而后四者间AUC差异均无统计学意义(P均>0.05);RF模型准确率最高,为88.24％.结论 基于CE-FLAIR影像组学模型可有效评估成人DLGG1p/19q状态;SVM、RF、XGBoost及LightGBM模型效能均较高,以RF模型准确率最高.

目的:评价在较低级别胶质瘤MRI影像组学研究中手动分割ROI特征提取的一致性和可重复性.方法:回顾性分析经病理证实的78例较低级别胶质瘤患者,由2名医师使用3D-Slicer软件对肿瘤ROI进行手动勾画分割,用Radiomics插件提取影像组学特征参数,包括一阶特征、形态特征和高阶纹理特征等.使用组内相关系数(ICC)、Bland-Altman图及Wilcoxon秩和检验行一致性分析和可重复性评价.结果:2名医师ROI勾画的ICC均>0.75,平均为0.937±0.068,其特征中能量和对比的一致性表现最佳(均ICC=0.998).同一医师ROI勾画的ICC均>0.9,平均为0.978±0.022,其特征中平均值和对比的一致性表现最佳(均ICC=1.000).Bland-Altman图可直观展示散点分布和一致性,获取重复性系数.Wilcoxon秩和检验均P>0.05,说明同一观察者对肿瘤ROI的手动分割具有可重复性.结论:基于手动分割ROI的较低级别胶质瘤的MRI影像组学特征参数提取具备较高的一致性及可重复性.

目的 基于磁共振T2WI构建影像组学模型,预测低级别胶质瘤1p/19q缺失状态的价值.方法 回顾性分析本院经病理证实的154例低级别胶质瘤患者(1p/19q共缺失100例,1p/19q非共缺失54例),按照分层抽样7:3分成训练集和验证集.使用3D-Slicer软件对肿瘤区域进行手动分割,用pyradiomics进行特征提取.临床资料的分析采用t检验/χ2检验;影像组学特征采用方差法和10折交叉验证的LASSO算法进行筛选,最后建立支持向量机、高斯朴素贝叶斯、K-近邻、逻辑回归模型,采用ROC曲线的曲线下面积值和sklearn分类报告中的参考指标(准确度、敏感度、特异性、F1分数)进行效能评价.结果 4种模型中,支持向量机的曲线下面积值最高,训练集和验证集分别为0.95、0.91;参考指标中表现最佳为K-近邻,其准确度、敏感度、特异性及F1分数分别为0.87、0.97、0.70、0.91;其次为支持向量机,各项指标与模型平均值相当.结论 基于T2WI影像组学模型可以有效地预测低级别胶质瘤1p/19q的缺失状态.

目的 探讨小细胞肺癌(smal cel lung cancer,SCLC)CT影像表现及临床表现特征.方法 采用64排螺旋CT对患者进行扫描成像,回顾性分析39例经病理证实的小细胞肺癌的临床及CT资料.结果小细胞肺癌的CT基本表现为肺门及(或)肺内肿块或结节,常伴有纵隔、肺门及肺内淋巴结肿大,肿大淋巴结融合常呈"冰冻纵隔""冰冻肺门",但肺不张及肿块内坏死、钙化少见.小细胞肺癌以中央型为主,周围型相对比较少见.结论 小细胞肺癌的 CT 影像表现形式多样,但具有一定的特征性,CT 平扫及多期增强扫描有助于诊断,回顾性分析 39 例经病理证实的小细胞肺癌的临床及 CT 资料,结合临床表现及实验室检查,可以提高对本病的认识和 CT 诊断,对临床的诊疗和预后评估有着积极的意义.

Objective To investigate the CT and MRI manifestations of primitive neuroectodermal tumors (PNET) in or‐der to improve understanding of PNET .Methods A retrospective analysis of imaging data of 45 cases of PNET confirmed pathologically ,which included 29 males ,19 females ,with a mean age of 36 .4 years old ,was conducted in this work .Re‐sults PNET can be divided into central (cPNET ) and peripheral (pPNET ) types according to the occurrence at different parts .9 cases of cPNET exhibited the larger supratentorial tumor ,cystic degeneration and necrosis ,none or mild edema a‐round the tumor with clear boundary ,and relative uniform density /signal .CT showed that the density was equal or slightly higher than that of brain gray matter ,MRI T1 WI showed equal or slightly lower signal ,and T2 WI showed a slightly higher or mixed signal ,which may be associated with CSF spread .As for the 36 cases of pPNET ,the clinical manifestations were the obvious soft tissue mass ,with unclear boundary ,the density /signal was not uniform ,a little hemorrhage was visible ,and the calcification was rare .CT showed that the density was equal or slightly lower than that of the surrounding muscle and the MRI signal was complex ,without significant specificity .Tumor involved in the skeletal system can appeare as a large sheet of osteolytic bone destruction ,without periosteal reaction .Conclusion Although the imaging diagnostics of PNET is not taken as a gold standard ,it can determine the extent of lesions ,which is of importance for the preoperative staging ,postoperative evaluation ,and differential diagnosis .