目的探讨动态对比增强MRI(dynamic contrast-enhancement MRI,DCE-MRI)瘤周血管特征结合瘤内血流动力学参数在乳腺影像报告和数据系统(breast imaging reporting and data system,BI-RADS)4类肿瘤中的鉴别诊断价值。材料与方法回顾性分...目的探讨动态对比增强MRI(dynamic contrast-enhancement MRI,DCE-MRI)瘤周血管特征结合瘤内血流动力学参数在乳腺影像报告和数据系统(breast imaging reporting and data system,BI-RADS)4类肿瘤中的鉴别诊断价值。材料与方法回顾性分析2018年8月至2023年3月于大连医科大学附属第一医院行乳腺MRI检查为BI-RADS 4类且病理结果明确肿瘤的女性病例102例,其中良性组43例,恶性组59例。记录患者年龄、病灶最大径(dmax)、乳腺DCE-MRI基本影像学特征、瘤周血管特征及瘤内血流动力学参数值。通过单因素和多因素logistic回归分析比较两组间多参数的差异,利用受试者工作特征(receiver operating characteristic,ROC)曲线以及曲线下面积(area under the curve,AUC)分析瘤周血管特征指标与瘤内参数值联合应用对BI-RADS 4类乳腺良恶性两组肿瘤鉴别的诊断效能。应用DeLong检验对AUC进行比较。结果乳腺良性组和恶性组病例在年龄、dmax、背景实质强化(background parenchymal enhancement,BPE)、纤维腺体组织量(fibroglandular tissue,FGT)、瘤周相邻血管征(adjacent vascular sign,AVS)数目、瘤周血管最大径、患侧瘤周与健侧同一象限血管直径差值(△d)、瘤周血管出现期相以及瘤内容积转移常数(volume transfer constant,K^(trans))、速率常数(flux rate constant,K_(ep))、最大增强斜率(maximum slope of increase,MSI)和时间-信号强度曲线(time-signal intensity curve,TIC)类型的差异均具有统计学意义(P<0.05),而病变位置、信号增强率(signal enhancement ratio,SER)和血管外细胞外间隙容积比(volume fraction of extravascular extra vascular space,V_(e))差异无统计学意义(P>0.05)。通过多因素logistic回归分析结果显示,△d、dmax、MSI和K^(trans)为区分两组间的独立影响因素,其中优势比最大的是MSI值(AUC为0.923)。将瘤周血管特征△d分别与dmax、MSI和K^(trans)进行两者联合模型比较,以△d与MSI联合模型的诊断效能最高(AUC为0.933,敏感度和特异度分别为93.2%和83.7%),且△d联合MSI与△d联合K^(trans)比较的差异具有统计学意义(P=0.001);其他联合指标在两两比较时差异无统计学意义(P>0.05),联合模型高于单独MSI模型的诊断效能。结论瘤周血管特征指标(△d)联合瘤内半定量(MSI)血流动力学参数对评价BI-RADS 4类乳腺肿瘤具有较好的鉴别诊断价值。展开更多
Background: Chronic kidney disease patients are at a greater risk for nephropathy requiring dialysis after percutaneous coronary intervention. Such patients are usually deferred due to fear of “Renalism”.?Objectives...Background: Chronic kidney disease patients are at a greater risk for nephropathy requiring dialysis after percutaneous coronary intervention. Such patients are usually deferred due to fear of “Renalism”.?Objectives This study assesses the outcome of Low dose contrast protocol during PCI in CKD patients whose e-GFR 60 ml/min/1.72 m and investigates a safety margin for contrast use in these high-risk categories.?Methods: Patients were into three groups according to CV/e-GFR ratio: Group (A) low-dose: CV/e-GFR ratio 2.0 Group (B) medium-dose: CV/e-GFR ratio > 2.0 and × bodyweight\s.creatinine). Group (C) high-dose: CV/e-GFR ratio > MACD. Results: A total of 73 patients were enrolled. Average age was 54 ± 8 years,81.4% were male and 18.6% were females and 52% were diabetic. Mean baseline e-GFR was 40 ± 8.0 ml/min/1.73m2. Contrast Volume used in group A was (58.26 ± 15.05) (n = 24), in group B (109.42 ± 17.11) (n = 26) and in group C (304.5 ± 60.30) (n = 23), respectively. The incidences of CI-AKI in the 3 groups were 0%, 11.5% and 35%, respectively (p = 0.02). All-cause death 0%, 17% and introduction of maintenance hemo dialysis was 0%, 11.5% and 26%, respectively (p Conclusion: Low dose contrast protocol is safe, effective and easily applicable technique without CI-AKI or death.展开更多
文摘具有混合记忆的自步对比学习(Self-paced Contrastive Learning,SpCL)通过集群聚类生成不同级别的伪标签来训练网络,取得了较好的识别效果,然而该方法从源域和目标域中捕获的行人数据之间存在典型的分布差异,使得训练出的网络不能准确区别目标域和源域数据域特征。针对此问题,提出了双分支动态辅助对比学习(Dynamic Auxiliary Contrastive Learning,DACL)框架。该方法首先通过动态减小源域和目标域之间的局部最大平均差异(Local Maximum Mean Discrepancy,LMMD),以有效地学习目标域的域不变特征;其次,引入广义均值(Generalized Mean,GeM)池化策略,在特征提取后再进行特征聚合,使提出的网络能够自适应地聚合图像的重要特征;最后,在3个经典行人重识别数据集上进行了仿真实验,提出的DACL与性能次之的无监督域自适应行人重识别方法相比,mAP和rank-1在Market1501数据集上分别增加了6.0个百分点和2.2个百分点,在MSMT17数据集上分别增加了2.8个百分点和3.6个百分点,在Duke数据集上分别增加了1.7个百分点和2.1个百分点。
文摘目的探讨动态对比增强MRI(dynamic contrast-enhancement MRI,DCE-MRI)瘤周血管特征结合瘤内血流动力学参数在乳腺影像报告和数据系统(breast imaging reporting and data system,BI-RADS)4类肿瘤中的鉴别诊断价值。材料与方法回顾性分析2018年8月至2023年3月于大连医科大学附属第一医院行乳腺MRI检查为BI-RADS 4类且病理结果明确肿瘤的女性病例102例,其中良性组43例,恶性组59例。记录患者年龄、病灶最大径(dmax)、乳腺DCE-MRI基本影像学特征、瘤周血管特征及瘤内血流动力学参数值。通过单因素和多因素logistic回归分析比较两组间多参数的差异,利用受试者工作特征(receiver operating characteristic,ROC)曲线以及曲线下面积(area under the curve,AUC)分析瘤周血管特征指标与瘤内参数值联合应用对BI-RADS 4类乳腺良恶性两组肿瘤鉴别的诊断效能。应用DeLong检验对AUC进行比较。结果乳腺良性组和恶性组病例在年龄、dmax、背景实质强化(background parenchymal enhancement,BPE)、纤维腺体组织量(fibroglandular tissue,FGT)、瘤周相邻血管征(adjacent vascular sign,AVS)数目、瘤周血管最大径、患侧瘤周与健侧同一象限血管直径差值(△d)、瘤周血管出现期相以及瘤内容积转移常数(volume transfer constant,K^(trans))、速率常数(flux rate constant,K_(ep))、最大增强斜率(maximum slope of increase,MSI)和时间-信号强度曲线(time-signal intensity curve,TIC)类型的差异均具有统计学意义(P<0.05),而病变位置、信号增强率(signal enhancement ratio,SER)和血管外细胞外间隙容积比(volume fraction of extravascular extra vascular space,V_(e))差异无统计学意义(P>0.05)。通过多因素logistic回归分析结果显示,△d、dmax、MSI和K^(trans)为区分两组间的独立影响因素,其中优势比最大的是MSI值(AUC为0.923)。将瘤周血管特征△d分别与dmax、MSI和K^(trans)进行两者联合模型比较,以△d与MSI联合模型的诊断效能最高(AUC为0.933,敏感度和特异度分别为93.2%和83.7%),且△d联合MSI与△d联合K^(trans)比较的差异具有统计学意义(P=0.001);其他联合指标在两两比较时差异无统计学意义(P>0.05),联合模型高于单独MSI模型的诊断效能。结论瘤周血管特征指标(△d)联合瘤内半定量(MSI)血流动力学参数对评价BI-RADS 4类乳腺肿瘤具有较好的鉴别诊断价值。
文摘Background: Chronic kidney disease patients are at a greater risk for nephropathy requiring dialysis after percutaneous coronary intervention. Such patients are usually deferred due to fear of “Renalism”.?Objectives This study assesses the outcome of Low dose contrast protocol during PCI in CKD patients whose e-GFR 60 ml/min/1.72 m and investigates a safety margin for contrast use in these high-risk categories.?Methods: Patients were into three groups according to CV/e-GFR ratio: Group (A) low-dose: CV/e-GFR ratio 2.0 Group (B) medium-dose: CV/e-GFR ratio > 2.0 and × bodyweight\s.creatinine). Group (C) high-dose: CV/e-GFR ratio > MACD. Results: A total of 73 patients were enrolled. Average age was 54 ± 8 years,81.4% were male and 18.6% were females and 52% were diabetic. Mean baseline e-GFR was 40 ± 8.0 ml/min/1.73m2. Contrast Volume used in group A was (58.26 ± 15.05) (n = 24), in group B (109.42 ± 17.11) (n = 26) and in group C (304.5 ± 60.30) (n = 23), respectively. The incidences of CI-AKI in the 3 groups were 0%, 11.5% and 35%, respectively (p = 0.02). All-cause death 0%, 17% and introduction of maintenance hemo dialysis was 0%, 11.5% and 26%, respectively (p Conclusion: Low dose contrast protocol is safe, effective and easily applicable technique without CI-AKI or death.
