The Siberian-Icelandic hotspot track is the only preserved continental hotspot track. Although the track and its associated age progression between 160 Ma and 60 Ma are not yet well understood, this section of the tra...The Siberian-Icelandic hotspot track is the only preserved continental hotspot track. Although the track and its associated age progression between 160 Ma and 60 Ma are not yet well understood, this section of the track is closely linked to the tectonic evolution of Amerasian Basin, the Alpha-Mendeleev Ridge and Baffin Bay. Using paleomagnetic data, volcanic structures and marine geophysical data, the paleogeography of Arctic plates （Eurasian plate, North American Plate, Greenland Plate and Alaska Microplate） was reconstructed and the Siberian-Icelandic hotspot track was interlinked between 160 Ma and 60 Ma. Our results suggested that the Alpha-Mendeleev Ridge could be a part of the hotspot track that formed between 160 Ma and 120 Ma. During this period, the hotspot controlled the tectonic evolution of Baffin Bay and the distribution of mafic rock in Greenland. Throughout the Mesozoic Era, the aforementioned Arctic plates experienced clockwise rotation and migrated northeast towards the North Pacific. The vertical influence from the ancient Icelandic mantle plume broke this balance, slowing down some plates and resulting in the opening of several ocean basins. This process controlled the tectonic evolution of the Arctic.展开更多
目的基于心脏磁共振(cardiac magnetic resonance,CMR)电影序列提取影像组学特征,并将其与临床特征与图像特征相结合,旨在构建射血分数降低型心力衰竭(heart failure with reduced ejection fraciton,HFrEF)患者的预后预测模型,并对模...目的基于心脏磁共振(cardiac magnetic resonance,CMR)电影序列提取影像组学特征,并将其与临床特征与图像特征相结合,旨在构建射血分数降低型心力衰竭(heart failure with reduced ejection fraciton,HFrEF)患者的预后预测模型,并对模型效能进行验证。材料与方法回顾性纳入2018年1月至2023年4月期间根据指南诊断为HFrEF并接受CMR检查的患者共503例。收集所有患者的临床基线信息、实验室检查、心电图以及部分超声指标作为电子健康记录,并完成随访。复合终点事件定义为包括心源性死亡、心衰再入院以及心脏移植等在内的不良心血管事件。所有患者均行标准CMR检查。使用单因素Cox分析确定与结局最为相关的临床变量。使用无监督nnU-netv2算法提取CMR电影序列中的功能学参数作为图像特征,并使用开源软件包从同一序列中提取影像组学特征。经过组间及组内一致性检验后,使用最大相关性和最小冗余方法对于组学特征进行筛选及降维。选择性能最佳的机器学习(machine learning,ML)分类器构建最终预测模型。建立的模型包括单纯影像组学模型、组学+临床特征混合模型以及组学+图像特征混合模型。通过曲线下面积(area under the curve,AUC)、准确度、精确度、召回率以及F1分数等评估模型的预测效能。结果经过严格纳排后,共纳入389例HFrEF患者用于模型构建。在1041(212,1238)天的中位随访时间内,87例患者发生了既定的结局事件(事件率22.4%),中位生存时间为495(8,1900)天。通过单因素Cox回归共纳入了包含NYHAⅢ/Ⅳ以及BNP在内的12个临床特征。在经过特征筛选及降维后,共纳入4个图像特征以及9个组学特征。在所使用的6个不同的分类器中,集成学习分类器表现最佳。在该分类器方法得到的输出结果中,组学+图像特征模型达到了最佳的预后预测效能,AUC为0.789,准确度为81.6%,精确度为72.5%,召回率为71.6%,F1分数为72.0%。结论本研究基于非增强CMR电影序列,创新性地使用影像组学方法,构建了具有较好预测效能的HFrEF的预后预测模型。在单纯影像组学模型基础上分别增加临床特征及反映心功能的图像特征可以提升模型的预测效能。展开更多
基金supported by a grant from the China Ocean Mineral Resources Research and Development Association Project(Grant No.DY125-12-R-03)
文摘The Siberian-Icelandic hotspot track is the only preserved continental hotspot track. Although the track and its associated age progression between 160 Ma and 60 Ma are not yet well understood, this section of the track is closely linked to the tectonic evolution of Amerasian Basin, the Alpha-Mendeleev Ridge and Baffin Bay. Using paleomagnetic data, volcanic structures and marine geophysical data, the paleogeography of Arctic plates （Eurasian plate, North American Plate, Greenland Plate and Alaska Microplate） was reconstructed and the Siberian-Icelandic hotspot track was interlinked between 160 Ma and 60 Ma. Our results suggested that the Alpha-Mendeleev Ridge could be a part of the hotspot track that formed between 160 Ma and 120 Ma. During this period, the hotspot controlled the tectonic evolution of Baffin Bay and the distribution of mafic rock in Greenland. Throughout the Mesozoic Era, the aforementioned Arctic plates experienced clockwise rotation and migrated northeast towards the North Pacific. The vertical influence from the ancient Icelandic mantle plume broke this balance, slowing down some plates and resulting in the opening of several ocean basins. This process controlled the tectonic evolution of the Arctic.
文摘目的基于心脏磁共振(cardiac magnetic resonance,CMR)电影序列提取影像组学特征,并将其与临床特征与图像特征相结合,旨在构建射血分数降低型心力衰竭(heart failure with reduced ejection fraciton,HFrEF)患者的预后预测模型,并对模型效能进行验证。材料与方法回顾性纳入2018年1月至2023年4月期间根据指南诊断为HFrEF并接受CMR检查的患者共503例。收集所有患者的临床基线信息、实验室检查、心电图以及部分超声指标作为电子健康记录,并完成随访。复合终点事件定义为包括心源性死亡、心衰再入院以及心脏移植等在内的不良心血管事件。所有患者均行标准CMR检查。使用单因素Cox分析确定与结局最为相关的临床变量。使用无监督nnU-netv2算法提取CMR电影序列中的功能学参数作为图像特征,并使用开源软件包从同一序列中提取影像组学特征。经过组间及组内一致性检验后,使用最大相关性和最小冗余方法对于组学特征进行筛选及降维。选择性能最佳的机器学习(machine learning,ML)分类器构建最终预测模型。建立的模型包括单纯影像组学模型、组学+临床特征混合模型以及组学+图像特征混合模型。通过曲线下面积(area under the curve,AUC)、准确度、精确度、召回率以及F1分数等评估模型的预测效能。结果经过严格纳排后,共纳入389例HFrEF患者用于模型构建。在1041(212,1238)天的中位随访时间内,87例患者发生了既定的结局事件(事件率22.4%),中位生存时间为495(8,1900)天。通过单因素Cox回归共纳入了包含NYHAⅢ/Ⅳ以及BNP在内的12个临床特征。在经过特征筛选及降维后,共纳入4个图像特征以及9个组学特征。在所使用的6个不同的分类器中,集成学习分类器表现最佳。在该分类器方法得到的输出结果中,组学+图像特征模型达到了最佳的预后预测效能,AUC为0.789,准确度为81.6%,精确度为72.5%,召回率为71.6%,F1分数为72.0%。结论本研究基于非增强CMR电影序列,创新性地使用影像组学方法,构建了具有较好预测效能的HFrEF的预后预测模型。在单纯影像组学模型基础上分别增加临床特征及反映心功能的图像特征可以提升模型的预测效能。
