目的:探讨基于深度学习(deep learning,DL)的ResNet+VST模型在超声心动图关键帧智能检测方面的可行性。方法:选取南京大学医学院附属鼓楼医院超声医学科采集的663个动态图像含心尖二腔(apical two chambers,A2C)、心尖三腔(apical three...目的:探讨基于深度学习(deep learning,DL)的ResNet+VST模型在超声心动图关键帧智能检测方面的可行性。方法:选取南京大学医学院附属鼓楼医院超声医学科采集的663个动态图像含心尖二腔(apical two chambers,A2C)、心尖三腔(apical three chambers,A3C)与心尖四腔(apical four chambers,A4C)3类临床检查常用切面以及EchoNet⁃Dynamic公开数据集中280个A4C切面动态图像,分别建立南京鼓楼医院数据集与EchoNet⁃Dynamic⁃Tiny数据集,各类别图像按4∶1方式划分为训练集和测试集,进行ResNet+VST模型的训练以及与多种关键帧检测模型的性能对比,验证ResNet+VST模型的先进性。结果:ResNet+VST模型能够更准确地检测心脏舒张末期(end⁃diastole,ED)与收缩末期(end⁃systole,ES)图像帧。在南京鼓楼医院数据集上,模型对A2C、A3C和A4C切面数据的ED预测帧差分别为1.52±1.09、1.62±1.43、1.27±1.17,ES预测帧差分别为1.56±1.16、1.62±1.43、1.45±1.38;在EchoNet⁃Dynamic⁃Tiny数据集上,模型对A4C切面数据的ED预测帧差为1.62±1.26,ES预测帧差为1.71±1.18,优于现有相关研究。此外,ResNet+VST模型有良好的实时性表现,在南京鼓楼医院数据集与EchoNet⁃Dynamic⁃Tiny数据集上,基于GTX 3090Ti GPU对16帧的超声序列片段推理的平均耗时分别为21 ms与10 ms,优于以长短期记忆单元(long short⁃term memory,LSTM)进行时序建模的相关研究,基本满足临床即时处理的需求。结论:本研究提出的ResNet+VST模型在超声心动图关键帧检测的准确性、实时性方面,相较于现有研究有更出色的表现,该模型原则上可推广到任何超声切面,有辅助超声医师提升诊断效率的潜力。展开更多
Endorepellin plays a key role in the regulation of angiogenesis,but its effects on angiogenesis after traumatic brain injury are unclear.This study explored the effects of endorepellin on angiogenesis and neurobehavio...Endorepellin plays a key role in the regulation of angiogenesis,but its effects on angiogenesis after traumatic brain injury are unclear.This study explored the effects of endorepellin on angiogenesis and neurobehavioral outcomes after traumatic brain injury in mice.Mice were randomly divided into four groups:sham,controlled cortical impact only,adeno-associated virus(AAV)-green fluorescent protein,and AAV-shEndorepellin-green fluorescent protein groups.In the controlled cortical impact model,the transduction of AAV-shEndorepellin-green fluorescent protein downregulated endorepellin while increasing the number of CD31+/Ki-67+proliferating endothelial cells and the functional microvessel density in mouse brain.These changes resulted in improved neurological function compared with controlled cortical impact mice.Western blotting revealed increased expression of vascular endothelial growth factor and angiopoietin-1 in mice treated with AAV-shEndorepellin-green fluorescent protein.Synchrotron radiation angiography showed that endorepellin downregulation promoted angiogenesis and increased cortical neovascularization,which may further improve neurobehavioral outcomes.Furthermore,an in vitro study showed that downregulation of endorepellin increased tube formation by human umbilical vein endothelial cells compared with a control.Mechanistic analysis found that endorepellin downregulation may mediate angiogenesis by activating vascular endothelial growth factor-and angiopoietin-1-related signaling pathways.展开更多
文摘目的:探讨基于深度学习(deep learning,DL)的ResNet+VST模型在超声心动图关键帧智能检测方面的可行性。方法:选取南京大学医学院附属鼓楼医院超声医学科采集的663个动态图像含心尖二腔(apical two chambers,A2C)、心尖三腔(apical three chambers,A3C)与心尖四腔(apical four chambers,A4C)3类临床检查常用切面以及EchoNet⁃Dynamic公开数据集中280个A4C切面动态图像,分别建立南京鼓楼医院数据集与EchoNet⁃Dynamic⁃Tiny数据集,各类别图像按4∶1方式划分为训练集和测试集,进行ResNet+VST模型的训练以及与多种关键帧检测模型的性能对比,验证ResNet+VST模型的先进性。结果:ResNet+VST模型能够更准确地检测心脏舒张末期(end⁃diastole,ED)与收缩末期(end⁃systole,ES)图像帧。在南京鼓楼医院数据集上,模型对A2C、A3C和A4C切面数据的ED预测帧差分别为1.52±1.09、1.62±1.43、1.27±1.17,ES预测帧差分别为1.56±1.16、1.62±1.43、1.45±1.38;在EchoNet⁃Dynamic⁃Tiny数据集上,模型对A4C切面数据的ED预测帧差为1.62±1.26,ES预测帧差为1.71±1.18,优于现有相关研究。此外,ResNet+VST模型有良好的实时性表现,在南京鼓楼医院数据集与EchoNet⁃Dynamic⁃Tiny数据集上,基于GTX 3090Ti GPU对16帧的超声序列片段推理的平均耗时分别为21 ms与10 ms,优于以长短期记忆单元(long short⁃term memory,LSTM)进行时序建模的相关研究,基本满足临床即时处理的需求。结论:本研究提出的ResNet+VST模型在超声心动图关键帧检测的准确性、实时性方面,相较于现有研究有更出色的表现,该模型原则上可推广到任何超声切面,有辅助超声医师提升诊断效率的潜力。
基金supported by the National Natural Science Foundation of China,Nos.81801236(to ZX),81974189(to HT)a grant from Shanghai Sixth People’s Hospital Affiliated to Shanghai Jiao Tong University School of Medicine,No.ynlc201719(to QZ).
文摘Endorepellin plays a key role in the regulation of angiogenesis,but its effects on angiogenesis after traumatic brain injury are unclear.This study explored the effects of endorepellin on angiogenesis and neurobehavioral outcomes after traumatic brain injury in mice.Mice were randomly divided into four groups:sham,controlled cortical impact only,adeno-associated virus(AAV)-green fluorescent protein,and AAV-shEndorepellin-green fluorescent protein groups.In the controlled cortical impact model,the transduction of AAV-shEndorepellin-green fluorescent protein downregulated endorepellin while increasing the number of CD31+/Ki-67+proliferating endothelial cells and the functional microvessel density in mouse brain.These changes resulted in improved neurological function compared with controlled cortical impact mice.Western blotting revealed increased expression of vascular endothelial growth factor and angiopoietin-1 in mice treated with AAV-shEndorepellin-green fluorescent protein.Synchrotron radiation angiography showed that endorepellin downregulation promoted angiogenesis and increased cortical neovascularization,which may further improve neurobehavioral outcomes.Furthermore,an in vitro study showed that downregulation of endorepellin increased tube formation by human umbilical vein endothelial cells compared with a control.Mechanistic analysis found that endorepellin downregulation may mediate angiogenesis by activating vascular endothelial growth factor-and angiopoietin-1-related signaling pathways.