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
Objective To explore the value of deep learning(DL)models semi-automatic training system for automatic optimization of clinical image quality control of transthoracic echocardiography(TTE).Methods Totally 1250 TTE vid...Objective To explore the value of deep learning(DL)models semi-automatic training system for automatic optimization of clinical image quality control of transthoracic echocardiography(TTE).Methods Totally 1250 TTE videos from 402 patients were retrospectively collected,including 490 apical four chamber(A4C),310 parasternal long axis view of left ventricle(PLAX)and 450 parasternal short axis view of great vessel(PSAX GV).The videos were divided into development set(245 A4C,155 PLAX,225 PSAX GV),semi-automated training set(98 A4C,62 PLAX,90 PSAX GV)and test set(147 A4C,93 PLAX,135 PSAX GV)at the ratio of 5∶2∶3.Based on development set and semi-automatic training set,DL model of quality control was semi-automatically iteratively optimized,and a semi-automatic training system was constructed,then the efficacy of DL models for recognizing TTE views and assessing imaging quality of TTE were verified in test set.Results After optimization,the overall accuracy,precision,recall,and F1 score of DL models for recognizing TTE views in test set improved from 97.33%,97.26%,97.26%and 97.26%to 99.73%,99.65%,99.77%and 99.71%,respectively,while the overall accuracy for assessing A4C,PLAX and PSAX GV TTE as standard views in test set improved from 89.12%,83.87%and 90.37%to 93.20%,90.32%and 93.33%,respectively.Conclusion The developed DL models semi-automatic training system could improve the efficiency of clinical imaging quality control of TTE and increase iteration speed.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。
文摘Objective To explore the value of deep learning(DL)models semi-automatic training system for automatic optimization of clinical image quality control of transthoracic echocardiography(TTE).Methods Totally 1250 TTE videos from 402 patients were retrospectively collected,including 490 apical four chamber(A4C),310 parasternal long axis view of left ventricle(PLAX)and 450 parasternal short axis view of great vessel(PSAX GV).The videos were divided into development set(245 A4C,155 PLAX,225 PSAX GV),semi-automated training set(98 A4C,62 PLAX,90 PSAX GV)and test set(147 A4C,93 PLAX,135 PSAX GV)at the ratio of 5∶2∶3.Based on development set and semi-automatic training set,DL model of quality control was semi-automatically iteratively optimized,and a semi-automatic training system was constructed,then the efficacy of DL models for recognizing TTE views and assessing imaging quality of TTE were verified in test set.Results After optimization,the overall accuracy,precision,recall,and F1 score of DL models for recognizing TTE views in test set improved from 97.33%,97.26%,97.26%and 97.26%to 99.73%,99.65%,99.77%and 99.71%,respectively,while the overall accuracy for assessing A4C,PLAX and PSAX GV TTE as standard views in test set improved from 89.12%,83.87%and 90.37%to 93.20%,90.32%and 93.33%,respectively.Conclusion The developed DL models semi-automatic training system could improve the efficiency of clinical imaging quality control of TTE and increase iteration speed.
