深度学习技术识别腹腔镜手术视频中肾动脉的可行性  

Feasibility of deep learning for renal artery detection in laparoscopic video

作　　者：赵欣[1] 廖章诚 王旭[1] 马琳[1] 周敬敏[1] 樊华[1] 张玉石[1] 徐维锋[1] 纪志刚[1] 李汉忠[1] 花苏榕[2] 李佳颐 周家权[3] Zhao Xin;Liao Zhangcheng;Wang Xu;Ma Lin;Zhou Jingmin;Fan Hua;Zhang Yushi;Xu Weifeng;Ji Zhigang;Li Hanzhong;Hua Surong;Li Jiayi;Zhou Jiaquan(Department of Urology,Peking Union Medical College Hospital,Chinese Academy of Medical Sciences,Beijing 100730,China;Department of General Surgery,Peking Union Medical College Hospital,Chinese Academy of Medical Sciences,Beijing 100730,China;Department of Urology,Hainan General Hospital(Hainan Affiliated Hospital of Hainan Medical University),Haikou 570311,China)

机构地区：[1]中国医学科学院北京协和医院泌尿外科,北京100730 [2]中国医学科学院北京协和医院基本外科,北京100730 [3]海南省人民医院(海南医学院附属海南医院)泌尿外科,海口570311

出　　处：《中华泌尿外科杂志》2022年第10期751-757,共7页Chinese Journal of Urology

基　　金：中央高水平医院临床科研业务费(2022-PUMCH-B-008)。

摘　　要：目的探讨深度学习技术在经腹膜后入路腹腔镜肾脏手术视频中对肾动脉识别的可行性。方法回顾性分析2020年1月至2021年7月北京协和医院实施的87例腹腔镜经腹膜后入路肾脏手术视频资料,包括根治性肾切除术、肾部分切除术、半尿路切除术。由2名泌尿外科医生筛选出包含肾动脉的视频片段,经过抽帧、标注、审核、校对后,用随机数字表法按4∶1比例将标注目标分为训练集和测试集。训练集用于训练神经网络模型,测试集用于测试不同难度场景下神经网络识别肾动脉的能力,统一输送至YOLOv3卷积神经网络模型进行训练。测试集根据2名高年资医生意见,按肾动脉与周围组织的区分度分为高、中、低辨认度。高辨认度即肾动脉干净,暴露面积大;中辨认度为肾动脉有一定程度浸血,暴露面积中等;低辨认度为肾动脉暴露面积小,常位于镜头边缘,浸血严重,可能存在镜头模糊。标注人员在手术视频中逐帧标注肾动脉真值框。所有图像经过归一化、预处理后输入神经网络模型进行训练。神经网络输出肾动脉预测框,与真值框重合的交并比(IOU)高于设定的阈值则判定为预测正确。记录测试集神经网络测试结果,根据IOU计算敏感性、精确率。结果本研究87个视频共提取5457帧图像,其中49个视频片段共4490个目标为训练集,38个视频片段共1135个目标为测试集。训练集中13个视频1149个目标为高辨认度,17个视频1891个目标为中辨认度,19个视频1450个目标为低辨认度。测试集中,9个视频267个目标为高辨认度,11个视频519个目标为中辨认度,18个视频349个目标为低辨认度。测试集IOU阈值为0.1时,敏感性和精确率分别为52.78%和82.50%;IOU阈值为0.5时,敏感性和精确率分别为37.80%和59.10%。IOU阈值为0.1时,高、中、低辨认度组的敏感性和精确率分别为89.14%和87.82%、45.86%和78.03%、32.95%和76.67%。实时�Objective To explore the feasibility of deep learning technology for renal artery recognition in retroperitoneal laparoscopic renal surgery videos.Methods From January 2020 to July 2021,the video data of 87 cases of laparoscopic retroperitoneal nephrectomy,including radical nephrectomy,partial nephrectomy,and hemiurorectomy,were retrospectively analyzed.Two urological surgeons screened video clips containing renal arteries.After frame extraction,annotation,review,and proofreading,the labeled targets were divided into training set and test set by the random number table in a ratio of 4∶1.The training set was used to train the neural network model.The test set was used to test the ability of the neural network to identify the renal artery in scenes with different difficulties,which was uniformly transmitted to the YOLOv3 convolutional neural network model for training.According to the opinion of two senior doctors,the test set was divided into high,medium,and low discrimination of renal artery and surrounding tissue.High identification means a clean renal artery and a large exposed area.For middle recognition degree,the renal artery had a certain degree of blood immersion,and the exposed area was medium.Low identification means that the exposed area of the renal artery was small,often located at the edge of the lens,and the blood immersion was severe,which may lead to lens blurring.In the surgical video,the annotator annotated the renal artery truth box frame by frame.After normalization and preprocessing,all images were input into the neural network model for training.The neural network output the renal artery prediction box,and if the overlap ratio(IOU)with the true value box was higher than the set threshold,it was judged that the prediction was correct.The neural network test results of the test set were recorded,and the sensitivity and accuracy were calculated according to IOU.Results In the training set,1149 targets of 13 videos had high recognition degree,1891 targets of 17 videos had medium recognition de

关 键 词：肾动脉 腹腔镜经腹膜后入路肾脏手术 人工智能 深度学习 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TP18[自动化与计算机技术—计算机科学与技术] R699[医药卫生—泌尿科学]