Automatic adjustment of laparoscopic pose using deep reinforcement learning

IF 1 4区 工程技术 Q4 ENGINEERING, MECHANICAL
Lingtao Yu, Yongqiang Xia, Pengcheng Wang, Lining Sun
{"title":"Automatic adjustment of laparoscopic pose using deep reinforcement learning","authors":"Lingtao Yu, Yongqiang Xia, Pengcheng Wang, Lining Sun","doi":"10.5194/ms-13-593-2022","DOIUrl":null,"url":null,"abstract":"Abstract. Laparoscopic arm and instrument arm control tasks are usually accomplished by an operative doctor. Because of intensive workload and long operative time, this method not only causes the operation not to be flow,\nbut also increases operation risk. In this paper, we propose a method for\nautomatic adjustment of laparoscopic pose based on vision and deep\nreinforcement learning. Firstly, based on the Deep Q Network framework, the raw laparoscopic image is taken as the only input to estimate the Q values\ncorresponding to joint actions. Then, the surgical instrument pose\ninformation used to formulate reward functions is obtained through object-tracking and image-processing technology. Finally, a deep neural network\nadopted in the Q-value estimation consists of convolutional neural networks\nfor feature extraction and fully connected layers for policy learning. The\nproposed method is validated in simulation. In different test scenarios, the\nlaparoscopic arm can be well automatically adjusted so that surgical\ninstruments with different postures are in the proper position of the field\nof view. Simulation results demonstrate the effectiveness of the method in learning the highly non-linear mapping between laparoscopic images and the optimal action policy of a laparoscopic arm.\n","PeriodicalId":18413,"journal":{"name":"Mechanical Sciences","volume":" ","pages":""},"PeriodicalIF":1.0000,"publicationDate":"2022-06-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Mechanical Sciences","FirstCategoryId":"5","ListUrlMain":"https://doi.org/10.5194/ms-13-593-2022","RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q4","JCRName":"ENGINEERING, MECHANICAL","Score":null,"Total":0}
引用次数: 0

Abstract

Abstract. Laparoscopic arm and instrument arm control tasks are usually accomplished by an operative doctor. Because of intensive workload and long operative time, this method not only causes the operation not to be flow, but also increases operation risk. In this paper, we propose a method for automatic adjustment of laparoscopic pose based on vision and deep reinforcement learning. Firstly, based on the Deep Q Network framework, the raw laparoscopic image is taken as the only input to estimate the Q values corresponding to joint actions. Then, the surgical instrument pose information used to formulate reward functions is obtained through object-tracking and image-processing technology. Finally, a deep neural network adopted in the Q-value estimation consists of convolutional neural networks for feature extraction and fully connected layers for policy learning. The proposed method is validated in simulation. In different test scenarios, the laparoscopic arm can be well automatically adjusted so that surgical instruments with different postures are in the proper position of the field of view. Simulation results demonstrate the effectiveness of the method in learning the highly non-linear mapping between laparoscopic images and the optimal action policy of a laparoscopic arm.
使用深度强化学习的腹腔镜姿势自动调整
摘要腹腔镜臂和器械臂的控制任务通常由手术医生完成。由于工作量大、手术时间长,这种方法不仅导致手术不流畅,而且增加了手术风险。在本文中,我们提出了一种基于视觉和深度强化学习的腹腔镜姿势自动调整方法。首先,基于Deep Q网络框架,将原始腹腔镜图像作为唯一的输入来估计对关节动作的Q值。然后,通过目标跟踪和图像处理技术获得用于制定奖励函数的手术器械姿态信息。最后,Q值估计中采用的深度神经网络由用于特征提取的卷积神经网络和用于策略学习的全连接层组成。仿真验证了该方法的有效性。在不同的测试场景中,腹腔镜臂可以很好地自动调整,使不同姿势的外科器械处于视野的适当位置。仿真结果证明了该方法在学习腹腔镜图像和腹腔镜臂的最佳动作策略之间的高度非线性映射方面的有效性。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
求助全文
约1分钟内获得全文 求助全文
来源期刊
Mechanical Sciences
Mechanical Sciences ENGINEERING, MECHANICAL-
CiteScore
2.20
自引率
7.10%
发文量
74
审稿时长
29 weeks
期刊介绍: The journal Mechanical Sciences (MS) is an international forum for the dissemination of original contributions in the field of theoretical and applied mechanics. Its main ambition is to provide a platform for young researchers to build up a portfolio of high-quality peer-reviewed journal articles. To this end we employ an open-access publication model with moderate page charges, aiming for fast publication and great citation opportunities. A large board of reputable editors makes this possible. The journal will also publish special issues dealing with the current state of the art and future research directions in mechanical sciences. While in-depth research articles are preferred, review articles and short communications will also be considered. We intend and believe to provide a means of publication which complements established journals in the field.
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术官方微信