基于形状感知的显微图像目标定位方法  

Object Localization Method in Microscopic Image Based on Shape Perception

作　　者：马静超 康文彬 李欣凝 陈薪 班晓娟[1,2,3,6,8] MA Jingchao;KANG Wenbin;LI Xinning;CHEN Xin;BAN Xiaojuan(Beijing Advanced Innovation Center for Materials Genome Engineering,University of Science and Technology Beijing,Beijing 100083,China;Shunde Innovation School,University of Science and Technology Beijing,Foshan 528300,China;School of Intelligence Science and Technology,University of Science and Technology Beijing,Beijing 100083,China;Collaborative Innovation Center of Steel Generic Technology,Beijing University of Science and Technology,Beijing 100083,China;HBIS Group Company Limited,Shijiazhuang 050023,China;Key Laboratory of Intelligent Bionic Unmanned Systems,Ministry of Education,University of Science and Technology Beijing,Beijing 100083,China;Department of Reproductive Medicine Center,Shunde Hospital,Southern Medical University(The First People's Hospital of Shunde),Foshan 528399,China;Institute of Materials Intelligent Technology,Liaoning Academy of Materials,Shenyang 110004,China)

机构地区：[1]北京科技大学北京材料基因工程高精尖创新中心,北京100083 [2]北京科技大学顺德创新学院,佛山528300 [3]北京科技大学智能科学与技术学院,北京100083 [4]北京科技大学钢铁共性技术协同创新中心,北京100083 [5]河钢集团有限公司,石家庄050023 [6]北京科技大学智能仿生无人系统教育部重点实验室,北京100083 [7]生殖医学中心南方医科大学顺德医院(顺德第一人民医院),佛山528399 [8]辽宁材料实验室材料智能技术研究所,沈阳110004

出　　处：《北京邮电大学学报》2023年第6期33-38,共6页Journal of Beijing University of Posts and Telecommunications

基　　金：佛山市科技创新专项资金项目(BK21BF002,BK22BF010)。

摘　　要：针对卵细胞浆内单精子注射显微图像中注射针针尖点定位不准确的问题,提出了一种基于形状感知的显微图像针尖定位方法。通过将目标形状和边界信息作为约束条件引入图像分割网络的损失函数中,使卷积神经网络在训练过程中更关注目标的形状和边界特征,从而提高针尖点的定位准确度。实验结果表明,基于形状感知的显微图像针尖定位方法在3种常用基线分割模型上的性能提高了7%以上,即所提方法可作为即插即用的模块以提高算法的泛化能力,并且对针尖的定位效果优于领域内其他方法。To deal with the inaccurate positioning issue of the needle tip in Intracytoplasmic sperm injection microscopic images,a shape-sensing-based microscopic image needle tip positioning method is proposed.In particular,the target shape and boundary information are proposed as constraints on the loss function of the image segmentation network,driving the convolutional neural network to pay more attention to the shape and boundary features of the target during the training process.The experimental results show that the performance of the proposed loss function is increased by more than 7%compared to the three commonly used baseline segmentation models.Thus,the method can be used as a plug-and-play module to improve the generalization ability of the algorithm.In addition,its localization accuracy exceeds the state-of-the-art methods in the field.

关 键 词：卵细胞浆内单精子注射 显微影像 针尖点定位 深度学习 形状感知 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]