融合通道层注意力机制的UNet的衍射极限荧光点检测和定位  

Channel-Wise Attention Mechanism Relevant UNet-Based Diffraction-Limited Fluorescence Spot Detection and Localization

作　　者：余永建 王越 李寰 周文超 舒风风 高明 吴一辉 Yu Yongjian;Wang Yue;Li Huan;Zhou Wenchao;Shu Fengfeng;Gao Ming;Wu Yihui(School of Ophthalmology&Optometry,Wenzhou Medical University,Wenzhou 325035,Zhejiang,China;Key Laboratory of Optical System Advanced Manufacturing Technology,Changchun Institute of Optics,Fine Mechanics and Physics,Chinese Academy of Sciences,Changchun 130033,Jilin,China;University of Chinese Academy of Sciences,Beijing 100049,China)

机构地区：[1]温州医科大学眼视光学院,浙江温州325035 [2]中国科学院长春光学精密机械与物理研究所光学系统先进制造技术重点实验室,吉林长春130033 [3]中国科学院大学,北京100049

出　　处：《激光与光电子学进展》2023年第14期245-254,共10页Laser & Optoelectronics Progress

基　　金：国家自然科学基金(U21A20395)。

摘　　要：针对高通量荧光显微成像中高密度、低信噪比、亚衍射极限荧光斑点的自动化精准检测和定位问题,基于UNet提出一种轻量级神经网络方法。该方法采用挤压和激发通道层注意力机制和残差模块优化特征信息,构建密度图和偏移量多输出架构,直接执行检测和亚像素定位。在公开数据集和模拟数据集进行实验,所提方法对低信噪比和高密度的荧光点检测优于当前算法,尤其对于达到衍射极限的高密度荧光点,有很好的检测性能,比如在128×128像素具有1200个荧光点并且大部分点达到衍射极限的图像下。所提算法对斑点的识别精度F1分数超过97.6%,定位误差为0.115 pixel,相比最新deepBlink方法,F1提升16.2个百分点并且定位误差减小0.63 pixel。This paper proposes a lightweight neural network method based on UNet to accurately detect and localize highdensity,low signal-to-noise ratio(SNR)sub-diffraction fluorescence spots in high-throughput fluorescence microscopy imaging.This method combines a squeeze and excitation channel-wise attention mechanism with a residual module to optimize feature information.A density map and offset multioutput architecture are also constructed for direct detection and subpixel localization.The proposed method has been verified on public and simulated datasets,and outperforms current algorithms for low SNR and high-density fluorescent spot detection.Notably,the detection performance of the proposed method is excellent for high-density fluorescent spot that reaches the diffraction limit,such as in images with a resolution of 128×128 pixels having 1200 fluorescent spots.The spot detection accuracy(F1 score)of the proposed algorithm exceeds 97.6%,and the localization error is 0.115 pixel.Compared with the latest deepBlink method,the F1 of the proposed algorithm has improved by 16.2 percentage points,and the localization error has been reduced by 0.63 pixel.

关 键 词：荧光显微镜 数字图像处理 模式识别 神经网络 医学和生物成像 

分 类 号：TP751.1[自动化与计算机技术—检测技术与自动化装置]