一种多尺度图像融合的冷冻电镜颗粒挑选方法

A METHOD FOR SINGLE PARTICLE SELECTION IN CRYO-EM BASED ON MULTI-SCALE IMAGE FUSION

当前主流的冷冻电镜颗粒挑选方法往往需要大量人工生成的训练集或者优质颗粒模板,或者颗粒挑选过程极为复杂。为了提高冷冻电镜颗粒挑选的效率,简化颗粒挑选流程,提出一种自动挑选颗粒方法,在图像预处理阶段使用基于Lanczos采样图像融合方法提高图像质量,随后使用基于最大类间方差的图像阈值分割方法分离颗粒与背景,实现颗粒挑选。在EMPAIR公共数据集的实验结果表明,该方法与其他方法相比,具有更高的召回率与精确率。

The current mainstream particle selection methods for Cryo-EM often require massive artificially generated training sets or high-quality particle templates,or the particle selection process is extremely complicated.In order to improve the efficiency of cryo-electron microscopy particle selection and simplify the particle selection process,this paper proposes an automatic particle selection method.In the image preprocessing stage,the Lanczos sampling image fusion method was used to improve the image quality,and the image threshold segmentation method based on the maximum inter-class variance was used to separate the particles and the background to achieve particle selection.The experimental results on the EMPAIR public data set show that the proposed method has a higher recall rate and accuracy rate compared with other methods.