基于焦平面估计的快速扫片方法
2020-05-21卢祎张雪媛何勇军
卢祎 张雪媛 何勇军



摘 要:随着仪器自动化的发展,基于显微镜的自动阅片系统被广泛应用于医学领域。该系统需要在显微镜视野下移动多个位置并拍摄标本图像,以完成整个扫描任务。但是由于载物台平面与物镜中轴线不能保证绝对的垂直,或者观测样本表面的起伏,使得各个视野的焦点不在一个平面内,有必要通过聚焦移动以获得清晰图像。传统方法在每个位置都进行聚焦,导致聚焦移动次数过多,扫片的效率较低。为了减少扫片的时间,提出了一种基于焦平面估计的快速扫片方法。通过采集有代表性位置的焦点,估计其附近位置的焦点,在扫描载玻片时可以直接移动到估计的焦点位置完成聚焦。实验表明此方法能有效减少焦点搜索的次数,提高扫片效率。
关键词:自动聚焦;自动显微镜;焦平面估计;多点平面分割方法;平面判断方法
DOI:10.15938/j.jhust.2020.01.008
中图分类号: TH742
文献标志码: A
文章编号: 1007-2683(2020)01-0051-08
Abstract:With the development of instrument automation, the microscope-based automatic reading system is used widely in the medical fieldThis system needs moving many positions under the microscope and capturing the images to complete the scanning taskHowever, because it is difficult to ensure the plane of the platform and the axis of the object lens are vertical, and the surface of the observed samples may be not even, the focuses in different locations are not in a planeTherefore, it is necessary to run auto focusing to obtain clear imagesConventional methods perform focusing at each location, causing excessive movements and inefficient scanningIn order to reduce the scanning time, this paper presents a fast scanning method based on focal plane estimationBy acquiring the focuses at several representative positions, the focuses of the position near them are estimated, and the platform(or the lens) can be moved directly to the estimated positionExperiments show that this method can reduce the number of focus searching effectively and improve the scanning efficiency-
Keywords:automatic focusing; automatic microscope; focal plane estimation; multi-point plane segmentation method; method of determining the plane
0 引 言
隨着电子显微镜的出现和图像处理技术的发展,基于人工智能的病理分析已经成为当下最主流的研究方向。这种技术首先在显微镜下采集脱落细胞或者组织切片的图片,然后分析图像病变情况,为病理诊断提供辅助。由于相机视野宽度有限,分析一个样本需要采集多幅图像。一个载玻片样本的扫描一般需要采集300张以上的图片。
全自动显微镜是通过自动移动载物平台使电子显微镜采集不同位置视野的图像。由于无法确保载物台平面与显微镜观察方向完全垂直,或者观测样本表面的起伏使焦点不在同一个平面内,导致显微镜下的图像不清晰。……
