CHARACTERIZING AND CORRECTING IMAGE DISTORTIONS ON THE TECNAI MICROSCOPES

这个子项目是许多研究子项目中利用 资源由NIH/NCRR资助的中心拨款提供。子项目和 调查员(PI)可能从NIH的另一个来源获得了主要资金， 并因此可以在其他清晰的条目中表示。列出的机构是 该中心不一定是调查人员的机构。 自从开始在我们的Tecnai显微镜上获取数据以来，我们一直担心图像失真会损害我们的断层重建质量。对于Gatan成像滤镜(GIF)的图像，这个问题尤其严重，因为使用GIF和使用GIF所需的低倍率镜头模式都会导致失真。我们开发了一个程序来求解从两个阶段位置的一组重叠的样本图像的图像变形场。一对图像的两个图像应该重叠约55%，并且必须获得四个不同方向的图像对。我们的数据采集程序SerialEM有一个很容易获取这些对的例程，所以现在可以直接获取这样的数据并计算图像畸变场。另一种扭曲发生在倾斜的样本上，因为它位于一个随Z高度变化的放大率和图像旋转的梯度中。当蒙太奇拍摄倾斜的样品的重叠图像时，这种梯度的效果会非常明显。在尝试将帧移入寄存器时存在大量错误；当对渐变失真进行校正时，错误被消除。通过在两个不同的Z高度拍摄样品的图像并确定它们之间的大小变化和旋转，可以很容易地测量梯度。该梯度仅取决于光束强度，因此通过记录每个倾斜序列中的光束强度，可以校正该梯度的图像。这两种梯度的校正都已纳入我们的标准层析成像处理流程，因此用户很容易应用它们。到目前为止，已经发现对GIF图像畸变场的校正显著地改善了图像对齐，而对梯度的校正则给出了不显著的改进。

This subproject is one of many research subprojects utilizing the resources provided by a Center grant funded by NIH/NCRR. The subproject and investigator (PI) may have received primary funding from another NIH source, and thus could be represented in other CRISP entries. The institution listed is for the Center, which is not necessarily the institution for the investigator. Since starting to acquire data on our Tecnai microscopes, we have been concerned about image distortions impairing the quality of our tomographic reconstructions. The problem is particularly severe with images from the Gatan Imaging Filter (GIF), because of distortions induced in both the GIF and the low magnification lens mode needed to use the GIF. We developed a program to solve for the image distortion field from a set of pairs of overlapping images of a specimen at two stage positions. The two images of a pair should overlap by about 55%, and pairs at four distinct orientations must be obtained. Our data acquisition program, SerialEM, has a routine to acquire these pairs easily, so it is now straightforward to acquire such data and compute an image distortion field. Another kind of distortion occurs with a tilted specimen, because it is sitting in a gradient in which magnification and image rotation vary with Z height. The effect of this gradient shows up prominently when a montage is taken of overlapping images of a tilted specimen. There are substantial errors when trying to shift the frames into register; the errors are eliminated when gradient distortions are corrected for. The gradient can be measured easily by taking images with the specimen at two different Z heights and determining the size change and rotation between them. The gradient depends only on beam intensity, so by recording beam intensity during each tilt series it is possible to correct images for this gradient. Corrections for both kinds of gradients have been incorporated into our standard processing stream for tomography, so they are easy for users to apply. So far, correcting for the GIF image distortion field has been found to improve image alignment substantially, while correcting for the gradients has given insignificant improvements.