TEM Data Acquisition With Nonlinear Image Shift Calibration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for single particle analysis (SPA) in cryogenic electron microscopy face significant targeting errors due to non-linear components, sample height variations, and limitations in image shift calibration, which restrict the range of image shifts and increase the number of stage moves, thereby prolonging data acquisition time.
Innovation Solution
The proposed method improves image shift calibration by accounting for non-flatness and tilt of the sample, using non-linear models based on Zernike polynomials or Gaussian process regression to correct targeting errors, and allows for on-the-fly adjustment of calibration points, enabling longer-range image shifts and reducing the need for mechanical stage movements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If image shift calibration uses linear models, then the calibration process is simple, but targeting errors increase due to non-linear components
Solution Approach 1:
The patent applies Zernike polynomials to transform the calibration parameters from simple linear coefficients to polynomial coefficients that capture non-linear optical aberrations. This parameter transformation enables the calibration model to account for non-linear components while maintaining a systematic and manageable calibration process.
Solution Approach 2:
The patent introduces an intermediate calibration step where Zernike polynomial coefficients are computed from observed targeting errors. This intermediary calculation layer bridges the gap between simple linear calibration and accurate non-linear correction, allowing the system to progressively refine targeting accuracy.
2Measurement precision
If the range of image shifts is limited, then targeting errors are reduced, but the number of stage moves increases
Solution Approach 1:
The patent expands the effective range of accurate image shifts by transforming the calibration parameters to include non-linear terms. This allows the system to maintain targeting accuracy even when performing larger image shifts, thereby reducing the frequency of mechanical stage moves and accelerating data acquisition.
Solution Approach 2:
The patent performs preliminary calibration using Zernike polynomials to establish accurate targeting parameters before data acquisition. This preliminary action enables the system to execute larger image shifts with confidence, knowing that the non-linear corrections will maintain accuracy throughout the extended range.
3Speed
If linear calibration models are used, then calibration computation is fast, but targeting errors due to non-linear components cannot be corrected
Solution Approach 1:
The patent transforms the calibration parameters from linear to polynomial form using Zernike polynomials. This transformation maintains computational efficiency while enabling the model to capture non-linear optical aberrations, thus improving targeting accuracy without sacrificing calibration speed.
Solution Approach 2:
The patent replaces the mechanical trial-and-error calibration approach with a mathematical model based on Zernike polynomials. This substitution allows for rapid computation of calibration parameters while accurately accounting for non-linear effects, achieving both speed and precision.
4Device complexity
If sample height variations are not accounted for, then calibration is simpler, but targeting errors increase
Solution Approach 1:
The patent incorporates sample height variations by adding z-offset parameters to the Zernike polynomial model. This parameter extension allows the calibration to account for non-flat samples while maintaining the systematic structure of the polynomial approach, balancing complexity and accuracy.
Solution Approach 2:
The patent extends the calibration model from two-dimensional (x, y) coordinates to three-dimensional (x, y, z) by incorporating sample height (z-offset) variations. This dimensional extension enables the model to account for non-flat sample surfaces, improving targeting accuracy for samples with height variations.
Data Source
AI summary
A method of automated data acquisition for a transmission electron microscope, the method comprising: obtaining a reference image of a sample at a first magnification; for each of a first plurality of target locations identified in the reference image: steering an electron beam of the transmission electron microscope to the target location, obtaining a calibration image of the sample at a second magnification greater than the first magnification, and using image processing techniques to identify an apparent shift between an expected position of the target location in the calibration image and an observed position of the target location in the calibration image, training a non-linear model using the first plurality of target locations and the corresponding apparent shifts; based on the non-linear model, calculating a calibrated target location for a next target location; steering the electron beam to the calibrated target location and obtaining an image at a third magnification greater than the first magnification.