Setting Up Image Correlation Drift Correction

The input for drift correction is the localization file (*_LOC.txt) for a DeltaVision localization microscopy data set.

The outputs from drift correction include a drift-corrected localization file and a drift correction log, which are saved. Additionally, a Drift Correction Results plot and a Localization Analysis Log are displayed to help the user determine the quality of the drift correction. If desired, users can save the Drift Correction Results image as a .jpeg file.

To set up image correlation drift correction:

1. From the Localization Image Analysis window, click the Correction tab to select the Correction portion of the window.

2. Click the Localization Data button to browse and select the localization file for correction. Alternatively, you can click and drag an image window number into this field. This file must be a localization file (*_LOC.txt).

3. Activate the Do drift correction check box.

4. Activate the Use weighted center-of-mass calculation check box to use this faster, but somewhat less precise calculation instead of the default, Gaussian fit method.

5. In the Time window field, enter a time window (in seconds) to divide up the data. This field determines how much time passes between each drift-corrected point. A good starting point is choosing a time that breaks the data into 10 to 15 time windows.

Notes: 1) The Time window field indicates the elapsed time between drift measurements. In most cases, the drift will be slow and linear so it can be corrected between time points by doing a linear interpolation. Therefore, the drift correction time window does not limit the time-resolution of the drift correction, as a drift measurement for each frame is inferred from interpolation.

2) A shorter time period means that less time has elapsed between corrections. This will increase the time resolution of the drift correction, allowing for a more accurate correction of the lateral stage drift. If the window is too short, however, there may not be enough data in the time window to reconstruct images with recurring features between time windows.

6. In the Pixel size field, enter a pixel size for the reconstruction. Image registration is performed to determine how much the image has shifted in the time that has elapsed in the time window. The pixel size helps to determine the sharpness of the features used to do image registration. A smaller pixel size will make the features sharper, but will also increase the time required to perform the drift correction.

Note: The pixel size does not affect the resolution of the drift correction. It only affects the quality of the correlation functions used to determine the drift. Therefore, the drift is still determined to sub-pixel resolution because the centroid of the correlation function for image registration is determined using either Gaussian fitting or a weighted center-of-mass calculation.

7. Select the Show drift correction results check box to display a graph of the x- and y-drift values measured by the drift correction algorithm.

8. Click the Run the Correction Step(s) button to perform the drift correction. When the process is complete, the Drift Results graph and the Localization Analysis Log are displayed.

The drift-corrected localization file is saved as a text file denoted *_LDC.txt with each row corresponding to a single detected fluorophore. The *_LDC.txt file includes ten columns where the x-coordinate and the y-coordinate of each fluorophore have been shifted appropriately to correct for the measured drift.