During FIB-SEM image acquisition, the alignment of the image slices, typically along the Z-axis, is not maintained due to, e.g., specimen drift or image distortions. Accordingly, a Slice Alignment is necessary.

Please make sure that the image is oriented with the misalignments in Z-direction before starting to align the slices. This is best observed by sliding through the 2D slices in Z-direction. Typically, offsets are visible when looking at images in X- and Y-direction. If the misalignments occur in another direction, either rotate or permute the image first.

Two options are available for the Alignment Method:

• For Correlation the algorithm tries to shift the slices in X- and Y-direction till the regions are as similar as possible within the defined region of interest (ROI).
• The SIFT (Scale-Invariant Feature Transform) algorithm can recognize features and tries to map the features to each other by shifting and rotating the slices. This works very well, if the features are very similar in the different slices. Find more detailed information on this method here.

The choices for ROI First Slice and ROI Last Slice for X and Y define the region, in which the alignment is examined. Based on the findings for that region, the slice alignment is corrected for the entire image. Please note that this might lead to unavoidable boundary effects that might require a Crop at the image boundaries. The rectangle resulting from the values for First Slice and Last Slice is visualized in red in the 2D Slice Visualization section in the X-Y-plane if Slice Alignment is chosen for Overlay.

For the Correlation method, the Radius defines the maximum shift between two slices to search for. For many FIB-SEM images, the default value of 10 is enough. However, if there are slices shifted more than 10 voxels, the Radius setting needs to be increased.

For SIFT, the SIFT Threshold for the image features used can be defined as Low, Medium, or High. With a higher threshold only more significant image features are used and with a lower threshold more image features are used. If the alignment does not modify the image, choose a lower threshold. If on the other hand, the alignment is not correct, a higher threshold can be better. When setting the threshold to Automatic, the thresholds are adjusted automatically.

The example below shows a 2D slice from X-direction of an artificially misaligned gray value image (left). After application of the Slice Alignment tool at the suggested default settings, the misalignment is entirely corrected (right).

Observe the offsets on the left and right boundaries of the aligned image (right). Accordingly, the image size in X- and Y-direction has increased in order to include all aligned slices in the 3D image without loss of information.