Using the Navigator

The Navigator window and its associated menu entries provide the capability to save stage positions and return to them, to make and save maps of areas of interest, mark positions on these maps and go to the marked positions, accurately return to positions on maps, and transform positions after moving the grid in the specimen holder. This section explains some of the concepts involved in using the Navigator.

Navigator Items
Maps
Groups of Items
Returning to a Position Reliably
Imaging States
Registrations
A Step-by-Step Procedure for Mapping
Finding Points after Grid Rotation
A Counter-intuitive Point about Misaligned Montage Frames
Memory Considerations

Navigator Items

The Navigator maintains a list of items which are shown in a box. It considers items to be of three types: points, polygons, and maps. A point is a single location, which will be displayed with a colored cross on images that contain the point. A polygon is set of connected points enclosing a region, which will be displayed as a set of lines forming a closed contour. A map is an image that has been saved to a file; the Navigator keeps track of where the image was saved as well as the stage position of the image and the conditions under which the image was acquired. The location of a map is displayed as a rectangle on other images. The Navigator makes it easy to reload a selected map image, regardless of whether all of your maps are in one file or separate files, and of whether the files are open or not. Maps can be either single frame or montaged images.

Points can be added in two different ways. You can move the stage to a specific location and then press a button ('Add Stage Pos') to add the location as a point. Alternatively, once you have taken a low-mag image at a known stage location, you can press a button ('Add Points') to start adding points, then click the left mouse button at a series of locations in the image. Each location becomes a new point item with a separate stage position deduced from its location in the image. A polygon can be generated in a similar fashion, by pressing 'Add Polygon' and clicking points on an image.

Maps

A map can be created from an acquired image simply by pushing a button after saving it to a file, or even before saving it if the image is compatible with the current open file. A montaged image can be made into a map as long as the montage overview is still in buffer B. There is also a feature for automatically acquiring a series of maps at selected points. Again, these maps can be either single frame images or montages. They will all be saved into a single file that must be set up before starting the automatic acquisition.

The large-area images used for maps are usually generated by taking montages with stage movements between the frames. The Navigator menu provides three ways to define the area that will be acquired in a montage. One option is simply to define an area that will capture the entire grid ('Setup Full Montage'). Another option is to move the stage to a series of points at the corners of the region of interest, such as the corners of the first and last sections of a ribbon. When each of these points is marked as a 'corner' point, the 'Setup Corner Montage' menu entry will set up a montage to encompass all of the area bounded by these points. The third option is to draw a polygon; the 'Setup Polygon Montage' menu entry will set up a montage to capture the area delineated by the polygon. When you select one of these options, the Montage Setup dialog will open in a special mode in which it will adjust the number of frames to fit the selected area whenever you change the magnification or the overlap between adjacent montage frames. Note that in all of these cases, a full rectangular array of frames need not be acquired; the program will omit pieces that fall outside of the defined area.  After a montage has been set up to acquire inside a polygon, it can be used for acquiring the same shape of area in multiple locations without drawing a polygon at those locations, as long as they are acquired as additional sections in the same file.

Groups of Items

When points are added in a single operation, they are defined as belonging to a group. A group is created when you add a series of points by clicking in an image, add a grid of points with a menu command, or define a set of positions for 'supermontaging' (acquiring a set of overlapping montages). The dialog box has an option for collapsing the listing of group items into a single line summarizing those items; this can help when there are many points in a group. Several commands operate on all the items in a group.

Returning to a Position Reliably

The Navigator incorporates several features to help you get to a desired location despite the imperfections of the microscope. Image misalignments between the magnifications can become quite large at very low magnification. These misalignments limit the ability to mark a feature in an image at one magnification and go to that feature at a much higher magnification. If the image shifts needed to compensate for the misalignments have been calibrated, you can choose to have these image shifts applied so that features stay centered when you change magnification. Even if you do not turn on this option, the Navigator will use these image shifts to adjust stage positions appropriately for the particular magnification. With these adjustments, if you mark a feature in an image taken at, say 200x, and then go to that position while at a magnification like 3000x, you should be able to find the feature in an image taken at that magnification. In contrast, if you go to the position while still at 200x, the stage will be moved to the appropriate position for centering the feature at 200x, and it may not appear at the higher magnification. Note also that these adjustments are of limited accuracy and will generally not be adequate for keeping a feature in the field of view across a 100-fold range of magnification.

A second problem with returning to a marked position is that the actual stage position at which a feature is centered may change by a few microns over time, possibly due to drift, shrinkage, or accumulation of errors from extensive stage movements. There are two features that help deal with this problem. One is the ability to shift the stage positions of all of the items by the same amount. This can be done by adding a point at a defined feature in an existing map, then taking a current image that shows the feature and clicking on the feature there. The 'Shift to Marker' menu entry is then used to shift the points. After this, you should be able to click on any point in that map and go to the point, without the error that was just eliminated by the shift. However, a different shift correction may be needed for distant points.

The second feature that helps in returning to a marked position is a routine for 'realigning' to a position by correlation with an existing map taken in regular mag mode. In order to return reliably to a position, it must be located in a map that is at least as big as the maximum possible stage error, and preferably twice as large. Thus, points to be realigned to should probably be located in a map at least 10 microns in extent. They can be near the edge of such a map, since it is the overall size of the map and not the distance of the point from the edge that matters. Do not expect reliable realignment to a point in a map that is only 2 microns in extent.

If you wish to realign precisely to a high-magnification image, such as for doing a tilt series, then you will need two maps: the medium-mag map for reliably returning to the vicinity of the point, and a high-mag image stored as a map. Three approaches to doing this are described next, along with their abilities to deal with the three problems that limit the reliability of realign routine.  These problems are referred to as: "intrinsic error in returning to the position", which was described above; "mags not well-aligned", which means that there is a misalignment between the medium and high mags and that there is either no image shift offset calibration or an inaccurate one; and "backlash", which refers to the extent that the stage does not return to the same position when it is approached from a different direction from the original one.  Note that if you take the medium-mag map in Low Dose mode with View images (single frame or montage), it is possible to correct a misalignment between the mags with a View shift offset (see Shift offset for View in Low Dose). 

  1. Anchor maps.  The simplest method is just to take two map images at every desired location, the high-mag image and one at, say, 3000x. The latter map is referred to as an 'anchor map'. You can store these in the same file or different files. After you have stored one map at the medium mag, you can mark this as a map to be used as a template for making more anchor maps. Thereafter, you can create a new anchor map with a single button press.  The specific steps to follow are:
       a) Find the first desired location.  Acquire and save a high-mag image and make it a map with the New Map button.
       b) Go to the medium mag suitable for realigning, acquire and save an image, and make it a map.
       c) Turn on the For anchor state checkbox.
       d) Find the next desired location.  At the high mag, take an image, shift with the right mouse button to show the desired area, and take a new image.  Press the Anchor Map button.  The current image will be saved as a map and the program will acquire and save the anchor map too.
       e) Repeat step (d) at  other locations.
    This method works well even if the mags are not well-aligned, and backlash will not contribute to errors in returning to the position if you allow the program to adjust for backlash when it takes the anchor map (or use a montage for the anchor map).  The method is most suitable when you do not already have a medium-mag map.
  2. Mark points and use Acquire at Points to get high-mag maps.  This approach takes advantage of an existing montaged map covering a larger area. Add one or more points on one or more medium-mag maps and select 'Acquire' for each one.  Run the 'Acquire at Points' menu item to acquire at map at each point and specify 'Realign to item' as an initial action.  This method will give robust return to this position and realignment to the high-mag map, but it is suitable only if the mags are well enough aligned so that the desired area is adequately centered in the high-mag map.
  3.  Move to positions marked in medium-mag maps, adjust positions and take maps at high mag.  In its simplest form, namely by clicking in the map, using Go To Marker, then moving around at high mag to center the area of interest, this method is vulnerable both in the initial repositioning to the medium-mag map and in the alignment to the high-mag map.  The initial repositioning is impaired to the extent there are already errors in returning to the medium-mag map position, which could be the case if numerous maps were taken before marking points on this one.  The alignment to the high-mag map could fail if the mags are misaligned by more than 1/3 of the extent of the high-mag map.  There are two ways to make this approach more reliable:
       a) Acquire a single image anchor map for each high-mag map: i.e., use method 1 above.  This eliminates the effects of pre-existing positioning errors and of misaligned mags.
       b) Add a point instead of just clicking at the desired location.  Be sure to be at the higher mag, and press Realign to Item. Then center on the desired location using image shift only, not stage shift.  Take an image and make it a map.  The errors found in realigning to the nearby point will be recorded for this map and taken into account when realigning again; backlash will also not be a problem.  Realignment to the high-mag map could still fail if the mags are misaligned too much.

The lessons from above are that methods relying an existing montaged map work best if the mags are well aligned, either intrinsically or with an accurate image offset calibration, and that using an Anchor map should give the most reliable repositioning.

On FEI microscopes, if a medium-magnification montaged map is taken with stage movement, there are some advantages in taking the montage with the option to 'Realign with image shift' (available when the option 'Use settings for high-quality stage montage' is selected).  Because of periodic, systematic errors in Compustage movement, stage montages with piece sizes around 5-10 microns may require overlaps of 20-25% to achieve sufficient overlap.  Using 'Realign with image shift' will allow the montage to be taken with an overlap of only 10%, provided that there are sufficient image features present for the this realignment to work.  The more important advantage for mapping is that when the Navigator moves to a point marked on a map acquired this way, it can adjust by the errors in stage position found during montaging.   The 'Realign to item' routine also takes account of the stage errors appropriately.  Taking montages with image shift realignment is particularly recommended for maps to be used for setting up supermontages.

Imaging States

The Imaging State dialog allows you to store and set the microscope state and Record acquisition parameters, thus making it easier to acquire images at different magnifications. Since this is particularly helpful when making the different kinds of Navigator maps, the dialog will generally open along with the Navigator. The dialog also allows you to set a state for acquiring another map into the same file as an existing map.

Registrations

The final concept that needs explaining is 'registration'. A registration is essentially one position of the grid in the specimen holder, or one relationship between positions on the grid and stage coordinates. Every item in the Navigator's list, as well as every image that you take, is associated with a registration. When you first start using the Navigator, all of the items that you add to the list are assumed to be at the first registration. If you disturb the grid in the holder (or even if you remove and replace the holder), you have changed the correspondence between positions on the grid and stage coordinates. You should then indicate to Navigator that the registration has changed by increasing the current registration number. At this point, items from previous registrations will no longer appear on newly acquired images (unless you select an option to make them appear) because they would not be in the right place. To get them into the right place and make them useful again, you need to transform them to the new registration.

To transform items to a new registration, you must first define some of the Navigator points at the first registration as registration points. You do this by selecting the 'Registration point' checkbox and, if necessary, specifying a point number as well for each of the relevant points. (Registration point numbers will increment automatically.) After you have gone to a new registration, you then move the stage to the same positions, add a new Navigator point at each, and identify each point as a registration point with the same number as in the first registration. Once you have identified a set of corresponding points in this way, select a menu entry to transform the items from the first registration to the new one. The program will find a transformation whose complexity depends on the number of registration points available. Just a shift will be found with 1 point, a rotation and shift with 2 points, rotation, scaling, and shift with 3 or 4 points, or a full linear transformation including stretching with 5 points. All items except the registration points will then be transformed and associated with the new registration. (The registration points are left behind to allow the data to be transformed back if necessary.) If you just need to adjust by a shift after removing and replacing the holder or after moving the stage around a lot, you can use just one registration point, or use the 'Shift to Marker' approach described above.

A Step-by-Step Procedure for Mapping

A suggested procedure for using the Navigator for mapping would be:

  1. Go to a magnification that can be used to map the area of interest in a reasonable number of large pixels, on the order of 10K x 10K (e.g., 10 by 10 pieces, each 1K x 1K). This might be as low as 160x.
  2. Move the stage to the corners of the area of interest and add the stage position as a point at each corner. Define each point as a corner point, then select 'Setup Corner Montage'. Alternatively, especially for cryomicroscopy, select 'Setup Full Montage'.
  3. Acquire the montage and designate it as a map.
  4. On the montage overview, add a point at each location of interest (e.g., corresponding location on serial sections, or grid squares with ice of the right thickness.) Select 'Acquire' for each point to set it up for automatic acquisition.
  5. At one of the locations of interest, draw a polygon around the area desired for a medium-mag map. Go to an intermediate mag (like 3000x) so that a map of this area will be a reasonable number of pixels. Select 'Setup Polygon Montage' to define a montage area.
  6. Acquire a montage to check the area.
  7. If necessary, shift the registration. Add a point on an identifiable feature on the low-mag map. Mark the same position on the new image with the left mouse button. Select 'Shift to Marker'.
  8. Select 'Acquire Areas' to acquire all of the areas automatically. Go do something else; these maps may take about 5 minutes each.
  9. When you load an individual map after it has been acquired, you can simply click in the map then press 'Go To Marker' to move the stage to the selected location. For more accurate positioning, go to the mag at which you want to view a feature, add a point item on the map, and use 'Realign to Item'. You could then acquire a tilt series at high mag.
  10. You do not need to close the montage file, but be sure to open a new file before trying to start a tilt series.

It is also possible to capture a series of intermediate mag maps or montages into a separate file for each.  The procedure would be slightly different for steps 4 and 5:

    4. At one of the locations of interest, draw a polygon around the area desired for a medium-mag map and select 'Acquire'. Go to the intermediate mag and select 'New file at item'.  In the little dialog that comes up, select to make a montage and fit it to the polygon.
   5. Mark all of the other desired locations with a series of points.  For each one, turn on 'Acquire', then select 'New file at item'.  Each new file will inherit the properties of the polygon montage.

If you need to have a different polygon at each location, such as when capturing from a curving ribbon of sections, then this will require a different montage file for each location.  At each location, draw a polygon, select 'Acquire'  then 'New file at item', and then go through the sequence for fitting a polygon to the montage.

Finding Points after Grid Rotation

If you want to find corresponding points after rotating a grid by 90 degrees, you could use the following procedure:

  1. Move the stage to five or six well-defined points (e.g., the corners of a section of interest), add the stage position as a Navigator point, and make each one a registration point.
  2. Add a point or map(s) at each location of interest (i.e., each place where a tilt series is taken.)
  3. Rotate the grid.
  4. Increase the current registration number to 2.
  5. Find the five or six registration points and add each one as a registration point in registration 2, making sure that the registration point numbers correspond to the first registration.
  6. Select 'Transform Items' in the Navigator menu.
  7. Select one of the points of interest in the Navigator list box and press 'Go To XY'. You should be close to the desired point. Errors of 5-10 microns are likely if you localized the registration points in low mag mode; but the error should be much less if you localized them at higher mag.
  8. If you saved medium-mag maps at any locations, you can use 'Realign to Item' to return to and align to those locations. The program will rotate the map image in order to align to it. You will probably need larger maps, 15-20 µm in extent, to realign reliably after a rotation.

This procedure is tedious because of the need for so many registration points. The number of points can be reduced to 3 or even 2 if the stretch in the stage coordinate system has been calibrated. It is calibrated simply using a transformation obtained from 6-8 registration points as just described, as long as the points fit with moderately low error (see Stage Stretch).

If you use the Align with Rotation dialog to align to points after rotation, you could use the following procedure to get a preliminary transformation with that dialog and replace it with one based on registration points:

  1. Make sure that you have medium mag maps at least 15-20 µm in extent at two or more locations separated by more than half of the extent of the area being studied.
  2. Rotate the grid. If you have a rotating holder and can watch the rotation, do the rotation at one of the places with a map if that will help you find it again.
  3. Increase the current registration to 2.
  4. Find one of the locations with a map. Take an image with the same mag.
  5. Load the map.
  6. Open the Align with Rotation dialog , set the angle to rotate the image by, and press Align to Map.
  7. Toggle between the aligned image and map ('Home' and 'Delete' keys) to check the alignment.
  8. If the alignment is good, press Apply Image Shift on Scope and Transform.
  9. In the Navigator, press Add Stage Pos.
  10. Select another appropriate map, load the map, and press Realign to Item.
  11. Take an image and verify that it is centered on the middle of the map. Press Add Stage Pos.
  12. You could repeat the last two steps at another position.
  13. In the Navigator menu, select 'Undo Transformation'.
  14. Make each of the maps that you aligned to and each of the added positions in the new registration be registration points with corresponding registration numbers.
  15. Run 'Transform Items' in the Navigator menu.
  16. If the realignment fails in step 10, run 'Undo Transformation' and reload the map. Then try to find the location. Align to it and add a stage position as in steps 6 to 9. Now you have a second point to use for a transformation.

A Counter-intuitive Point about Misaligned Montage Frames

SerialEM will sometimes fail to align all of the montage frames in the overview image properly, particularly when there are grid bars over the zones of overlap between adjacent pieces. It is important to realize that such a misalignment generally does not affect the accuracy with which a map can be used. A position marked on a montage map is converted to a stage coordinate by determining which frame the position is in and the pixel coordinate within that frame, and then deriving a stage coordinate solely from this pixel position and the stage coordinates at which that frame was acquired. The marked coordinate will thus not depend on whether the pieces in the overview are properly aligned, misaligned, or not aligned at all. (You can demonstrate this by adding a point at the same position on a map after reloading it with and without piece alignment turned on - the coordinates that appear in the Navigator table will be the same.) The accuracy of both moving to marked positions and using the Realign to Item routine at them should not depend on whether a map is well-aligned. Two actions do require a well-aligned map: adding a grid of points over a set of holes, and setting up an array of supermontage positions.

Memory Considerations

Large maps are memory intensive. If you keep many of them around by copying them to other buffers, you can easily run out of memory and start using swap file space. You should generally use binning to make them more information-dense, but the amount of binning depends on the kind of camera, the size of pixels in the detector, and the voltage.  The most binning is needed for CCD cameras at 200 and 300 KV, which should generally be binned to ~1Kx1K for the most space-efficient images.  At lower voltages, cameras have much better resolution and maps could be taken unbinned, although binning by 2 will still provide images with more information per pixel.  The same is true for direct detectors at higher voltages.  The K2 camera in counting mode is a special case; it has good information content in unbinned images but the dose rate needs to be kept in the appropriate range for counting mode.  Thus, you could take map images unbinned with relatively long exposures, or use binning by 2 with shorter exposures.  In any case, binning must be considered in conjunction with the magnification being used.  For mapping, a particular resolution or pixel size is needed to achieve the desired level of detail. If you double the binning you need to increase the magnification to compensate.  If you are constrained to stay above a certain magnification, that in turn may dictate the binning needed to prevent huge maps.

The memory use of maps is minimized by an option (in the Navigator menu) to convert maps to bytes, which is on by default. With this conversion, a map will typically occupy as many bytes as pixels (e.g., a 10K by 10K image will require 100 MB). Without the conversion, a map will take 3 times as much memory because the image is stored as 2-byte integers and an additional byte array of the same size is needed for display. You might need to turn off the option if the byte conversion is truncating the intensities inappropriately so that you cannot see light or dark details. If this occurs, turn off the conversion, read in the map, and type new values in the White or Black text boxes of the Image Display control panel to keep the desired intensities from being truncated. Be sure to turn the option back on.

Even when conversion to bytes is on, the acquisition of a new map will take 3 times as much memory as the number of pixels, because the scaling to bytes cannot be known until the whole montage is acquired. As soon as the montage is designated as a map, however, it will be converted to bytes and the extra memory will be available again. If the program cannot get memory for the initial overview image when acquiring a montage, it will increase the binning of this image to reduce the memory requirement. After you make such an image a map, you can then reload it to see it at full resolution, since this will require less memory.

If performance starts to suffer, use the Task Manager to assess the memory and swap file usage. Also notice the memory usage indicator in the Buffer Control Panel. To recover the memory occupied by a map in a higher buffer, make the image be the current one and use the Delete button in that control panel. If you are going to use the Navigator extensively, get as much memory as possible for the computer and make sure the swap file space is large (1 GB). Because the behavior of the computer is usually atrocious when significant swapping occurs, it is probably preferable to set the TotalMemoryLimitMB property to prevent the program from requesting large amounts of memory for montage overviews when memory use is near the limit (See General entries ).