Region of Interest

Before a scan can be carried out, the region of interest (ROI) must first be defined. This will be a rectangular area so that a succession of regular stage movements in the X and Y directions bring each part of the ROI into view of the camera. To complete the task the focal plane must also be defined to account for variation in sample height with respect to the objective.

To define a rectangular area on a plane we essentially need to specify the lengths of the sides and its position. The ROI panel gives 3 methods to do this:

Corners
Using the joystick to move to 2 points to specify opposite corners of the rectangle is enough to define the ROI. Click on 'Set start', use the joystick to move to one of the corners, then press 'OK' when in position. Repeat using 'Set end'.



Edges
Rather than moving specifically to 2 corners we can also use the 4 edges of the rectangle to locate our ROI. This requires moving to 4 points rather than 2 but has the advantage of not having to precisely find the corners. In any order, and as before, use the 'Set' buttons along with the joystick to move to any points along the 4 edges.



Square
For the special case of the ROI being a square we may specify the centre of the square and one point on the border. This is enough to uniquely define our ROI. Again the 'Set' buttons along with the joystick are used to set these 2 points.



Note: numerical input
Rather than use the joystick to physically move to the points required it is also possible to enter the desired co-ordinates directly. A mixture of the 2 methods can also be used.

After we defined the ROI it is time to define the focal plane.
This is necessary so that as the sample moves in X and Y we can change the stage Z position to maintain focus.

Use points inside/outside ROI - Focusing, which requires live imaging, may expose cells to harmful UV or cause bleaching. If this occurs inside the ROI then the cells scanned will not be 'equal' - some will have experienced different conditions than others. This would suggest that the better option would be to focus by imaging cells or objects outside the ROI. This assumes of course that there are cells to see which might not be the case. Similarly, cells on the outskirts of the ROI may not be representative and lead to incorrect focusing. So knowledge of the cell pattern will help deciding which option to take.

Setup focal plane
The stage will go to the first focal position and a live image is presented. Use the Z-drive to achieve focus, moving in X and Y if necessary. Clicking 'OK' stores the information and moves to the second point. After 3 points have been completed a focal plane will have been defined. The three points to which the stage automatically drove formed an equilateral triangle which minimises error in the calculation.

Set fixed offset
If it is known that the focus varies little across the ROI then it may be desirable to simply set a single focus to define a flat focal plane. This saves time and minimises cell exposure to the illumination. When selected, the stage drives to a single point, the user is asked to focus using the joystick, and on completion this Z value will be used for any susequent scan.

Checking the results
Before initiating the scan we may like to check how well the focusing set up is going to perform. The final tab of the ROI panel allows the user to go to various points and see how accurate the calculated focus position is. If, at this stage, it is found that the results are unsatisfactory then the process can be repeated.



Possible problems
Focusing can be a tricky business. There are 3 main sources of problems.

The 3 point procedure assumes that the objects lie on a flat surface at some angle with respect to the plane of the stage. So we are assuming the substrate is flat (at least across the ROI). Now if, for example, we are looking at cells on a sheet of mylar containing media there may be some sag in the centre due to the weight of the media and dependent on how tightly stretched the mylar is on the cell dish. In this case the cells will not lie on a plane but on a 3D parabolic surface and the method will fail. Similarly, if there were a wrinkle in the mylar then focusing on a point here will generate a spurious Z value which will lead to an erroneous result. Possible solutions, apart from ensuring correct construction of the cell dish, are to reduce the scan area (since the planar approximation will hold better over shorter distances), change the scan area position to a 'flatter' region, or to use single point focusing in the hope that a fixed stage height will provide adequate focusing for the majority of the cells.

Secondly, the 3 point procedure requires well-spaced points to be accurate. Any 2 points define the gradient of the plane in that direction - how much the Z height changes over the distance between points. Consider 2 points close together - with perfect focusing the gradient is still well-defined. However, if there is a small error in the Z values then the error in the gradient becomes large since any inaccuracy is then magnified at larger distances. This situation can arise when there are few objects at the nominal focal points and the user has to hunt for something to focus on. Often, since a sense of position can easily be lost when using the joystick, the user ends up at similar locations and this situation occurs. The only real solution here is to start again and, preferably with knowledge of where objects are likely to be found, take care to control the direction of the stage movement with the joystick.

Somewhat related to both of these issues is when one of the focal points is simply incorrect. If the user focuses on a foreign object which isn't in the cellular plane, for example, an incorrect focal plane will be set. Also, if the user moves too far from the ROI to find an object then the planar approximation, which is a local approximation, may no longer hold (as in the example of the sagging cell dish).

In all these situations, where focusing is not straightforward, it is worth looking at the sample live under a non-damaging illumination to see if there is anything unexpected or to reassess the position of the objects of interest.

Additionally, a problem may occur if the plane defined takes the Z drive beyond its limit of travel at one edge of the slide. The microscope should warn if this happens. In this case you should follow the instructions to refocus at a central position using a coarse Z drive. If the plane defined exceeds both limits of the Z drive, the software will advise that a smaller region should be scanned. You can also make sure that the sample is not mounted at an angle (e.g. as the cover slip is too close to the edge of the slide, and the slide it sitting on it, in the stage insert of an inverted microscope; the Open microscopes Abbe and Galileo are inverted).