In the current implementation of the thick-slice rendering, the image border is simply ignored.
The ImageVtkMapper2D uses instances of mitk::ExtractSliceFilter and vtkMitkThickSlicesFilter to calculate the thick-slice projection.
The ExtractSliceFilter is told to extract the current slice + 'thickSlicesNum' in both directions.
Thus, a 'thickSlicesNum' of 1 will add one slice on each side of the current slice, a value of 2 will add two slices on each side, etc.
This is fine for the 'central slices' of the image. On the border however, this is not very intuitive. If we are currently showing the first slice and we add an additional slice on top of that, we have no pixel-information to consider in the vtkMitkThickSlicesFilter.
The solution would be to make sure that the desired sliceThickness is applied by appending those slices, that cannot be displayed in one direction, to the other direction (and vice versa when reaching the other end of the image).
See Wiki-Page for more details...