Contrast Transfer Function, Points, and Pixels

Microscopical images are now almost always recorded digitally. To accomplish this, the flux of photons that forms the final image must be divided into small geometrical subunits called pixels. The light intensity in each pixel will be stored as a single number. Changing the objective magnification, the zoom magnification on your confocal control panel, or choosing another coupling tube magnification for your charge-coupled device (CCD) camera changes the size of the area on the object that is represented by one pixel. If you can arrange matters so that the smallest feature recorded in your image data is at least 4 to 5 pixels wide in each direction, then all is well. This process is diagrammed for a laser-scanning confocal in Figure 4.1, where the diameter of the scanning beam is shown to be at least four times the interline spacing of the scanning raster. This means that any individual fluorescent molecule should be excited by at least four overlapping, adjacent scan lines and that, along each scan line, it will contribute signal to at least four sequential pixels. Finally, it is important to remember that information stored following these rules will only properly resemble the original light pattern if it is first spatially filtered to remove noise signals that are beyond the spatial bandwidth of the imaging system. Image deconvolution is the most accurate way of imposing this reconstruction condition and this applies equally to data that have been collected by widefield or scanning techniques. If you do this right, your image should look like that in Figure 4.2. If you are already convinced of this, jump to page 71 for the second half of this chapter, on gray levels. But if it all seems to be irrelevant mumbo-jumbo, read on. Incorrect digitization can destroy data.