The horizontal axis in my histogram represents the quantity of information in my image, or the optical density values. In this particular instance, I have densities ranging from a value of 0 (absolute white) on the left to 256 (absolute black) on the right side of my image.
*Note: Keep in mind that this histogram was taken using "Image J" so the bit depth is limited. On a traditional CR system, and depending on the bit depth your system utilizes, you may see up to 16,000 plus shades of gray. This histogram is simply for demonstration.
The vertical axis represents the number of pixels that are assigned within each density value.
In the following image, one of the great aspects of the Image J software is displayed. Depending on where my cursor is placed on the histogram, you can tell on each image what the density value is for each column, as well as how many pixels were assigned that density value. In the image on the left, the value (or degree of density on a scale of 0-256) is 73. The count (or the number of pixels exposed with this density level) is 456. In the image on the right, my cursor was moved over a higher density value of 188, and a higher count of 5883. So in this image, the crosshair placement represents a darker density assigned to a greater number of pixels.
Just as a point of reference, this is the image represented by the histogram:
The CR system has a pre-programmed algorithm for each type of exam you do... this is what you are selecting when you input "chest lateral" before image plate scanning. This tells the computer that a histogram similar to the pre-programmed shape will be scanned. This is where the CR system can produce errors. Depending on the raw data that is scanned, the computer will assign a range on your histogram termed "values of interest" or VOI. Unfortunately, the Image J software did not include this, but I have represented the VOI with red lines in the following image:
The VOI on the histogram helps to determine your Exposure Index (Kodak) or S Number (Fuji), as well as how your image will be rescaled. In first generation CR systems, the operator could slide the VOI to the left or to the right in order to visualize recorded anatomy better, but most current applications do not allow the radiographer to do so. It may be possible, however, to apply a different LUT (lookup table) to the image to make adjustments. So when the computer applies automatic rescaling (the computer's attempt to adjust the image due to over/under exposure) to your image, the process may fail if the original histogram analysis is incorrect. Believe it or not, it is STILL very important to utilize the proper exposure factors.
This brings us to an important role of the software, histogram equalization. Equalization is performed by the computer in attempt to produce a more uniform histogram to increase the level of contrast in your image. After the original histogram is derived, an "inverse" histogram can be calculated and a spreading (or commonly termed flattening) of histogram values can be applied. Compare this image before histogram equalization and after:
This is a very basic explanation of a histogram and how it is utilized in a CR system. It would be easy to go onto many tangents from here (which I have a bad habit of doing in-person). A number of processing and post-processing errors can occur that I hope to dive into in the near future, but a basic understanding of histogram analysis is required. In the meantime, I would like to encourage everyone reading this to familiarize yourselves with the histograms utilized in your own imaging departments, and maniplulate them if you have the capability on your CR system and/or PACS terminals. There's not a lot of information about this in current textbooks, but I can imagine that we will all be responsible for knowing more about these things as we continue to replace conventional automatic processor technology with digital equipment.
