Theoretically, if the algorithms are correct, the image should have optimal contrast and brightness.However, even if the algorithms are in-correct, certain post-processing options may still be applied for specific image effects.After digital images have been captured by a digital image receptor, default algorithms (using a look up table) are used to process the image to display the radiographed anatomy in best possible way. One of the advantages of digital imaging technology is the ability to post-process the image.
Post-processing refers to the changing or enhancing the digital image in order to improve its diagnostic quality after the image as been acquired (post acquisition). During post-processing, various algorithms are applied to the original digital image to modify pixel values.Once viewed, the changes made can be saved or the image can be returned to its default setting.
Theoretically, if the algorithms are correct, the image should have optimal contrast and brightness.However, even if the algorithms are in-correct, certain post-processing options may still be applied for specific image effects.
An important concept to keep in mind is that many of the post proccessing algorithms are done automatically based upon pre-set look up tables and software settings. This makes the integrity of the raw data very important. The radiographer is responsible for insuring that the raw image is at its best. This can be done by ensuring a quality SNR (checking the exposure index of the image), providing a quality histogram (collimation, removal of metal, proper positioning, etc.), and the reduction of artifacts such as motion
There are a many types of post-processing activities available in medical imaging, here are a few.
Window & Level is the most common post processing control used by the technologist and radiologist. Essentially, these are two different adjustments that are derived from the human eye's ability to distinguish between two gray's. The human eye cannot perceive the same number of grays' that a computer image is able to produce. Window controls how light or dark the image is (density and brightness). Level controls the ratio of black to white (contrast).
Frequency uses un-sharp masking, or smoothing processing to enhance the details of a image. This happens by comparing the processed image to the raw image and if the details are outside a pre-set threshold, the pixels are subtracted as seen in the image to the right.
This is the adjustment of how much or little gray the image displays. Unlike the reliance on the sensitometric curve in film/screen, the gray scale response is governed by the user. Gray scale determines the overall appearance of the final image and is very similar to controlling the contrast.
Equalization is weighted processing to improve the display range and enhance local contrast. This is accomplished by altering local (small areas on the image) densities and brightness levels to even the image.
Once the signal is obtained for each pixel, they are averaged to speed up the processing time. The signal strength of one pixel is averaged with the signal strength of the nearby pixels. The more pixels averaged together, the smoother the image appears. Edge enhancement takes place when fewer pixels in the nearby area are included in the signal average. When the signal strength (frequency) of an area of interest is known, that signal can be amplified and other signals blocked out. This process is known as high-pass filtering and results in increases contrast and edge enhancement. High-pass filtering is useful for increasing the visibility of the edges of large structures like organs and soft tissues, however it can be noisy.
Another type of filtering gives the image a smooth appearance. This low-pass filtering smoothes images by averaging each pixel's frequency with surrounding pixel values to remove high-frequency noise. This filtering process reduces noise and contrast. Low-pass filtering is useful for viewing small structures such as fine bone tissues.
Noise supression reduces visibility of noise in a image by using smoothing algorithms to "blur" out defined noise. This can degrade edge clarity in a image, making proper exposure important.
Image orientation refers to the way anatomy is oriented on the image plate. Vendors mark the cassettes in different ways to indicate the top and sides of the imaging plate. In CR, the image reader scans and reads the image from the leading edge of the imaging plate to the opposite end. The image reader must be informed of the orientation of the anatomy with respect to the reader. Unless the reader is informed differently, the image is displayed exactly as it was read.
The largest size of CR imaging plate is 35x43 centimeters (14x17inches).Occasionally, an area of interest is too large to fit on one CR cassette, such as foot-to-hip or entire spine studies. Therefore, the entire area of interest must be imaged using several CR cassettes. Through a process called image stitching, multiple images can be "stitched" together using specialized software programs. Images are processed in computer programs that nearly seamlessly join the anatomy for display as one single image.
Many times it is necessary to include information to the image other than the standard identification. The image annotation function allows the radiographer to select preset terms and/or to manual input text. One disadvantage of image annotation is that they overlay the image as bitmap images and depending on how each system is set up, the annotations may not transfer to PACS. Image annotation to identify the patient's left or right side should never be used as a substitute for radiographer's anatomy markers.
Occasionally it is necessary to magnify an image to better visualize anatomy. With image magnification, all or part of an image can be magnified. When it is necessary to magnify only a small segment of anatomy, some systems allow a box to be placed over the anatomy of interest functioning as a magnifying glass to show a magnified version of the underlying anatomy. Both the size of the magnified area and the amount of magnification can be made larger or smaller. When it is necessary to magnify the entire image, a "zoom" function can be used. The image can be enlarged enough so that only parts of it are visible on the screen, however, the parts not visible can be seen by scrolling up/down and right/left with the mouse (depending on the software being used).
Gradient variations are similar to equalization only applied to the entire image with a specific pattern or gradient. Gradients are used for the gradual blending of color or gray scale from low to high values. Gradients are used to help even the intensity levels of various portions within the image. Each method of blending is a gradient type. A common gradient would be on a Femur where the density of the anatomy is much greater at the superior end of the leg.
Contrast enhancement increases the appearance of large scale light-dark transitions. Similar to sharpening but by removing scales of gray to create larger differences between grays.
Each time a image is "processed", it takes away or hides information. Altough processing is designed to enhance a image, processing can also hide pathologies. Consider a copy machine, if you keep making a copy of a copy, the image quality will eventially become to poor to recognize as the original. Because processing can only go so far, it is up to the technologist to make the best exposure possible in the first place.