Let's review how window, level, and filtering all contribute to computed tomography (CT) image display. Manipulating these options when viewing a brain CT image can help you minimize diagnostic imaging errors.
While there was a time when CT images were printed on large sheets of film, CT images are now stored electronically and viewed on a computer screen using a Picture Archive and Communication System (PACS). PACS not only makes the images widely available, but it also allows you to manipulate their appearance.
By using varying window, level, and filtering options, the same CT data can be displayed in different ways. Certain settings show bony structures well, while the brain is better visualized using other settings.
Keep in mind that the terms window and level refer only to the way we display the CT data. They should not be confused with the parameters used to acquire CT data such as x-ray tube current, peak voltage, and detector slice thickness.
Why adjust window and level on a CT scan image?
It is important to understand the role of window and level settings in CT diagnosis, since they can either reveal or hide important findings on CT. For example, when the same scan is viewed with two sets of window and level settings, it’s possible that only one will reveal an abnormality of the skull!
How CT scan data is displayed
Computed tomography scan data is collected using an x-ray source on one side of the patient, and multiple x-ray detectors on the other side. The CT scan image uses shades of gray or white to indicate the magnitude of the x-ray beam’s interaction with the scanned body parts.
For example, since the bony skull blocks (i.e., attenuates) the passage of x-rays considerably more than the brain, its high attenuation is represented on the CT image with a shade of white. But, the air-filled sinuses allow the x-ray beam to pass through unimpeded, so it provides little attenuation of the x-ray beam and is represented as a shade of gray on the image.
How x-ray attenuation is reported on a CT scan
The magnitude of the x-ray attenuation during the CT scan can be calculated and reported using Hounsfield units (HU). These attenuation values on a patient scan can range from -1000 to +1000 HU, reflecting the wide differences in x-ray attenuation in the skull and brain.
If you then try to display this data on one black and white image, it would require hundreds—if not thousands—of shades of gray! Unfortunately, our human eye-brain interface is limited in perception and can only distinguish about 20 shades of gray between black and white. This represents a substantial limitation to displaying all of the available CT information.
The solution? Display only a portion of the data at one time.
Defining window and level settings for CT images
The terms window and level when discussing CT imaging indicate how much of that information we choose to display on the image at one time. Window indicates the range of HU values you choose to display, and level represents the midpoint of that range. You might think it would be possible to show the most information at one time by using a very wide window, but that means tissues of dissimilar attenuation values will be represented with the same shade of gray.
How to choose the right window and level settings for a CT scan
In order to choose the best window and level settings to view CT images, consider first what it is that you are looking for. If you are looking for an acute brain infarction, it is best to use a very narrow window since we know that there is only a small difference in attenuation between the normal brain and an acute infarct. While it provides the best possible contrast for detection of an infarct, all the tissues with attenuation values either above and below the selected window to be lumped together into all gray or all white.
To put this another way, imagine you are a wine reviewer and know from experience that you can only perceive differences in five different wines a day. You are up against a deadline for a story about wines from a local vineyard that offers 30 wines that range in price from $20 to $500 a bottle. You decide to select five wines in a narrow but popular price range.
You choose a range of $30 to $50 bottles, resulting in only five wines to review. All wines below $30 will be lumped together as inexpensive and all those above $50 will be labeled as more expensive.
Now, let’s say that your editor tells you that your approach won’t do since the magazine subscribers are a very affluent crowd. She asks you to review all the wines from $90 to $500, but you have only one day left before the deadline and can only review five wines.
Since there are ten wines in that range, you decide to pour together the $90 and $100 bottles, the $150 and $200 bottles, and so on, until you have only five mixed wines to taste. Your window is much larger, and level is higher, but you can see that this approach of mixing wines is of limited value. You can no longer perceive differences between the $90 and $100 bottles, for example, because you poured them together.
This is conceptually how we display CT images. Whenever you choose a wide window for a CT image, there will be more mixing of attenuation values into the same shade of gray. As a result, tissues of similar attenuation values may be indistinguishable on a CT image.
Now, let’s leave the vineyard and go back to the clinic.
A brain CT scan of a 55-year-old woman with new neurological symptoms revealed no abnormal findings when viewed with standard CT soft tissue window and level settings. However, if you were to review the same CT scan with a more narrow window of only 40 HU, an area of abnormally low attenuation in the frontal lobe becomes more conspicuous because there is less blending of similar attenuation values.
Notably, the brain abnormality was even more evident on a magnetic resonance imaging (MRI) scan obtained for this patient on the same day. A brain biopsy revealed that this was due to a primary brain tumor.
Why use filtering when reading a CT scan?
You should also be familiar with how filtering CT data alters the images. The scanner collects an enormous volume of data during each scan. That data is then processed to form an image using a reconstruction algorithm such as back projection or mathematical modeling.
Filtering occurs after the data collection but before image reconstruction. Different filters, which are also called kernels, are selected to optimize imaging of either soft tissue or bone.
Once the data is filtered, no amount of adjustment of window and level will allow the images to look the same. So even if the soft tissue filtered image is windowed appropriately for bone, it will not have the same diagnostic quality as one that used data filtered for viewing bone.
One immediate advantage to using bone filtering to review the head CT data is that it makes fractures more obvious. The improved detail may allow us to see even subtle fractures like the one below (Fig. 11).
Changing the window and level settings will increase the detection of some abnormalities—but can obscure others. For example, when reviewing this patient’s scan with a standard CT window setting of 80 HU and a level of 25 HU, the right subdural hematoma blends in with the skull since both are assigned the same shade of white. At this setting, it looks as though the patient has a thickened skull on the right side!
When a wider window and higher level are used to display the same scan, you can now see the subdural hemorrhage. This illustrates why you need to review the CT scan using more than one window and level setting.
When you consider the benefit of reviewing images at several windows, as well as multiplanar reconstructions, you should accept that you will be reviewing a lot of images with each head CT scan! No one said this was going to be easy, but being conscientious and consistent in your review will maximize your ability to detect abnormalities.
That’s it for now. If you want to improve your understanding of key concepts in medicine, and improve your clinical skills, make sure to register for a free trial account, which will give you access to free videos and downloads. We’ll help you make the right decisions for yourself and your patients.
- Guinto, FC Jr, Garrabrant, EC, and Radcliffe, WB. 1972. Radiology of the persistent stapedial artery. Radiology. 105: 365–369. PMID: 5079662
- Kim, YI, Ahn, KJ, Chung, YA, et al. 2009. A new reference line for the brain CT: the tuberculum sellae-occipital protuberance line is parallel to the anterior/posterior commissure line. AJNR Am J Neuroradiol. 30: 1704–1708. PMID: 19762457