11.6 Scan Analysis: Region Placement, Window Level, and Reported Values
Key Takeaways
- Window level and window width adjust only the displayed brightness and contrast of a DXA image; they do not change bone mineral content, area, or BMD.
- Adjusting window level reveals artifacts, bone margins, and region boundaries that are invisible at default display settings, which is why it is an analysis tool at all three sites.
- Reported values at every site include BMC in grams, area in square centimeters, BMD in grams per square centimeter, T-score, Z-score, and percent of reference mean.
- Graphical displays plot the patient against reference curves and plot serial values over time, and they are interpretive aids rather than the basis for classification.
- Every manual region adjustment or bone-map edit must be documented so the identical analysis can be reproduced at follow-up.
11.6 Scan Analysis: Region Placement, Window Level, and Reported Values
Quick Answer: Analysis is the same discipline at spine, femur, and forearm: verify the bone map against the anatomy, verify region placement, adjust window level to inspect what the default display hides, sanity-check the reported values, and document every manual change. Window level changes only how the image is displayed — it never changes BMC, area, or BMD. The ARRT specifications effective January 1, 2027 add window level to scan analysis at all three sites.
Window Level and Window Width
What They Are
A DXA acquisition stores a range of attenuation values far wider than a monitor can display at once. Window width sets how much of that range is mapped to the available shades of grey; window level sets where the centre of that range sits.
- Narrow window width = high contrast. Small density differences become visible, but values outside the window saturate to pure black or pure white.
- Wide window width = low contrast. More of the range is visible at once, with less differentiation between similar densities.
- Raising window level generally darkens the image; lowering it generally brightens it.
The Critical Point
Windowing is a display operation. It changes nothing about the stored data, the bone map, the computed BMC, the projected area, or the BMD. A technologist who windows an image to look at an artifact has not altered the measurement in any way.
This is worth stating plainly because the opposite belief leads to two errors: refusing to window at all, and believing that windowing can "fix" a bad result.
Why It Is an Analysis Tool
The default display is optimized for a typical patient. Outside that typical range, information hides.
| Task | Windowing approach |
|---|---|
| Find a faint intervertebral disc space in a degenerative spine | Narrow the window to raise contrast at the bone-disc boundary |
| See the cortical margin in a severely osteoporotic bone | Brighten and narrow to bring low-density bone out of the dark end |
| Identify overlying aortic calcification | Adjust to separate vascular calcification from vertebral bone |
| Check whether the ischium underlies the femoral neck | Window to distinguish the two overlapping bone densities |
| Look for a non-removable artifact | Widen first to survey the whole field, then narrow onto the suspicious area |
| Examine soft tissue for asymmetry or a retained object | Window toward the soft tissue range |
| Verify a vertebral compression deformity | Adjust to make endplates and vertebral heights clearly visible |
| Check the distal reference and carpal boundary on a forearm | Narrow onto the distal end |
Practically: survey the whole field at a wide window, then narrow onto anything that looks wrong. If the image at default settings looks unremarkable but the number is implausible, windowing is the next step, not the last one.
What Windowing Cannot Do
- It cannot rescue a motion-corrupted acquisition.
- It cannot correct a mispositioned patient.
- It cannot remove a non-removable artifact from the measurement.
- It cannot change a region-of-interest placement — that is a separate editing operation.
- It does not change anything reported to the clinician.
Windowing lets you see the problem. Fixing it requires repositioning, re-acquiring, editing the region, or excluding the site.
Region of Interest Placement
Region placement is the analysis step that does change the numbers.
| Site | Placement essentials |
|---|---|
| Lumbar spine | Intervertebral markers in the disc spaces; correct levels labeled from T12 ribs and the iliac crest; global ROI boundaries per manufacturer convention; excluded vertebrae excluded |
| Proximal femur | Neck box at the narrowest part of the neck, perpendicular to the neck axis, clear of ischium and greater trochanter; global ROI boundaries; correct side |
| Forearm | Distal reference correctly identified at the end of the radius and ulna; percentage regions located from the entered forearm length; regions clear of the carpus; radius and ulna resolved separately |
Three rules apply everywhere:
- Compare the placement against the anatomy, not against what looks tidy.
- Adjust manually when automatic placement is wrong, using the manufacturer's tools.
- Document the adjustment and reproduce it at follow-up, using the baseline analysis as reference for routine clinical comparison.
Reported Values
Every site reports the same core set:
| Value | Unit | Meaning |
|---|---|---|
| BMC | grams | Total mineral within the region |
| Area | cm² | Projected area of the region |
| BMD | g/cm² | BMC ÷ area; the measured quantity |
| T-score | SD | Deviation from the young-adult reference mean |
| Z-score | SD | Deviation from the age-matched reference mean |
| % young adult | % | BMD as a percentage of the young-adult mean |
| % age-matched | % | BMD as a percentage of the age-matched mean |
Plausibility Checks
- Spine: do per-level values rise gradually from L1 to L4? A level far above its neighbors contains something dense.
- Femur: is the femoral neck value lower than the total hip, as it usually is? Is Ward's area lowest, as it always is — and is it correctly excluded from classification?
- Forearm: is the 33% radius higher than the ultradistal, as cortical composition predicts?
- All sites: does the BMD equal BMC divided by area? Does the T-score correspond to the BMD given the displayed reference mean?
- Serial: is the change plausible for the interval and the therapy, and does it exceed the LSC?
Graphical Displays
Most reports include two graphs:
- A reference curve plot, showing the patient's BMD against age-related reference curves with T-score or percentile bands. It communicates position at a glance.
- A serial trend plot, showing BMD at successive visits.
Both are interpretive aids. Classification comes from the T-score at a valid site, and significance of serial change comes from comparing absolute BMD difference against the LSC — never from whether a trend line looks like it is going up.
Be alert to a serial plot that silently mixes scanners, sides, or analyses with different exclusions. The graph will happily draw a line through incomparable points.
Documentation
For every analysis, record: regions used and any excluded; manual bone-map or region edits and why; artifacts present and their location; acquisition mode and any low-density option; positioning deviations; and for the forearm, the side and the entered forearm length. Windowing itself does not require documentation, because it changes nothing — but what you found by windowing usually does.
A technologist adjusts the window level on a lumbar spine image to make faint intervertebral spaces visible. What effect does this have on the reported BMD?
A forearm report shows an ultradistal BMD higher than the 33% radius BMD. What should the technologist suspect?
Which analysis action actually changes the values reported to the clinician?