9.5 Lumbar Spine Problems: Poor Bone Edge Detection and Obscured Intervertebral Spaces

9.5 Lumbar Spine Problems: Poor Bone Edge Detection and Obscured Intervertebral Spaces

Quick Answer: Two failure modes account for most unusable lumbar spine studies. Poor bone edge detection means the software cannot find where bone stops, so BMC and area are wrong. Obscured intervertebral spaces mean it cannot tell where one vertebra ends and the next begins, so the level assignments are wrong. Both fail silently — the software returns a number either way. The technologist's job is to look.

Failure Mode One: Poor Bone Edge Detection

The Mechanism

Edge detection thresholds the bone map: pixels above a mineral threshold are bone, below it are soft tissue. This works when the attenuation difference between bone and surrounding tissue is large. When that difference shrinks, the threshold lands in the wrong place.

Causes

The Two Directions of Error

This is the part that traps candidates. Under-detection and over-detection do not both lower BMD.

• Under-detection (algorithm misses a rim of low-density cortex): BMC falls and area falls, but area falls proportionally more, so BMD reads artifactually high.
• Over-detection (algorithm captures adjacent dense structures such as an osteophyte or calcification): BMC rises more than area rises, so BMD also reads artifactually high.

Both common errors bias upward, which is one reason a severely osteoporotic-looking spine returning a comfortable T-score deserves scrutiny rather than relief.

Recognition and Correction

1. Display the bone map over the image and compare it with the visible anatomy. Does the outline follow the vertebral margins? Are transverse processes appropriately included or excluded per the manufacturer's convention? Are there islands of "bone" in soft tissue?
2. Check for motion — look for a step or discontinuity running horizontally across the image.
3. Repeat the acquisition when motion or a wrong mode caused the problem. Editing cannot rescue a motion-corrupted acquisition.
4. Use the low-density analysis option where the manufacturer provides one, for genuinely osteoporotic bone.
5. Manually edit the bone map where the manufacturer's tools allow, restoring the true margin.
6. Document every edit. An undocumented manual edit is a precision error waiting to happen at the next visit.

A study that cannot be made reliable should be reported as such. "Technically limited, low bone density with suboptimal edge detection; hip preferred for monitoring" is useful information. A silently bad number is not.

Failure Mode Two: Obscured Intervertebral Spaces

The Mechanism

The software separates vertebrae by placing intervertebral markers in the disc spaces. It locates them by finding low-attenuation bands between high-attenuation vertebral bodies. When those bands are not visible, markers are misplaced, and mineral belonging to one vertebra is assigned to its neighbor.

Causes

The Correction Hierarchy

The order matters, because the right answer is usually "reposition," not "edit."

1. Reposition first. If lordosis is the cause, re-flexing the hips and knees over the leg block will open the disc spaces. Acquiring a properly positioned scan is always preferable to editing a poorly positioned one.
2. Verify levels using T12 ribs and the iliac crest before adjusting markers.
3. Adjust intervertebral markers manually when anatomy — not positioning — is the cause, placing them in the anatomically correct plane even where the lucency is faint.
4. Exclude the affected level when the boundary genuinely cannot be determined, following the exclusion rules, and document why.
5. Reproduce at follow-up. Manually placed markers must be placed the same way next time. Most systems allow the prior analysis to be displayed as a reference during follow-up analysis — for routine clinical follow-up this is appropriate and is how reproducibility is achieved. It is specifically not permitted during a precision study, where each scan must be analyzed independently.

What Not to Do

• Do not accept an automatic analysis whose markers sit visibly inside vertebral bodies.
• Do not "fix" residual lordosis by dragging markers when repositioning and rescanning is available.
• Do not exclude a level merely because it is inconvenient; exclusion follows defined criteria.
• Do not leave the fact of manual marker placement out of the record.

A Practical Review Checklist

Before accepting any lumbar spine analysis:

• Are the correct levels labeled, verified against T12 ribs and the iliac crest?
• Do the intervertebral markers sit in disc spaces, not in vertebral bodies?
• Does the bone map follow the actual bone margins?
• Are transverse processes handled per the manufacturer's convention?
• Are there islands of "bone" outside the vertebrae?
• Is there a motion step anywhere in the image?
• Do the per-level BMD values progress plausibly, generally rising from L1 to L4?
• Does any single level differ sharply from its neighbors in a way that suggests focal pathology?
• Has every manual intervention been documented?

That last question about per-level progression is a fast, powerful screen. Lumbar BMD normally increases from L1 through L4 because vertebral size increases. A level that breaks the pattern — much higher than its neighbors — usually contains something dense that does not belong to it.