How Bone Densitometry / DEXA Works

How a bone densitometer uses two X-ray energy ranges to estimate bone mineral content, separate bone from soft tissue, and calculate bone mineral density from a measurement that looks much simpler than it really is

A bone densitometer does not look nearly as intimidating as a CT scanner, angiography system, or MRI.

Back to Biomed Basics

What This Page Explains

This page covers:

The Simple Version

A DEXA scanner sends X-rays through the patient and measures how much radiation reaches the detector. The difficulty is that both bone and soft tissue attenuate the beam. With only one X-ray energy, the detector sees the combined attenuation and cannot easily determine how much came from mineralized bone versus surrounding tissue.

DEXA solves this by acquiring information at two different effective X-ray energies. Bone and soft tissue change their attenuation differently as photon energy changes. By comparing the measurements from the two energy ranges, the system mathematically separates much of the soft-tissue contribution and estimates the amount of mineral present in the bone.

The scanner then calculates bone mineral content and divides it by the projected bone area to produce an areal bone mineral density value, commonly expressed in g/cm². That number can then be compared with reference populations to generate T-scores or Z-scores. Because this is a quantitative measurement, calibration, patient positioning, analysis regions, detector stability, and repeatability matter just as much as whether the image looks acceptable. Preserve phantom trends, measured BMD, scan mode, positioning, analysis region, and whether a repeat scan reproduces the result before assuming the X-ray source or detector has failed.

Start With X-Ray Attenuation

As X-rays pass through the patient:

Some photons are absorbed or scattered.

Different Materials Attenuate Differently

Bone attenuates strongly.

Soft tissue attenuates less.

This Seems Easy

Why not simply measure how much radiation the bone blocks?

Because Soft Tissue Is in the Way

A lumbar spine scan includes:

Detector Measures the Combined Effect

It does not automatically know which attenuation came from bone.

One Measurement, Multiple Materials

That is the problem.

Dual Energy Adds Another Piece of Information

Instead of measuring attenuation at one X-ray energy:

Measure it at two.

Why That Helps

The relationship between attenuation and photon energy differs between:

Bone Contains Higher-Atomic-Number Elements

Especially:

Energy Dependence

Their attenuation changes differently with X-ray energy than the surrounding soft tissue.

Two Equations Help Solve Two Components

Conceptually:

Low-energy measurement

and:

High-energy measurement

provide enough information to estimate:

This Is the Core of DEXA

How Are Two Energies Produced?

The exact technique varies by manufacturer and system generation.

Possible Methods

Systems may use:

kVp Switching

The X-ray generator alternates between:

Lower kVp

and:

Higher kVp.

Detector Measures Both Conditions

The scanner compares the attenuation.

Filter-Based Separation

Some systems can shape the X-ray spectrum so two useful energy ranges are available.

Exact Hardware Varies

But the goal remains:

Acquire attenuation information at two distinct energy conditions.

This Is Not the Same as Two Separate Full X-Ray Exams

The system is designed to perform the dual-energy measurement efficiently with relatively low radiation output.

The X-Ray Source

Like other X-ray equipment:

The tube uses:

to produce radiation.

But DEXA Output Is Much Lower Than Many Diagnostic X-Ray Systems

Low Dose Does Not Mean Calibration Is Unimportant

In fact:

Quantitative measurements can be very sensitive to small systematic changes.

Detector

The detector measures transmitted X-ray intensity.

Detector Types Vary

Depending on scanner design.

Older Systems

May use:

Newer Systems

May use:

Pencil Beam

A very narrow X-ray beam scans across the patient.

Detector Moves or Aligns With Source

Data is built point by point.

Advantage

Low scatter.

Limitation

Longer acquisition.

Fan Beam

A wider beam covers more area.

Multiple detector elements collect data simultaneously.

Advantage

Faster scan.

But Geometry Becomes More important.

Fan-Beam Magnification

If anatomy is farther from the detector:

Projected size changes.

This Is Important Because DEXA Calculates Area

Remember the Formula

Conceptually:

BMD = Bone Mineral Content / Projected Bone Area

If Projected Area Changes

The reported BMD can change.

Positioning Matters Quantitatively

Not just cosmetically.

Bone Mineral Content

The scanner estimates the amount of mineral within the analyzed bone region.

This is:

BMC.

Usually Expressed in grams

Bone Mineral Density

The scanner divides BMC by projected area.

Common Unit

g/cm².

This Is Areal BMD

Important distinction.

DEXA Does Not Directly Measure True Volumetric Density

It produces an area-based measurement from a two-dimensional projection.

Why That Matters

Bone size can influence the measurement.

CT-Based Bone Measurements

Can estimate volumetric density differently.

But standard DEXA BMD is typically:

Areal.

T-Score

The T-score compares the patient's BMD with the average BMD of a young healthy reference population.

It Expresses Difference in Standard Deviations

Z-Score

The Z-score compares with an age- and population-matched reference group.

Scanner Measures BMD

Reference database creates the score.

This Is Important for Biomeds

If BMD is correct but the wrong demographic/reference data is used:

The score can still be wrong.

Measurement vs Interpretation Layer

Keep them separate.

Region of Interest

The scanner does not simply average the entire image.

Analysis software identifies specific anatomical regions.

Common Sites

Spine Scan

Software identifies:

Individual vertebrae.

BMD May Be Reported for

L1-L4

or other approved combinations depending on clinical analysis.

If Vertebral Labeling Is Wrong

The calculation can include the wrong anatomy.

Hardware Can Be Fine

Hip Scan

The software may define regions such as:

Positioning Matters Enormously

Internal or external rotation changes the projected femoral geometry.

Repeat Exam

If the hip is positioned differently:

BMD result may change even though bone itself did not.

This Is Why Patient Positioning Devices Exist

To reproduce geometry.

Serial Measurements

DEXA is often used to track changes over:

Months or years.

Small Change May Matter

That means repeatability must be excellent.

Precision

Precision describes how closely repeated measurements agree.

A Precise Scanner

Produces nearly the same BMD for unchanged conditions.

Accuracy

Accuracy describes how close the measurement is to the true or accepted value.

Precise but Wrong

Possible.

Example

Every scan reads:

5% high.

Very repeatable.

Not accurate.

Phantom Calibration

DEXA systems use calibration phantoms with known characteristics.

Phantom Provides a Stable Reference

Daily or Routine QC

Depending on system and program:

A phantom may be scanned regularly.

The System Tracks

Trend Matters

A single point slightly different may not mean failure.

Progressive Drift

Is more significant.

Control Chart

Some systems plot phantom results over time.

This Is Extremely Useful

A gradual drift can be seen before clinical studies are obviously affected.

Phantom BMD Shift

Could indicate changes in:

Calibration Does Not Mean “Make the Number Look Right”

The procedure establishes or verifies known system response.

Repeated Recalibration Can Hide Hardware Drift

Same lesson as other modalities.

Ask Why the Calibration Moved

X-Ray Tube Aging

Tube output can change over time.

Dual-Energy Relationship Matters

If the low- and high-energy spectra change differently:

Quantitative separation could be affected.

Detector Gain

If detector response changes:

Measured attenuation changes.

The System May Use Reference Detectors

Some designs monitor source output separately.

Why?

To distinguish source variation from patient attenuation.

Reference Channel

Can help normalize X-ray intensity.

I0

In attenuation measurements, the system needs some idea of:

Incident intensity before the patient.

Detector After Patient Measures

Transmitted intensity.

Relationship Produces Attenuation

If Source Output Changes

Normalization helps prevent that from being interpreted as patient change.

But Compensation Has Limits

Detector Nonuniformity

In multi-element systems:

Different detector channels can have different sensitivity.

Calibration Corrects Stable Differences

Bad Channel

Can produce:

Image May Look Mildly Abnormal

But ROI measurement could still be affected.

Scatter

Scatter adds unwanted radiation to the detector.

Pencil Beam

Naturally rejects much scatter because of narrow geometry.

Fan Beam

Requires more scatter management.

Scatter Correction

Software and detector geometry help compensate.

Patient Thickness

Large patient:

More attenuation.

More scatter.

Scan Mode May Change

System may use different acquisition technique depending on patient size.

This Can Be Normal

Scan Speed

Some systems offer:

modes.

Different Modes Can Use Different

Do Not Compare QC Across Modes Without Matching Conditions

Calibration Block

Some systems include an internal calibration reference that enters the beam during each scan.

Purpose

Provide known attenuation values.

This Helps Correct

Mechanical Position of Calibration Device Matters

If it fails to enter correctly:

Quantitative errors can result.

This Can Look Like Detector Failure

Scanner Arm Movement

DEXA scanners often move the source/detector assembly across the patient.

Position Encoder

The system needs to know:

Where each measurement was taken.

Motion Smoothness

Important for correct spatial sampling.

Encoder Failure

Could produce:

Image Can Reveal Motion Irregularity

Table

The patient table is part of the geometry.

Table Attenuation Is Known

System design accounts for it.

Foreign Material

Placed under patient can alter attenuation.

Positioning aids must be appropriate.

Clothing and External Objects

Metal can create artifacts.

Zippers, Coins, Buckles

Can interfere with analysis.

Artifact May Be Obvious

Surgical Hardware

Spinal instrumentation or hip replacement can affect measurements.

The analysis may need to exclude affected anatomy clinically.

Biomed Perspective

Do not interpret the patient study.

But understand why a technically good scanner can produce unusual values around metal.

Degenerative Change

Clinical factors can influence measured BMD.

Example

Calcification or degenerative changes may increase apparent spine BMD.

That Is Not Scanner Error

Again

Patient biology affects quantitative values.

Comparison Phantom Removes That Variable

Body Composition

Many DEXA systems can also estimate:

Same Dual-Energy Principle

Soft tissues have different attenuation characteristics.

Mathematical Modeling

Separates tissue components.

Body Composition Is Also Sensitive to

Whole-Body Scan

Uses larger field.

Detector uniformity becomes very important.

Edge of Field

If detector calibration differs across scan width:

Body composition or BMD could vary spatially.

Collimation

The X-ray beam is tightly controlled.

Field Size

Matches the scanning geometry.

Mechanical Alignment

Source and detector need correct relationship.

Misalignment

Can affect:

Fan-Beam Systems

Particularly depend on precise geometry.

Laser or Positioning Light

May help center the patient.

If Alignment Light Is Wrong

Patient positioning becomes less reproducible.

But X-Ray hardware may still be fine.

Scan Region

Operator defines:

Where to scan.

Wrong Scan Range

Can omit anatomy or change analysis.

Repeatability Requires Similar Scan Region

Software Analysis

After acquisition:

The system classifies:

and defines ROIs.

Edge Detection

Software identifies bone boundaries.

Poor Image Quality

Can cause incorrect segmentation.

Manual Adjustment

Technologists may correct analysis boundaries.

Operator Analysis Can Therefore Change Reported BMD

Hardware Service Needs to Distinguish

Acquisition problem

from:

Analysis problem.

Raw Image Good, Analysis Wrong

Software/ROI becomes more interesting.

Raw Image Artifact

Hardware/acquisition becomes more interesting.

Reference Database

T-scores and Z-scores depend on:

Software Upgrade

Could change:

BMD May Stay Similar

Scores may shift slightly.

Document Version Changes

Cross-Calibration

Suppose a facility replaces one DEXA scanner with another.

Same Patient

May not receive numerically identical BMD.

Why?

Different:

Cross-Calibration

May be required for reliable longitudinal comparison.

This Matters Clinically

Biomed Role

Understand that:

"Both scanners pass QC"

does not guarantee:

Their patient BMD values are directly interchangeable.

Least Significant Change

Clinical programs may calculate a threshold called:

Least Significant Change, or LSC.

Purpose

Determine whether change in serial BMD exceeds expected measurement variability.

This Depends on Precision

Not a Hardware Specification Alone

It includes facility measurement performance.

Still Useful Concept

A tiny numerical difference may be measurement noise rather than real biological change.

DEXA Is a Quantitative Modality

That is the core theme.

A Small Calibration Drift Matters

Even if the image looks completely acceptable.

Detector Failure Example

Suppose one detector channel becomes unstable.

Image

May show subtle streak.

Phantom

BMD measurement also changes.

Now the channel problem has quantitative consequence.

X-Ray Output Failure Example

Low-energy output drifts relative to high-energy output.

Image Still Forms

But material decomposition changes.

Phantom BMD moves outside historical range.

This is more important than simple brightness.

Positioning Example

Same patient scanned twice.

First Scan

Hip internally rotated according to protocol.

Second Scan

Femur less rotated.

Projected geometry changes.

BMD difference appears.

Hardware is unchanged.

Phantom is stable.

Strong evidence against scanner drift.

Daily QC

Phantom scan provides baseline.

If Daily QC Fails

Do not immediately scan patients and assume:

Probably fine.

Follow facility procedure.

Check Setup First

Reproduce

If appropriate.

Persistent Failure

Requires investigation.

Environmental Conditions

Detector electronics and mechanical systems operate within specified:

ranges.

Large Changes

Can affect performance.

Again

Facilities issues can become measurement issues.

Power Stability

The generator still depends on reliable electrical power.

But DEXA is generally lower power than large radiographic systems

Internal Power Supply Fault

Could affect:

Symptom Pattern Matters

Scan Aborts at Same Position

Think:

Scan Completes but Values Drift

Think:

Image Looks Bad Everywhere

More general acquisition problem.

Only ROI Analysis Wrong

Software/segmentation.

Real-World Example: Phantom Drift

Daily phantom BMD has slowly increased over several weeks.

Technologists have not reported image problems.

The system is repeatable but moving away from baseline.

This is exactly the kind of issue quantitative QC is supposed to detect before clinical confidence is affected.

Real-World Example: Patient Change, Phantom Stable

One patient's spine BMD changes unexpectedly between visits.

Daily QC is stable.

Images show different positioning and degenerative change.

There is little evidence of scanner calibration failure.

Real-World Example: One Scan Mode Fails

Standard scan mode passes phantom QC.

High-resolution mode produces a repeatable streak.

Detector readout configuration or mode-specific acquisition becomes more interesting than the entire scanner.

Real-World Example: BMD Wrong After Service

X-ray tube or detector service completed.

Images look fine.

Phantom BMD is shifted from historical baseline.

Required post-service calibration was not completed.

The system is imaging but not yet quantitatively validated.

Real-World Example: Position-Dependent Artifact

Scan repeatedly develops a streak at the same carriage location.

Artifact occurs with different phantoms.

Motion or cable behavior linked to scanner-arm position becomes suspicious.

Common Mistakes

Thinking DEXA directly measures the physical density of bone in g/cm³. Standard DXA usually reports areal BMD in g/cm².

Judging calibration from image appearance. A quantitative drift may exist with a perfectly recognizable image.

Ignoring patient positioning in serial studies. Geometry can change the measured projected area.

Assuming T-score is a direct detector measurement. The scanner measures BMD; the score comes from comparison with reference data.

Recalibrating every phantom shift without investigating why it occurred. Stable baseline changes deserve explanation.

Comparing two scanners as if their numbers must be identical. Cross-calibration may be necessary.

Treating all unusual patient results as equipment faults. Clinical anatomy, metal, degenerative change, and analysis choices can affect measurements.

A Useful DEXA Framework

Think:

Dual-Energy X-Ray Source

↓

Patient: Bone + Soft Tissue

↓

Low-Energy Attenuation

+

High-Energy Attenuation

↓

Detector

↓

Calibration / Reference Correction

↓

Material Separation

↓

Bone Mineral Content

↓

Projected Bone Area

↓

BMD

↓

Reference Database

↓

T-Score / Z-Score

That is the measurement chain.

Another Useful Troubleshooting Split

Ask:

Is the image wrong, the BMD number wrong, or only the final score/analysis wrong?

Then:

Does the phantom reproduce the change?

Then:

Is the problem stable across scan modes and positions?

Those questions separate:

far better than looking only at the patient report.

What Did You Actually Prove?

If the DEXA image looks normal:

You proved:

The scanner produced a recognizable acquisition under that condition.

You did not prove:

If the phantom BMD matches expected baseline:

You have evidence that:

The system's quantitative measurement chain is stable under that controlled phantom condition.

You did not prove:

A patient's positioning or analysis was correct.

If repeated phantom QC, calibration checks, positioning verification, and system-specific performance tests all pass:

You have much stronger evidence that the densitometer is functioning correctly.

Final Thoughts for Biomeds

DEXA looks simple because the scanner is small and the images are not dramatic.

But the measurement problem is sophisticated.

The detector sees:

Bone plus soft tissue.

The system uses two X-ray energies to separate their effects.

It estimates how much mineral is present.

It measures the projected bone area.

Then it calculates BMD.

After that, software compares the result with reference populations and produces scores used clinically.

That means there are several opportunities for a number to become wrong.

The X-ray spectra can drift.

The detector can drift.

Calibration can shift.

The patient can be positioned differently.

The software can analyze a different region.

The reference data can change.

The useful question is not:

Did it make a spine picture?

It is:

Did the entire quantitative measurement chain reproduce the known result it was supposed to reproduce?

That is why the phantom matters.

It gives you something the patient cannot:

A stable known reference.

And that is exactly what you need when the problem is not merely:

Can the scanner see bone?

but:

Can the scanner measure it accurately?

And, as always:

What did you actually prove?

— Jake

Important Note

DEXA/DXA X-ray generation methods, energy-separation techniques, detector types, beam geometry, calibration phantoms, BMD algorithms, body-composition analysis, reference databases, precision requirements, cross-calibration methods, QC procedures, and acceptable limits vary significantly by manufacturer, scanner model, software version, and clinical program. DEXA uses ionizing radiation. Follow current OEM documentation, approved bone-densitometry QC procedures, facility radiation-safety requirements, qualified medical-physics guidance, and authorized service scope when evaluating quantitative performance.

Related Biomed Basics

Back to Biomed Basics