5.1 Physical Principles of Dual-Energy X-ray Absorptiometry

5.1 Physical Principles of Dual-Energy X-ray Absorptiometry

Core Clinical Principle: Dual-Energy X-ray Absorptiometry (DXA) overcomes the fundamental limitation of single-energy radiography—the inability to differentiate between thin dense bone and thick soft tissue—by transmitting two distinct photon energy spectra through the patient. By exploiting differential photoelectric and Compton attenuation, DXA establishes two simultaneous equations that mathematically eliminate soft tissue attenuation to isolate areal Bone Mineral Density ($aBMD = BMC / Area$).

---

1. Foundational Physics of Photon Attenuation

The measurement of bone mineral relies on the predictable attenuation of electromagnetic radiation traversing biological matter. When a collimated monoenergetic x-ray beam passes through a uniform absorber of thickness $x$, beam attenuation follows the Beer-Lambert Law:

$$I = I_0 e^{-\mu x}$$

Where $I_0$ is the incident beam intensity, $I$ is the transmitted intensity reaching the detector, $\mu$ is the linear attenuation coefficient ($cm^{-1}$) representing the fractional reduction of photons per unit distance, and $x$ is absorber thickness ($cm$).

Because biological tissues vary in physical density ($\rho$ in $g/cm^3$) without altering elemental composition, medical physics expresses attenuation as the mass attenuation coefficient ($\mu/\rho$ in $cm^2/g$). The mass attenuation coefficient is independent of physical density or packing state; it depends solely on the absorbing material effective atomic number ($Z_{eff}$) and incident photon energy ($E$).

Diagnostic Interaction Mechanisms: Photoelectric vs. Compton

In diagnostic densitometry energies (30–140 keV), two interactions dominate:

1. Photoelectric Absorption: The incident photon expends all energy ejecting an inner-shell electron. Interaction probability ($\tau$) is strongly dependent on atomic number and photon energy:

   $$\tau \propto \frac{Z^3}{E^3}$$

   Mineralized bone contains calcium hydroxyapatite ($Ca_{10}(PO_4)_6(OH)_2$) with calcium ($Z = 20$) and phosphorus ($Z = 15$), producing an effective atomic number ($Z_{eff} \approx 13.8$) far exceeding that of soft tissue ($Z_{eff} \approx 7.4$). At low energies (~40–45 keV), photoelectric absorption in bone is 6 to 7 times greater per gram than in adjacent soft tissue.

2. Compton Scattering: The photon collides with an outer-shell valence electron, ejecting a recoil electron and deflecting the photon with reduced energy. Compton probability depends almost exclusively on electron density (around $3.0 \times 10^{23}$ electrons per gram for non-hydrogen biological elements) and scales inversely with energy ($1/E$), exhibiting virtually zero dependence on atomic number ($Z$).

At higher energies (~70–100 keV), Compton scattering dominates in both bone and soft tissue, significantly narrowing the attenuation difference between them.

---

2. Two-Component Tissue Model and Mathematical Solution

DXA models the human body as a two-compartment system:

1. Bone mineral compartment (calcium hydroxyapatite).
2. Soft tissue compartment (adipose and lean soft tissue).

In adjacent non-bone baseline regions, soft tissue is further resolved into fat and lean mass. The scanner determines the baseline soft tissue attenuation ratio ($R_{st} = \mu_{L,st} / \mu_{H,st}$) and extrapolates this value across the bone projection area.

The Single-Energy Limitation

A single x-ray beam yields one equation with two unknowns ($x_b$ for bone and $x_s$ for soft tissue):

$$\ln(I_0 / I) = \mu_b x_b + \mu_s x_s$$

A single measurement cannot distinguish between a thin dense bone and a thick soft tissue layer, as infinite combinations produce identical total attenuation.

The Dual-Energy Solution

Transmitting a low-energy spectrum ($L$, ~40–50 keV) and a high-energy spectrum ($H$, ~70–100 keV) provides two independent equations in two unknown areal densities ($M_b = \rho_b x_b$ and $M_s = \rho_s x_s$ in $g/cm^2$):

$$\ln(I_{0,L} / I_L) = (\mu/\rho)_{b,L} \cdot M_b + (\mu/\rho)_{s,L} \cdot M_s$$ $$\ln(I_{0,H} / I_H) = (\mu/\rho)_{b,H} \cdot M_b + (\mu/\rho)_{s,H} \cdot M_s$$

Because the mass attenuation coefficients are calibrated physical constants, solving this linear system isolates bone mineral areal density ($M_b$) while mathematically subtracting soft tissue attenuation from every pixel.

Areal BMD and the Bone Size Artifact

From pixel areal density ($M_b$), the scanner calculates:

• Bone Mineral Content (BMC): Sum of bone mineral mass within the region of interest ($BMC = \sum [M_b \times \text{pixel area}]$ in grams, $g$).
• Projected Bone Area: Total bone pixel area ($cm^2$).
• Areal Bone Mineral Density ($aBMD$):

$$aBMD = \frac{BMC}{\text{Area}} \quad \left[\frac{g}{cm^2}\right]$$

DXA measures areal density ($g/cm^2$), not true volumetric density ($g/cm^3$). Because larger bones have a longer anterior-posterior path length, they contain more total mineral within their projected silhouette. Consequently, larger bones display an artificially elevated $aBMD$ despite identical true volumetric density. This bone size artifact causes systematic underestimation of bone strength in petite adults or pediatric patients and overestimation in large individuals.

---

3. Dual-Energy Generation Technologies

The ARRT content outline names the two methods of x-ray production for dual-energy separation as k-edge filtration and energy switching. They are the two rows of the table below, and each manufacturer uses one or the other.