Hounsfield Unit Scale, Linear Attenuation & Windowing (WW/WL)

---

Biophysical Principles: Linear Attenuation Coefficient ($\mu$) & Energy Dependence

Computed tomography measures the degree to which an x-ray beam is attenuated as it traverses matter. In the diagnostic energy range ($80\text{--}140\text{ kVp}$), x-ray attenuation in biological tissues is governed primarily by two physical interaction mechanisms:

1. Compton Scattering: The dominant interaction in soft tissues, where x-ray photons interact with loosely bound outer-shell electrons. The probability of Compton scattering depends almost entirely on the tissue's physical density ($\rho$) and electron density (electrons per unit volume), exhibiting minimal dependence on atomic number ($Z$).
2. Photoelectric Absorption: Occurs when an incident x-ray photon ejects an inner-shell electron, depositing its entire energy. Photoelectric absorption is proportional to the cube of the effective atomic number ($Z^3$) and inversely proportional to the cube of photon energy ($1 / E^3$). It plays a prominent role in dense cortical bone ($Z_{\text{eff}} \approx 13.8$) and in iodinated contrast media ($Z = 53$).

The Linear Attenuation Coefficient ($\mu$)

The linear attenuation coefficient ($\mu$, measured in $\text{cm}^{-1}$) characterizes the fraction of x-ray photons removed from a monochromatic beam per centimeter of absorber. However, diagnostic CT scanners utilize polychromatic x-ray spectra generated by bremsstrahlung and characteristic emissions. For a standard $120\text{ kVp}$ tube potential with routine filtration, the effective (mean) photon energy is dependent on the spectrum, filtration and measurement definition; it is not one fixed 70–75 keV value.

Because the numerical value of $\mu$ varies significantly with tube potential ($kVp$), beam filtration, and beam hardening along the patient path, reporting raw $\mu$ values in medical reports would be completely unstandardized. An identical kidney cyst might measure $\mu = 0.205\text{ cm}^{-1}$ at $100\text{ kVp}$ and $\mu = 0.185\text{ cm}^{-1}$ at $140\text{ kVp}$. To provide a standardized water-relative scale, while retaining spectrum and measurement dependence, Sir Godfrey Hounsfield established the CT Number / Hounsfield Unit scale.

---

The Hounsfield Unit (HU) Mathematical Formulation

The Hounsfield Unit is an internationally standardized, normalized linear transformation of the linear attenuation coefficient of a tissue relative to that of pure water:

$$HU = \frac{\mu_{\text{tissue}} - \mu_{\text{water}}}{\mu_{\text{water}}} \times K$$

where $K$ is an arbitrary scaling constant established as $1000$ (the Hounsfield multiplier).

Reference values and energy dependence

Water defines 0 HU and air is approximately −1000 HU in the conventional scale. Bone is not an immutable +1000 HU calibration anchor. Its value depends on mineral content, spectrum, reconstruction and measurement. From the formula HU = 1000(μmaterial − μwater)/μwater, twice the water attenuation gives +1000 HU and three times gives +2000 HU, under the stated effective attenuation comparison.

Sensitivity of the Scale

Because $K = 1000$, each individual Hounsfield Unit represents exactly a $0.1\%$ difference in linear attenuation coefficient relative to water. A tissue measuring $+50\text{ HU}$ attenuates the diagnostic x-ray beam by $5.0\%$ more than water, whereas a tissue measuring $-100\text{ HU}$ attenuates $10.0\%$ less than water.

---

Diagnostic Spectrum: Characteristic HU Values of Human Tissues

The diagnostic spectrum of computed tomography spans from $-1000\text{ HU}$ to $> +3000\text{ HU}$. The following are illustrative conventional-CT attenuation ranges, not diagnostic cutoffs. A tissue’s attenuation is influenced by its physical density, water content, lipid concentration, and effective atomic number.

Comprehensive Biological Tissue Attenuation Reference

Blood appearance is not a precise clock

Acute clotted blood commonly has greater attenuation than brain, while evolving blood can become isodense or hypodense. Hematocrit, dilution, clotting and time affect the appearance. Do not date a hematoma by a universal 1.5 HU/day decline. Review location, morphology, mass effect, prior images and clinical context, using appropriate brain and extra-axial windows.

Window Width (WW) and Window Level (WL): Perception & Display Physics

The fundamental rationale for windowing (gray-level mapping) stems from the physiological limitations of the human visual system.

Human Contrast Perception vs. CT Dynamic Range

Stored image bit depth and scaling determine the numerical range, while a display presents a selected grayscale mapping. Raw detector sampling depth is a separate property. Mapping a very broad attenuation range onto a limited display makes small tissue differences less conspicuous. There is no universal count of gray shades visible to every observer; monitor calibration, luminance, surroundings and the task affect perception. Brain gray-white differences require suitable contrast settings even when the underlying data contain them.

Windowing solves this limitation by selecting a narrow diagnostic slice of HUs and expanding it across the monitor's full black-to-white dynamic range.

Window Width (WW) — Image Contrast

Window Width (WW) defines the range (total span) of Hounsfield Units assigned across the full grayscale spectrum from pure black to pure white.

• WW controls Contrast: The narrower the window width, the fewer HUs distributed across the grayscale ramp, resulting in higher image contrast (a steeper grayscale gradient). Subtly differing tissues are driven toward opposite ends of the grayscale.
• Narrow WW (e.g., $80\text{ HU}$): High contrast. Used when target tissues exhibit almost identical attenuation, such as brain gray and white matter.
• Wide WW (e.g., $2000\text{ HU}$): Low contrast, wide latitude. Used when the anatomical region contains tissues with extreme attenuation differences (e.g., bone trabeculae adjacent to air or dense cortical bone, or lung parenchyma adjacent to thoracic wall soft tissue), preventing image saturation.

Window Level (WL / Window Center) — Image Brightness

Window Level (WL), or Window Center, defines the midpoint (center) of the Hounsfield range displayed on the monitor.

• WL controls Brightness: Increasing the WL shifts the display toward higher CT numbers, making the overall image darker. Decreasing the WL shifts the display toward lower CT numbers, making the overall image brighter.
• Golden Clinical Rule: A useful initial Window Level is often centered at or near the mean Hounsfield Unit of the primary organ or tissue of interest (e.g., $+40\text{ HU}$ for brain, $+50\text{ HU}$ for liver, $-600\text{ HU}$ for lung parenchyma, $+400\text{ HU}$ for bone).

Mathematical Formulas for Display Range Boundaries

The lower and upper thresholds of the displayed grayscale ramp are approximated in the continuous linear teaching model by the following equations. Exact DICOM linear windowing uses half-unit offsets and WW − 1; displays can also support other functions. The examples here consistently use the simplified model:

$$\text{Lower Display Limit (Pure Black)} = WL - \frac{WW}{2}$$ $$\text{Upper Display Limit (Pure White)} = WL + \frac{WW}{2}$$

• Any voxel with an attenuation value less than or equal to the Lower Limit is displayed as pure black ($0\%\text{ luminance}$).
• Any voxel with an attenuation value greater than or equal to the Upper Limit is displayed as pure white ($100\%\text{ luminance}$).
• Voxels falling between the Lower and Upper Limits are distributed linearly across intermediate shades of gray in proportion to their HU value.

---

Standard Clinical Window Settings

The following are illustrative starting presets for specialized Window Width and Window Level presets tailored to the diagnostic task:

---

Step-by-Step Worked Calculations & Practice Scenarios

Worked Example 1: Calculating Displayed Grayscale Range for Brain vs. Blood Windows

Clinical Scenario: A patient with acute trauma undergoes a non-contrast head CT. The technologist reviews the scan on routine brain settings ($WW = 80$, $WL = 40$) and then switches to a dedicated subdural/blood window ($WW = 200$, $WL = 75$). What are the displayed HU ranges, and how does an acute $+70\text{ HU}$ subdural clot appear on each?

• Step 1: Compute Routine Brain Display Range:

$$\text{Lower} = 40 - \frac{80}{2} = 40 - 40 = 0\text{ HU}$$ $$\text{Upper} = 40 + \frac{80}{2} = 40 + 40 = +80\text{ HU}$$

Range: $[0\text{ to }+80\text{ HU}]$. An acute clot at $+70\text{ HU}$ falls near the very top of this narrow range ($70/80 = 87.5\%$ luminance), appearing nearly bright white, easily blending into the adjacent $+1000\text{ HU}$ calvarium.

• Step 2: Compute Subdural / Blood Display Range:

$$\text{Lower} = 75 - \frac{200}{2} = 75 - 100 = -25\text{ HU}$$ $$\text{Upper} = 75 + \frac{200}{2} = 75 + 100 = +175\text{ HU}$$

Range: $[-25\text{ to }+175\text{ HU}]$.

• Interpretation: On the blood window, the upper limit is $+175\text{ HU}$. The dense skull bone ($> +1000\text{ HU}$) remains saturated at pure white, while the acute clot ($+70\text{ HU}$) sits comfortably at the midpoint of the grayscale ($[70 - (-25)] / 200 = 95/200 = 47.5\%$ gray), clearly separated from both brain tissue and bone.

Worked Example 2: Determining Linear Attenuation Coefficient from Hounsfield Units

Clinical Scenario: In an explicitly simplified attenuation example, a tissue has a measured value of $+65\text{ HU}$. If the reference linear attenuation coefficient of water at $120\text{ kVp}$ is known to be $\mu_{\text{water}} = 0.190\text{ cm}^{-1}$, calculate the exact linear attenuation coefficient of the tissue ($\mu_{\text{tissue}}$).

• Step 1: Rearrange the Hounsfield Formula for $\mu_{\text{tissue}}$:

$$HU = \frac{\mu_{\text{tissue}} - \mu_{\text{water}}}{\mu_{\text{water}}} \times 1000 \implies \frac{HU}{1000} = \frac{\mu_{\text{tissue}}}{\mu_{\text{water}}} - 1$$ $$\mu_{\text{tissue}} = \mu_{\text{water}} \times \left(1 + \frac{HU}{1000}\right)$$

• Step 2: Substitute Known Numerical Values:

$$\mu_{\text{tissue}} = 0.190\text{ cm}^{-1} \times \left(1 + \frac{65}{1000}\right) = 0.190 \times 1.065 \approx 0.20235\text{ cm}^{-1}$$

• Interpretation: The tissue exhibits an attenuation coefficient of $\approx 0.202\text{ cm}^{-1}$, which is exactly $6.5\%$ greater than that of pure water.

---