Hounsfield Unit Measurements in Abdominal CT
Quantitative Sniffing for Diagnostic Certainty: Beyond visual assessment into the realm of material verification.
The Central Thesis
"Radiology advances not by seeing more, but by verifying better. HU measurement is the radiologist’s equivalent of a 'sniff test'—a deliberate act of verification before acceptance."
The Tissue Characterization Spectrum
Organ-Based Clinical Applications
The Fatty Liver Logic
Visual assessment of steatosis is notoriously unreliable due to display window settings. HU provides the objective standard.
- Steatosis: Liver < Spleen by >10 HU on non-contrast scans.
- Cysts: Uniform water attenuation (~0-15 HU).
In oncology, a decrease in HU without a change in size often indicates "liquefaction" or response to anti-angiogenic therapy (Choi Criteria).
The Physicist's Warning: Avoid these Pitfalls
Partial Volume
ROIs placed on edges of small lesions will mix with surrounding fat/water, creating false readings.
kVp Variability
80 kVp scans make everything "brighter" (higher HU) than 120 kVp due to the photoelectric effect.
Phase Confusion
Never use a portal venous phase HU to "rule out" a lipid-rich adrenal adenoma. Non-contrast is required.
Pseudo-enhancement
Small renal cysts can show a false 10-15 HU rise due to reconstruction artifacts from adjacent high-density cortex.
The Spectral Era: HU Evolved
In Spectral (Dual-Energy) CT, HU is no longer a fixed number. It becomes a function of energy. We transition from "Density Imaging" to "Material Decomposition."
Virtual Non-Contrast (VNC)
Subtracting iodine algorithmically to see the "true" underlying tissue without rescanning the patient.
Monoenergetic (MonoE)
Viewing images at 40 keV to boost iodine signal, making subtle tumors "pop" compared to standard 120 kVp.
Spectral HU Curve
Iodine HU spikes at low energy (keV), while water remains flat. This "Slope" is a fingerprint for iodine content.
Diagnostic "Sniff Test" Quiz
You see a 2cm adrenal nodule. The non-contrast HU is 8. What is your diagnosis?
Comments
Post a Comment