CT Grading of Fatty Liver
On unenhanced CT, hepatic steatosis can be graded using specific Hounsfield unit (HU) thresholds: liver attenuation ≤40 HU indicates at least moderate steatosis (≥30% fat), while progressive decreases at 64.4 HU, 59.1 HU, 41.9 HU, and 25.0 HU correlate with 0%, 1-25%, 26-50%, and >50% fatty infiltration respectively. 1
Absolute Liver Attenuation Criteria (Unenhanced CT)
The most clinically useful thresholds for grading hepatic steatosis on non-contrast CT are:
- ≤40 HU: Correlates with ≥30% steatosis (moderate-to-severe), with 100% specificity and 82% sensitivity 1
- ≤48 HU: Alternative threshold with 100% specificity but only 53.8% sensitivity for moderate-to-severe steatosis 2
- <50 HU: Identifies grade 1 steatosis with high specificity 3
- <45 HU: Identifies grade 2 steatosis 3
- <35 HU: Identifies grade 3 steatosis 3
- <20 HU: Identifies grade 4 (severe) steatosis 3
The graded attenuation scale provides more granular assessment: Normal liver (0% fat) measures approximately 64.4 HU, mild steatosis (1-25% fat) at 59.1 HU, moderate steatosis (26-50% fat) at 41.9 HU, and severe steatosis (>50% fat) at 25.0 HU 1.
Liver-to-Spleen Comparison
A liver-to-spleen attenuation difference >10 HU (with liver lower than spleen) and a liver-to-spleen ratio <1.0 diagnose moderate-to-severe steatosis 1. This comparative method helps account for variations in CT technique, though it requires both organs to be free of confounding pathology.
On contrast-enhanced CT, a liver-minus-spleen difference of ≤-20.5 HU at 80-100 seconds or ≤-18.5 HU at 101-120 seconds indicates fatty liver 4. However, these values are protocol-specific and less reliable than unenhanced measurements.
Critical Limitations and Pitfalls
CT is NOT appropriate as a first-line modality for diagnosing mild steatosis (<30% fat) due to:
- Poor sensitivity for early disease (fails to detect steatosis <30%) 1
- Radiation exposure concerns 2
- Lack of accuracy and reliability for mild disease 1
Contrast-enhanced CT should be avoided for steatosis quantification because HU values vary with contrast timing, injection site, protocol differences, scan duration, and patient hemodynamics 1. The attenuation is affected primarily by blood flow rather than fat content 2.
Important confounders that invalidate CT measurements:
- Iron deposition (increases attenuation, masking steatosis)
- Copper accumulation
- Glycogen storage
- Amiodarone deposition
- Any infiltrative liver disease 2
Practical Application Algorithm
- Use only unenhanced (non-contrast) CT for steatosis assessment
- Measure volumetric liver attenuation using either whole-liver segmentation or four-point sampling (no significant difference between methods) 5
- Apply the 40 HU threshold as the most clinically validated cutoff:
- ≤40 HU = moderate-to-severe steatosis (≥30% fat) - high specificity, warrants clinical action
40 HU = excludes moderate-to-severe steatosis
- For borderline cases (40-48 HU), consider liver-to-spleen comparison as confirmatory
- If more precise grading needed, use the progressive scale (64.4,59.1,41.9,25.0 HU)
Tube Voltage Considerations
CT attenuation values are affected by tube voltage - measurements at 80 kVp differ from 120 kVp 6. When comparing serial studies or applying published thresholds, ensure consistent kVp settings or apply voltage-specific adjusted cutoffs using validated regression equations 6.
Clinical Context
While CT provides specific quantitative thresholds, MRI with proton density fat fraction (PDFF) remains superior for detecting and grading steatosis, particularly for mild disease (≥5% fat), with sensitivity 76.7-90.0% and specificity 87.1-91% 1. Ultrasound is the appropriate first-line screening modality 2, with CT reserved for opportunistic detection when performed for other indications 7.
The high specificity of the ≤40 HU threshold makes CT useful for retrospective identification of moderate-to-severe steatosis in large cohorts for epidemiological studies 8, 9, but the low sensitivity precludes effective population screening 8.