Impact of contrast administration and CT reconstruction plane on Hounsfield units for assessing underlying bone quality in the lumbar spine.

Abstract

Objective: Hounsfield units (HUs) may better predict biomechanical complications of instrumented fusion than conventional bone quality measures. Typically, noncontrast axial slices are used. This study aims to address the influence of reconstruction plane and contrast administration on measured HUs in patients undergoing lumbar spine imaging.

Methods: Patients without prior spine surgery, infection, or tumor were identified for 3 groups: 1) 150 patients with high-resolution noncontrast lumbar CT, 2) 109 patients with noncontrast and contrast-enhanced lumbar CT, and 3) 100 patients with noncontrast lumbar CT and lumbar CT myelograms. Noncontrast HU measures in group 1 were compared between axial, sagittal, and coronal reconstructions. HU measures for groups 2 and 3 were compared on axial slice averages. HU measures between groups were compared using the Wilcoxon signed-rank test to investigate the presence of a statistically significant difference between groups (α = 0.05). Linear regression was also used to determine the degree of correlation between HU measures on noncontrast axial CT slices and HU measures on other sequences.

Results: The mean patient age was 58.8 ± 15.9 years for group 1 (54.0% male), 65.5 ± 15.9 years for group 2 (61.5% male), and 65.2 ± 14.2 years (53.5% male) for group 3. Comparison of HU measures in group 1 showed significant differences across measurement modalities with the exception of axial average versus coronal average measures (p = 0.257) and sagittal average versus midsagittal measures (p = 0.726). There were significant differences in average axial HUs for each vertebral body between contrast and noncontrast lumbar CT and between CT myelography and noncontrast CT. Linear regression analysis demonstrated an extremely high correlation between measures for all methodologies (r2 = 0.950-0.986, all p < 0.001), and between contrast-enhanced and noncontrast studies (r2 = 0.870, p < 0.001). Measurements on CT myelography were significantly correlated with those in noncontrast studies, although to a lesser degree (r2 = 0.745, p < 0.001).

Conclusions: HU measures on noncontrast CT appear similar across multiple different planes, suggesting that a simpler methodology such as single-slice measurement on midsagittal reconstruction may allow for more rapid assessment of underlying bone quality. Consistent with prior work, contrast-enhanced CT sequences appear to reliably mirror underlying bone quality, although CT myelogram measures may vary in a more unpredictable way that precludes their use to interpret underlying bone quality. HU measures on contrast-enhanced studies are on average greater, suggesting the need for higher cutoff values.