BACKGROUND: Although measuring vertebral bone density with the use of the computed tomography-based Hounsfield units (HU) is becoming increasingly popular, a much-desired standardized and robust method for measuring HU values accurately, quickly and conveniently was yet to exist. METHODS: Lumbar CT images of patients with degenerative lumbar diseases were analyzed. The HU values of the vertebral bodies were measured in three planes: mid-axial (including or not including the posterior venous plexus (PVP)), mid-sagittal and mid-frontal
the HU values of the vertebral pedicles were measured in two planes: axial and sagittal. The existing method was improved to measure the HU values of the S1 pedicles. A comparative analysis was performed, the inter- and intra-observer agreements were calculated. RESULTS: The HU values of the lumbar vertebrae in the mid-axial planes were the highest (p <
0.001), but showed the most reliable, reproducible and stable values of the intra-class correlation coefficient (ICC) (0.80-0.99 for vertebral bodies and 0.80-0.94 for pedicles). Non-inclusion of the PVP in the ROI did not affect HU values (p >
0.05), and the measurement of the HU values on a slice with the PVP showed high reproducibility (the ICC was 0.93-0.94). Measuring the HU values of the S1 pedicles with the use of the improved method led to an increase in ICC from 0.48 to 0.68. CONCLUSIONS: Measuring HU values in the L3 alone rather than in the entire lumbar spine or other vertebrae works up to expectations and may be recommended. The ROI should be placed in the mid-axial plane on a slice including the PVP. The most challenging HU measurement is for the S1 pedicles, but the proposed improved method makes it easier to do. These considerations may be particularly relevant to spinal surgeons in their routine practice, including planning the lumbar fusion and screw fixation.