Poster Presentation 51st International Society for the Study of the Lumbar Spine Annual Meeting 2025

Introduction:

Hounsfield units (HU) derived from computed tomography (CT) imaging have been reported to correlate positively with bone mineral density (BMD) and are useful for evaluating osteoporosis before spine surgery. However, accurate measurement of HU values is often hindered by artifacts caused by pedicle screw insertion after spinal fusion surgery. Recently, the use of a metal artifact reduction (MAR) algorithm has enabled the evaluation of bone quality even in vertebral bodies with pedicle screws. Nevertheless, detailed HU values using the MAR algorithm remain unclear. This study aimed to investigate HU values obtained through MAR algorithms in patients with spinal fusion surgery and to assess their association with osteoporosis.

Methods:

Patients who underwent spinal fusion surgery and had CT imaging performed with both conventional and MAR algorithms were analyzed. CT imaging using both algorithms was conducted on the same day or within one month. HU values were evaluated on mid-sagittal images.

Study 1: HU values obtained using the MAR algorithm were statistically compared to those obtained using the conventional algorithm. To investigate the effects of metal artifacts by screw insertion, HU values were evaluated in vertebral bodies with and without pedicle screw insertion.

Study 2: The correlation between HU values and T-scores was assessed.

Results:

A total of 25 patients (mean age: 72.8 years [range: 43–88], 90 vertebral bodies) were analyzed.

Study 1: In 61 vertebral bodies without pedicle screw insertion, HU values obtained using the MAR algorithm showed a strong positive correlation with those from the conventional algorithm (P < 0.01, r = 0.92). However, HU values were significantly lower with the MAR algorithm than with the conventional algorithm (P < 0.01; MAR: 67 ± 55, conventional: 98 ± 61). Similarly, in 29 vertebral bodies with pedicle screw insertion, HU values from both algorithms showed a strong positive correlation (P < 0.01, r = 0.86), but HU values from the MAR algorithm were significantly lower (P < 0.01; MAR: 80 ± 45, conventional: 111 ± 58).

Study 2: HU values obtained from the MAR and conventional algorithms positively correlated with T-scores (P < 0.05, r = 0.66–0.69).

Discussion:

HU values were successfully measured using the MAR algorithm, showing a strong correlation with the conventional algorithm and a significant association with T-scores. This indicates that HU values derived from the MAR algorithm can be used to evaluate BMD. Furthermore, the MAR algorithm allowed for the assessment of HU values in vertebral bodies with pedicle screw insertion by mitigating implant artifacts. This method offers a valuable tool for evaluating BMD not only before spinal surgery but also in patients who have undergone spinal fusion surgery.