Diagnostic accuracy of dual energy computed tomography for suspected pyogenic spondylodiscitis

While magnetic resonance imaging (MRI) is the diagnostic method of choice, we here analyze the diagnostic potential of dual-energy computed tomography (DECT) in differentiating abnormal discs from normal-appearing discs and in differentiating between infectious and degenerative conditions. Twenty-eight patients with suspected spondylodiscitis who underwent DECT and MRI of the spine were retrospectively included. Eighteen patients were diagnosed with spondylodiscitis and ten patients with degenerative disc disease. A combined clinical reference standard for the diagnosis was used. One abnormal disc and one normal-appearing disc per patient were included. Three blinded readers analyzed CT, DECT cMaps and MRI images. Quantitative analysis was performed in standardized regions of interest placed in each of the two discs included. Mixed-model analysis was used to identify correlations between CT density alterations and spondylodiscitis or degenerative disc disease. Diagnostic accuracy for differentiating abnormal discs and normal-appearing discs: 69.7% (95% CI, 56.0 to 81.2) for CT, 76.8% (95% CI, 63.6 to 87.0) for CT + DECT, 58.9% (95% CI, 45.0 to 71.9) for MRI; for differentiating spondylodiscitis and degenerative disc disease: 64.3% (95% CI, 44.1 to 81.4) for CT, 60.7% (95% CI, 40.6 to 78.5) for CT + DECT, 53.6% (95% CI, 33.9 to 72.5) for MRI. Mixed-model analysis revealed that normal-appearing discs had higher average density than abnormal discs in DECT (mean difference = 47.0 HU (95% CI, 32.8 to 61.3), p = < 0.001). In summary, both qualitative and quantitative DECT can distinguish normal-appearing discs from abnormal discs. Spondylodiscitis and degenerative disc disease were not distinguished accurately by DECT in this study.