Low-dose multi-detector computed tomography for periradicular infiltrations at the cervical and lumbar spine

Abstract

Periradicular infiltrations are frequently performed in daily neuroradiological routine and are often guided by multi-detector computed tomography (MDCT), thus leading to radiation exposure. The purpose of this study was to evaluate MDCT with low dose (LD) and model-based iterative reconstruction for image-guided periradicular infiltrations at the cervical and lumbosacral spine. We retrospectively analyzed 204 MDCT scans acquired for the purpose of cervical or lumbosacral periradicular interventions, which were either derived from scanning with standard dose (SD; 40 mA and 120 kVp) or LD (20–30 mA and 120 kVp) using a 128-slice MDCT scanner. The SD cases were matched to the LD cases considering sex, age, level of infiltration, presence of spinal instrumentation, and body diameter. All images were reconstructed using model-based iterative image reconstruction and were evaluated by two readers (R1 and R2) using 5- or 3-point Likert scales (score of 1 reflects the best value per category). Furthermore, noise in imaging data was quantitatively measured by the standard deviation (StDev) of muscle tissue. The dose length product (DLP) was statistically significantly lower for LD scans (6.75 ± 6.43 mGy*cm vs. 10.16 ± 7.70 mGy*cm; p < 0.01; reduction of 33.5%). Image noise was comparable between LD and SD scans (13.13 ± 3.66 HU vs. 13.37 ± 4.08 HU; p = 0.85). Overall image quality was scored as good to very good with only minimal artifacts according to both readers, and determination of the nerve root was possible in almost all patients (LD vs. SD: p > 0.05 for all items). This resulted in high confidence for intervention planning as well as periprocedural intervention guidance for both SD and LD scans. The inter-reader agreement was at least substantial (weighted Cohen’s κ ≥ 0.62), except for confidence in intervention planning for LD scans (κ = 0.49). In conclusion, considerable dose reduction for planning and performing periradicular infiltrations with MDCT using model-based iterative image reconstruction is feasible and can be performed without clinically relevant drawbacks regarding image quality or confidence for planning.