LAUSR

OBJECTIVE Application of Medishield to the nerve root is common during spinal surgery to create a mechanical barrier from pain mediators and reduce scar formation. However, Medishield's signal characteristics on magnetic resonance imaging (MRI) have not yet been examined. METHODS Microsurgical interlaminotomy was performed on 2 lower lumbar segments in 17 adult New Zealand white rabbits. After dural exposure, applications of 1 mL (autologous blood clot or Medishield) were randomized for each level. On postoperative days 1 through 3, various MRI sequences in 1.5T were performed including T1-weighted, T2-w, T1-gadolinium-weighted, susceptibility-weighted and turbo inversion recovery magnitude (TIRM) sequence. Signaling characteristics were analyzed by 3 blinded observers. Inter-rater agreement was calculated using Fleiss's kappa coefficient (κ). Positive and negative likelihood ratios in detecting Medishield by MRI were determined. RESULTS Of 24 MRIs performed, TIRM sequence identified Medishield with the highest likelihood ratio. Medishield's positive likelihood ratio was highest (5.8) on postoperative day 1 with interobserver agreement of 93% (κ = 0.75); these rates declined to 2.5 and 1.4 on postoperative days 2 and 3 with interobserver agreements of 71% (κ = 0.43) and 83% (κ = 0.67), respectively. Medishield adherence was confirmed in each rabbit by histologic examinations. CONCLUSION Understanding that radiologic detection of Medishield diminished over time as its signal characteristics became less distinguishable from a blood clot is essential in clinical practice. Medishield was detected on postoperative day 1 but not 2 days later after hemodynamic changes had occurred. These results may provide a guide for postoperative findings, such as differential diagnosis of hematoma.