Limitations of diffusion-weighted imaging in the diagnosis of postoperative infections

OBJECTIVE: Diffusion-weighted imaging (DWI) has assumed a rapidly emerging role in the diagnosis of intracranial infection; however, its usefulness in the recognition of postoperative infection has been largely unexplored. We sought to determine the ability of DWI to accurately detect a broad range of postneurosurgical infections as well as identify individual factors that may limit its applicability. METHODS: We retrospectively identified 65 patients who had undergone surgery for the confirmed diagnosis of infection between August 2001 and February 2005 and had received preoperative magnetic resonance imaging evaluation, including DWI. Fifty patients developed infections after a primary neurosurgical procedure ("postoperative" infections), whereas 15 infections occurred without antecedent intervention ("spontaneous" infections). Logistic regression analysis was used to identify factors associated with false-negative DWI findings. Additionally, we investigated the false-positive rate of DWI by retrospectively reviewing a series of 30 consecutive patients who underwent craniotomy and received postoperative DWI. RESULTS: Spontaneously developing cranial infections exhibited evidence of restricted diffusion in 14 out of 15 (93%) patients; however, infections that occurred postoperatively were associated with a significant false-negative rate using DWI (36%; P < 0.01). Within the subset of patients with postoperative infection, location of infection significantly correlated with the DWI false-negative rate. Infections located extradurally were less likely to demonstrate restricted diffusion compared with those located primarily within the subdural or intraparenchymal spaces. Additionally, false-positive DWI findings were observed in 11 of the 30 patients (37%) who had DWI obtained postoperatively in the absence of infection. CONCLUSION: Utilization of DWI for the diagnosis of infection after primary neurosurgical intervention is associated with an elevated false-negative rate. The absence of restricted diffusion is not sufficient to exclude the presence of pyogenic postcraniotomy infection and should not be used as the principle determinant of patient management in this clinical setting.