Magnetic resonance imaging as an alternative to computed tomography in select patients with traumatic brain injury: a retrospective comparison

OBJECT Traumatic head injury (THI) is a highly prevalent condition in the United States, and concern regarding excess radiation-related cancer mortality has placed focus on limiting the use of CT in the evaluation of pediatric patients with THI. Given the success of rapid-acquisition MRI in the evaluation of ventriculoperitoneal shunt malfunction in pediatric patient populations, this study sought to evaluate the sensitivity of MRI in the setting of acute THI. METHODS Medical records of 574 pediatric admissions for THI to a Level 1 trauma center over a 10-year period were retrospectively reviewed to identify patients who underwent both CT and MRI examinations of the head within a 5-day period. Thirty-five patients were found, and diagnostic images were available for 30 patients. De-identified images were reviewed by a neuroradiologist for presence of any injury, intracranial hemorrhage, diffuse axonal injury (DAD, and skull fracture. Radiology reports were used to calculate interrater reliability scores. Baseline demographics and concordance analysis was performed with Stata version 13. RESULTS The mean age of the 30-patient cohort was 8.5 +/- 6.7 years, and 63.3% were male. The mean Injury Severity Score was 13.7 +/- 9.2, and the mean Glasgow Coma Scale score was 9 +/- 5.7. Radiology reports noted 150 abnormal findings. CT scanning missed findings in 12 patients; the missed findings included DAI (n = 5), subarachnoid hemorrhage (n = 6), small subdural hematomas (n = 6), cerebral contusions (n = 3), and an encephalocele. The CT scan was negative in 3 patients whose subsequent MRI revealed findings. MRI missed findings in 13 patients; missed findings included skull fracture (n = 5), small subdural hematomas (n = 4), cerebral contusions (n = 3), subarachnoid hemorrhage (n = 3), and DAI (n = 1). MRI was negative in 1 patient whose preceding CT scan was read as positive for injury. Although MRI more frequently reported intracranial findings than CT scanning, there was no statistically significant difference between CT and MRI in the detection of any intracranial injury (p = 0.63), DAI (p = 0.22), or intracranial hemorrhage (p = 0.25). CT scanning tended to more frequently identify skull fractures than MRI (p = 0.06). CONCLUSIONS MRI may be as sensitive as CT scanning in the detection of THI, DAI, and intracranial hemorrhage, but missed skull fractures in 5 of 13 patients. MRI may be a useful alternative to CT scanning in select stable patients with mild THI who warrant neuroimaging by clinical decision rules.