Predicting outcomes of traumatic brain injury by imaging modality and injury distribution

Early prediction of outcomes after traumatic brain injury (TBI) is often difficult. To improve prognostic accuracy soon after trauma, we compared different radiological modalities and anatomical injury distribution in a group of adult TBI patients. The four methods studied were computed tomography (CT), magnetic resonance imaging (MRI) with T2-weighted imaging (T2WI), fluid-attenuated inversion recovery (FLAIR) imaging, and susceptibility weighted imaging (SWI). The objective of this study was to identify which modality and anatomic model best predict outcome. The patient population consisted of 38 adults admitted between February 2001 and May 2003. Early CT, T2WI, FLAIR, and SWI were obtained for each patient as well as a Glasgow Outcome Score (GOS) between 0.1 and 22 months (mean 9.2 months) after injury. Using a semi-automated computer method, intraparenchymal lesions were traced, measured, and converted to lesion volumes based on slice thickness and pixel size. Lesions were assigned to zones and regions. Outcomes were dichotomized into good (GOS 4-5) and poor (GOS 1-3) outcome groups. Brain injury detected by imaging was analyzed by median total lesion volume, median volume per lesion, and median number of lesions per outcome group. T2WI and FLAIR imaging most consistently discriminated between good and poor outcomes by median total lesion volume, median volume per lesion, and median number of lesions. In addition, T2WI and FLAIR imaging most consistently discriminated between good and poor outcomes by zonal distribution. While SWI rarely discriminated by outcome, it was very sensitive to intraparenchymal injury and its optimal use in evaluating TBI is unclear. SWI and other new imaging modalities should be further studied to fully evaluate their prognostic utility in TBI evaluation.